source: palm/trunk/DOC/app/chapter_4.1.html @ 61

Last change on this file since 61 was 61, checked in by raasch, 16 years ago

further preliminary changes for revision 3.2

  • Property svn:keywords set to Id
File size: 136.1 KB
Line 
1<!DOCTYPE html PUBLIC "-//W3C//DTD HTML 4.01 Transitional//EN">
2<html><head>
3<meta http-equiv="content-type" content="text/html; charset=ISO-8859-1"><title>PALM
4chapter 4.1</title></head>
5<body><h3><a name="chapter4.1"></a>4.1
6Initialization parameters</h3>
7<br><table style="text-align: left; width: 100%;" border="1" cellpadding="2" cellspacing="2"> <tbody>
8<tr> <td style="vertical-align: top;"><font size="4"><b>Parameter name</b></font></td>
9<td style="vertical-align: top;"><font size="4"><b>Type</b></font></td>
10<td style="vertical-align: top;"> <p><b><font size="4">Default</font></b> <br> <b><font size="4">value</font></b></p> </td>
11<td style="vertical-align: top;"><font size="4"><b>Explanation</b></font></td>
12</tr> <tr> <td style="vertical-align: top;">
13<p><a name="adjust_mixing_length"></a><b>adjust_mixing_length</b></p>
14</td> <td style="vertical-align: top;">L</td>
15<td style="vertical-align: top;"><span style="font-style: italic;">.F.</span></td> <td style="vertical-align: top;"> <p style="font-style: normal;">Near-surface adjustment of the
16mixing length to the Prandtl-layer law.&nbsp; </p> <p>Usually
17the mixing length in LES models l<sub>LES</sub>
18depends (as in PALM) on the grid size and is possibly restricted
19further in case of stable stratification and near the lower wall (see
20parameter <a href="#wall_adjustment">wall_adjustment</a>).
21With <b>adjust_mixing_length</b> = <span style="font-style: italic;">.T.</span>
22the Prandtl' mixing length l<sub>PR</sub> = kappa * z/phi
23is calculated
24and the mixing length actually used in the model is set l = MIN (l<sub>LES</sub>,
25l<sub>PR</sub>). This usually gives a decrease of the
26mixing length at
27the bottom boundary and considers the fact that eddy sizes
28decrease in the vicinity of the wall.&nbsp; </p> <p style="font-style: normal;"><b>Warning:</b> So
29far, there is
30no good experience with <b>adjust_mixing_length</b> = <span style="font-style: italic;">.T.</span> !&nbsp; </p>
31<p>With <b>adjust_mixing_length</b> = <span style="font-style: italic;">.T.</span> and the
32Prandtl-layer being
33switched on (see <a href="#prandtl_layer">prandtl_layer</a>)
34<span style="font-style: italic;">'(u*)** 2+neumann'</span>
35should always be set as the lower boundary condition for the TKE (see <a href="#bc_e_b">bc_e_b</a>),
36otherwise the near-surface value of the TKE is not in agreement with
37the Prandtl-layer law (Prandtl-layer law and Prandtl-Kolmogorov-Ansatz
38should provide the same value for K<sub>m</sub>). A warning
39is given,
40if this is not the case.</p> </td> </tr> <tr>
41<td style="vertical-align: top;"> <p><a name="alpha_surface"></a><b>alpha_surface</b></p>
42</td> <td style="vertical-align: top;">R<br> </td>
43<td style="vertical-align: top;"><span style="font-style: italic;">0.0</span><br> </td>
44<td style="vertical-align: top;"> <p style="font-style: normal;">Inclination of the model domain
45with respect to the horizontal (in degrees).&nbsp; </p> <p style="font-style: normal;">By means of <b>alpha_surface</b>
46the model domain can be inclined in x-direction with respect to the
47horizontal. In this way flows over inclined surfaces (e.g. drainage
48flows, gravity flows) can be simulated. In case of <b>alpha_surface
49</b>/= <span style="font-style: italic;">0</span>
50the buoyancy term
51appears both in
52the equation of motion of the u-component and of the w-component.<br>
53</p> <p style="font-style: normal;">An inclination
54is only possible in
55case of cyclic horizontal boundary conditions along x AND y (see <a href="#bc_lr">bc_lr</a>
56and <a href="#bc_ns">bc_ns</a>) and <a href="#topography">topography</a> = <span style="font-style: italic;">'flat'</span>. </p>
57<p>Runs with inclined surface still require additional
58user-defined code as well as modifications to the default code. Please
59ask the <a href="http://www.muk.uni-hannover.de/%7Eraasch/PALM_group/PALM_group.html#0">PALM
60developer&nbsp; group</a>.</p> </td> </tr>
61<tr> <td style="vertical-align: top;"> <p><a name="bc_e_b"></a><b>bc_e_b</b></p> </td>
62<td style="vertical-align: top;">C * 20</td> <td style="vertical-align: top;"><span style="font-style: italic;">'neumann'</span></td>
63<td style="vertical-align: top;"> <p style="font-style: normal;">Bottom boundary condition of the
64TKE.&nbsp; </p> <p><b>bc_e_b</b> may be
65set to&nbsp;<span style="font-style: italic;">'neumann'</span>
66or <span style="font-style: italic;">'(u*) ** 2+neumann'</span>.
67<b>bc_e_b</b>
68= <span style="font-style: italic;">'neumann'</span>
69yields to
70e(k=0)=e(k=1) (Neumann boundary condition), where e(k=1) is calculated
71via the prognostic TKE equation. Choice of <span style="font-style: italic;">'(u*)**2+neumann'</span>
72also yields to
73e(k=0)=e(k=1), but the TKE at the Prandtl-layer top (k=1) is calculated
74diagnostically by e(k=1)=(us/0.1)**2. However, this is only allowed if
75a Prandtl-layer is used (<a href="#prandtl_layer">prandtl_layer</a>).
76If this is not the case, a warning is given and <b>bc_e_b</b>
77is reset
78to <span style="font-style: italic;">'neumann'</span>.&nbsp;
79</p> <p style="font-style: normal;">At the top
80boundary a Neumann
81boundary condition is generally used: (e(nz+1) = e(nz)).</p> </td>
82</tr> <tr> <td style="vertical-align: top;">
83<p><a name="bc_lr"></a><b>bc_lr</b></p>
84</td> <td style="vertical-align: top;">C * 20</td>
85<td style="vertical-align: top;"><span style="font-style: italic;">'cyclic'</span></td>
86<td style="vertical-align: top;">Boundary
87condition along x (for all quantities).<br> <br>
88By default, a cyclic boundary condition is used along x.<br> <br>
89<span style="font-weight: bold;">bc_lr</span> may
90also be
91assigned the values <span style="font-style: italic;">'dirichlet/neumann'</span>
92(inflow from left, outflow to the right) or <span style="font-style: italic;">'neumann/dirichlet'</span>
93(inflow from
94right, outflow to the left). This requires the multi-grid method to be
95used for solving the Poisson equation for perturbation pressure (see <a href="http://www.muk.uni-hannover.de/%7Eraasch/PALM_group/doc/app/chapter_4.2.html#psolver">psolver</a>)
96and it also requires cyclic boundary conditions along y (see<br> <a href="#bc_ns">bc_ns</a>).<br> <br>
97In case of these non-cyclic lateral boundaries, a Dirichlet condition
98is used at the inflow for all quantities (initial vertical profiles -
99see <a href="#initializing_actions">initializing_actions</a>
100- are fixed during the run) except u, to which a Neumann (zero
101gradient) condition is applied. At the outflow, a Neumann (zero
102gradient) condition is used for all quantities except v, which is set
103to its horizontal average along the outflow (e.g. v(k,:,nx+1) =
104average_along_y( v(k,:,nx)), and except w, which is set to zero
105(Dirichlet condition). These conditions ensure the velocity field to be
106free of divergence at the inflow and at the outflow. For perturbation
107pressure Neumann (zero gradient) conditions are assumed both at the
108inflow and at the outflow.<br> <br>
109When using non-cyclic lateral boundaries, a filter is applied to the
110velocity field in the vicinity of the outflow in order to suppress any
111reflections of outgoing disturbances (see <a href="#km_damp_max">km_damp_max</a>
112and <a href="#outflow_damping_width">outflow_damping_width</a>).<br>
113<br>
114In order to maintain a turbulent state of the flow, it may be
115neccessary to continuously impose perturbations on the horizontal
116velocity field in the vicinity of the inflow throughout the whole run.
117This can be switched on using <a href="http://www.muk.uni-hannover.de/%7Eraasch/PALM_group/doc/app/chapter_4.2.html#create_disturbances">create_disturbances</a>.
118The horizontal range to which these perturbations are applied is
119controlled by the parameters <a href="#inflow_disturbance_begin">inflow_disturbance_begin</a>
120and <a href="#inflow_disturbance_end">inflow_disturbance_end</a>.
121The vertical range and the perturbation amplitude are given by <a href="http://www.muk.uni-hannover.de/%7Eraasch/PALM_group/doc/app/chapter_4.2.html#psolver">disturbance_level_b</a>,
122<a href="http://www.muk.uni-hannover.de/%7Eraasch/PALM_group/doc/app/chapter_4.2.html#psolver">disturbance_level_t</a>,
123and <a href="http://www.muk.uni-hannover.de/%7Eraasch/PALM_group/doc/app/chapter_4.2.html#psolver">disturbance_amplitude</a>.
124The time interval at which perturbations are to be imposed is set by <a href="http://www.muk.uni-hannover.de/%7Eraasch/PALM_group/doc/app/chapter_4.2.html#dt_disturb">dt_disturb</a>.<br>
125<br>
126In case of non-cyclic horizontal boundaries <a href="http://www.muk.uni-hannover.de/%7Eraasch/PALM_group/doc/app/chapter_4.2.html#call_psolver_at_all_substeps">call_psolver
127at_all_substeps</a> = .T. should be used.<br> <br> <span style="font-weight: bold;">Note:</span><br>
128Using non-cyclic lateral boundaries requires very sensitive adjustments
129of the inflow (vertical profiles) and the bottom boundary conditions,
130e.g. a surface heating should not be applied near the inflow boundary
131because this may significantly disturb the inflow. Please check the
132model results very carefully.</td> </tr> <tr> <td style="vertical-align: top;"> <p><a name="bc_ns"></a><b>bc_ns</b></p>
133</td> <td style="vertical-align: top;">C * 20</td>
134<td style="vertical-align: top;"><span style="font-style: italic;">'cyclic'</span></td>
135<td style="vertical-align: top;">Boundary
136condition along y (for all quantities).<br> <br>
137By default, a cyclic boundary condition is used along y.<br> <br>
138<span style="font-weight: bold;">bc_ns</span> may
139also be
140assigned the values <span style="font-style: italic;">'dirichlet/neumann'</span>
141(inflow from rear ("north"), outflow to the front ("south")) or <span style="font-style: italic;">'neumann/dirichlet'</span>
142(inflow from front ("south"), outflow to the rear ("north")). This
143requires the multi-grid
144method to be used for solving the Poisson equation for perturbation
145pressure (see <a href="chapter_4.2.html#psolver">psolver</a>)
146and it also requires cyclic boundary conditions along x (see<br> <a href="#bc_lr">bc_lr</a>).<br> <br>
147In case of these non-cyclic lateral boundaries, a Dirichlet condition
148is used at the inflow for all quantities (initial vertical profiles -
149see <a href="#initializing_actions">initializing_actions</a>
150- are fixed during the run) except v, to which a Neumann (zero
151gradient) condition is applied. At the outflow, a Neumann (zero
152gradient) condition is used for all quantities except u, which is set
153to its horizontal average along the outflow (e.g. u(k,ny+1,:) =
154average_along_x( u(k,ny,:)), and except w, which is set to zero
155(Dirichlet condition). These conditions ensure the velocity field to be
156free of divergence at the inflow and at the outflow. For perturbation
157pressure Neumann (zero gradient) conditions are assumed both at the
158inflow and at the outflow.<br> <br>
159For further details regarding non-cyclic lateral boundary conditions
160see <a href="#bc_lr">bc_lr</a>.</td> </tr>
161<tr> <td style="vertical-align: top;"> <p><a name="bc_p_b"></a><b>bc_p_b</b></p> </td>
162<td style="vertical-align: top;">C * 20</td> <td style="vertical-align: top;"><span style="font-style: italic;">'neumann'</span></td>
163<td style="vertical-align: top;"> <p style="font-style: normal;">Bottom boundary condition of the
164perturbation pressure.&nbsp; </p> <p>Allowed values
165are <span style="font-style: italic;">'dirichlet'</span>,
166<span style="font-style: italic;">'neumann'</span>
167and <span style="font-style: italic;">'neumann+inhomo'</span>.&nbsp;
168<span style="font-style: italic;">'dirichlet'</span>
169sets
170p(k=0)=0.0,&nbsp; <span style="font-style: italic;">'neumann'</span>
171sets p(k=0)=p(k=1). <span style="font-style: italic;">'neumann+inhomo'</span>
172corresponds to an extended Neumann boundary condition where heat flux
173or temperature inhomogeneities near the
174surface (pt(k=1))&nbsp; are additionally regarded (see Shen and
175LeClerc
176(1995, Q.J.R. Meteorol. Soc.,
1771209)). This condition is only permitted with the Prandtl-layer
178switched on (<a href="#prandtl_layer">prandtl_layer</a>),
179otherwise the run is terminated.&nbsp; </p> <p>Since
180at the bottom boundary of the model the vertical
181velocity
182disappears (w(k=0) = 0.0), the consistent Neumann condition (<span style="font-style: italic;">'neumann'</span> or <span style="font-style: italic;">'neumann+inhomo'</span>)
183dp/dz = 0 should
184be used, which leaves the vertical component w unchanged when the
185pressure solver is applied. Simultaneous use of the Neumann boundary
186conditions both at the bottom and at the top boundary (<a href="#bc_p_t">bc_p_t</a>)
187usually yields no consistent solution for the perturbation pressure and
188should be avoided.</p> </td> </tr> <tr> <td style="vertical-align: top;"> <p><a name="bc_p_t"></a><b>bc_p_t</b></p>
189</td> <td style="vertical-align: top;">C * 20</td>
190<td style="vertical-align: top;"><span style="font-style: italic;">'dirichlet'</span></td>
191<td style="vertical-align: top;"> <p style="font-style: normal;">Top boundary condition of the
192perturbation pressure.&nbsp; </p> <p style="font-style: normal;">Allowed values are <span style="font-style: italic;">'dirichlet'</span>
193(p(k=nz+1)= 0.0) or <span style="font-style: italic;">'neumann'</span>
194(p(k=nz+1)=p(k=nz)).&nbsp; </p> <p>Simultaneous use
195of Neumann boundary conditions both at the
196top and bottom boundary (<a href="#bc_p_b">bc_p_b</a>)
197usually yields no consistent solution for the perturbation pressure and
198should be avoided. Since at the bottom boundary the Neumann
199condition&nbsp; is a good choice (see <a href="#bc_p_b">bc_p_b</a>),
200a Dirichlet condition should be set at the top boundary.</p> </td>
201</tr> <tr> <td style="vertical-align: top;">
202<p><a name="bc_pt_b"></a><b>bc_pt_b</b></p>
203</td> <td style="vertical-align: top;">C*20</td>
204<td style="vertical-align: top;"><span style="font-style: italic;">'dirichlet'</span></td>
205<td style="vertical-align: top;"> <p style="font-style: normal;">Bottom boundary condition of the
206potential temperature.&nbsp; </p> <p>Allowed values
207are <span style="font-style: italic;">'dirichlet'</span>
208(pt(k=0) = const. = <a href="#pt_surface">pt_surface</a>
209+ <a href="#pt_surface_initial_change">pt_surface_initial_change</a>;
210the user may change this value during the run using user-defined code)
211and <span style="font-style: italic;">'neumann'</span>
212(pt(k=0)=pt(k=1)).&nbsp; <br>
213When a constant surface sensible heat flux is used (<a href="#surface_heatflux">surface_heatflux</a>), <b>bc_pt_b</b>
214= <span style="font-style: italic;">'neumann'</span>
215must be used, because otherwise the resolved scale may contribute to
216the surface flux so that a constant value cannot be guaranteed.</p>
217</td> </tr> <tr> <td style="vertical-align: top;"> <p><a name="pc_pt_t"></a><b>bc_pt_t</b></p>
218</td> <td style="vertical-align: top;">C * 20</td>
219<td style="vertical-align: top;"><span style="font-style: italic;">'initial gradient'</span></td>
220<td style="vertical-align: top;"> <p style="font-style: normal;">Top boundary condition of the
221potential temperature.&nbsp; </p> <p>Allowed are the
222values <span style="font-style: italic;">'dirichlet' </span>(pt(k=nz+1)
223does not change during the run), <span style="font-style: italic;">'neumann'</span>
224(pt(k=nz+1)=pt(k=nz)), and <span style="font-style: italic;">'initial_gradient'</span>.
225With the 'initial_gradient'-condition the value of the temperature
226gradient at the top is
227calculated from the initial
228temperature profile (see <a href="#pt_surface">pt_surface</a>,
229<a href="#pt_vertical_gradient">pt_vertical_gradient</a>)
230by bc_pt_t_val = (pt_init(k=nz+1) -
231pt_init(k=nz)) / dzu(nz+1).<br>
232Using this value (assumed constant during the
233run) the temperature boundary values are calculated as&nbsp; </p>
234<ul> <p style="font-style: normal;">pt(k=nz+1) =
235pt(k=nz) +
236bc_pt_t_val * dzu(nz+1)</p> </ul> <p style="font-style: normal;">(up to k=nz the prognostic
237equation for the temperature is solved).<br>
238When a constant sensible heat flux is used at the top boundary (<a href="chapter_4.1.html#top_heatflux">top_heatflux</a>),
239<b>bc_pt_t</b> = <span style="font-style: italic;">'neumann'</span>
240must be used, because otherwise the resolved scale may contribute to
241the top flux so that a constant value cannot be guaranteed.</p> </td>
242</tr> <tr> <td style="vertical-align: top;">
243<p><a name="bc_q_b"></a><b>bc_q_b</b></p>
244</td> <td style="vertical-align: top;">C * 20</td>
245<td style="vertical-align: top;"><span style="font-style: italic;">'dirichlet'</span></td>
246<td style="vertical-align: top;"> <p style="font-style: normal;">Bottom boundary condition of the
247specific humidity / total water content.&nbsp; </p> <p>Allowed
248values are <span style="font-style: italic;">'dirichlet'</span>
249(q(k=0) = const. = <a href="#q_surface">q_surface</a>
250+ <a href="#q_surface_initial_change">q_surface_initial_change</a>;
251the user may change this value during the run using user-defined code)
252and <span style="font-style: italic;">'neumann'</span>
253(q(k=0)=q(k=1)).&nbsp; <br>
254When a constant surface latent heat flux is used (<a href="#surface_waterflux">surface_waterflux</a>), <b>bc_q_b</b>
255= <span style="font-style: italic;">'neumann'</span>
256must be used, because otherwise the resolved scale may contribute to
257the surface flux so that a constant value cannot be guaranteed.</p>
258</td> </tr> <tr> <td style="vertical-align: top;"> <p><a name="bc_q_t"></a><b>bc_q_t</b></p>
259</td> <td style="vertical-align: top;"><span style="font-style: italic;">C
260* 20</span></td> <td style="vertical-align: top;"><span style="font-style: italic;">'neumann'</span></td>
261<td style="vertical-align: top;"> <p style="font-style: normal;">Top boundary condition of the
262specific humidity / total water content.&nbsp; </p> <p>Allowed
263are the values <span style="font-style: italic;">'dirichlet'</span>
264(q(k=nz) and q(k=nz+1) do
265not change during the run) and <span style="font-style: italic;">'neumann'</span>.
266With the Neumann boundary
267condition the value of the humidity gradient at the top is calculated
268from the
269initial humidity profile (see <a href="#q_surface">q_surface</a>,
270<a href="#q_vertical_gradient">q_vertical_gradient</a>)
271by: bc_q_t_val = ( q_init(k=nz) - q_init(k=nz-1)) / dzu(nz).<br>
272Using this value (assumed constant during the run) the humidity
273boundary values
274are calculated as&nbsp; </p> <ul> <p style="font-style: normal;">q(k=nz+1) =q(k=nz) +
275bc_q_t_val * dzu(nz+1)</p> </ul> <p style="font-style: normal;">(up tp k=nz the prognostic
276equation for q is solved). </p> </td> </tr> <tr>
277<td style="vertical-align: top;"> <p><a name="bc_s_b"></a><b>bc_s_b</b></p> </td>
278<td style="vertical-align: top;">C * 20</td> <td style="vertical-align: top;"><span style="font-style: italic;">'dirichlet'</span></td>
279<td style="vertical-align: top;"> <p style="font-style: normal;">Bottom boundary condition of the
280scalar concentration.&nbsp; </p> <p>Allowed values
281are <span style="font-style: italic;">'dirichlet'</span>
282(s(k=0) = const. = <a href="#s_surface">s_surface</a>
283+ <a href="#s_surface_initial_change">s_surface_initial_change</a>;
284the user may change this value during the run using user-defined code)
285and <span style="font-style: italic;">'neumann'</span>
286(s(k=0) =
287s(k=1)).&nbsp; <br>
288When a constant surface concentration flux is used (<a href="#surface_scalarflux">surface_scalarflux</a>), <b>bc_s_b</b>
289= <span style="font-style: italic;">'neumann'</span>
290must be used, because otherwise the resolved scale may contribute to
291the surface flux so that a constant value cannot be guaranteed.</p>
292</td> </tr> <tr> <td style="vertical-align: top;"> <p><a name="bc_s_t"></a><b>bc_s_t</b></p>
293</td> <td style="vertical-align: top;">C * 20</td>
294<td style="vertical-align: top;"><span style="font-style: italic;">'neumann'</span></td>
295<td style="vertical-align: top;"> <p style="font-style: normal;">Top boundary condition of the
296scalar concentration.&nbsp; </p> <p>Allowed are the
297values <span style="font-style: italic;">'dirichlet'</span>
298(s(k=nz) and s(k=nz+1) do
299not change during the run) and <span style="font-style: italic;">'neumann'</span>.
300With the Neumann boundary
301condition the value of the scalar concentration gradient at the top is
302calculated
303from the initial scalar concentration profile (see <a href="#s_surface">s_surface</a>, <a href="#s_vertical_gradient">s_vertical_gradient</a>)
304by: bc_s_t_val = (s_init(k=nz) - s_init(k=nz-1)) / dzu(nz).<br>
305Using this value (assumed constant during the run) the concentration
306boundary values
307are calculated as </p> <ul> <p style="font-style: normal;">s(k=nz+1) = s(k=nz) +
308bc_s_t_val * dzu(nz+1)</p> </ul> <p style="font-style: normal;">(up to k=nz the prognostic
309equation for the scalar concentration is
310solved).</p> </td> </tr> <tr> <td style="vertical-align: top;"> <p><a name="bc_uv_b"></a><b>bc_uv_b</b></p>
311</td> <td style="vertical-align: top;">C * 20</td>
312<td style="vertical-align: top;"><span style="font-style: italic;">'dirichlet'</span></td>
313<td style="vertical-align: top;"> <p style="font-style: normal;">Bottom boundary condition of the
314horizontal velocity components u and v.&nbsp; </p> <p>Allowed
315values are <span style="font-style: italic;">'dirichlet' </span>and
316<span style="font-style: italic;">'neumann'</span>. <b>bc_uv_b</b>
317= <span style="font-style: italic;">'dirichlet'</span>
318yields the
319no-slip condition with u=v=0 at the bottom. Due to the staggered grid
320u(k=0) and v(k=0) are located at z = - 0,5 * <a href="#dz">dz</a>
321(below the bottom), while u(k=1) and v(k=1) are located at z = +0,5 *
322dz. u=v=0 at the bottom is guaranteed using mirror boundary
323condition:&nbsp; </p> <ul> <p style="font-style: normal;">u(k=0) = - u(k=1) and v(k=0) = -
324v(k=1)</p> </ul> <p style="font-style: normal;">The
325Neumann boundary condition
326yields the free-slip condition with u(k=0) = u(k=1) and v(k=0) =
327v(k=1).
328With Prandtl - layer switched on, the free-slip condition is not
329allowed (otherwise the run will be terminated)<font color="#000000">.</font></p>
330</td> </tr> <tr> <td style="vertical-align: top;"> <p><a name="bc_uv_t"></a><b>bc_uv_t</b></p>
331</td> <td style="vertical-align: top;">C * 20</td>
332<td style="vertical-align: top;"><span style="font-style: italic;">'dirichlet'</span></td>
333<td style="vertical-align: top;"> <p style="font-style: normal;">Top boundary condition of the
334horizontal velocity components u and v.&nbsp; </p> <p>Allowed
335values are <span style="font-style: italic;">'dirichlet'</span>
336and <span style="font-style: italic;">'neumann'</span>.
337The
338Dirichlet condition yields u(k=nz+1) = ug(nz+1) and v(k=nz+1) =
339vg(nz+1),
340Neumann condition yields the free-slip condition with u(k=nz+1) =
341u(k=nz) and v(k=nz+1) = v(k=nz) (up to k=nz the prognostic equations
342for the velocities are solved).</p> </td> </tr> <tr>
343<td style="vertical-align: top;"><span style="font-weight: bold;"><a name="building_height"></a>building_height</span></td>
344<td style="vertical-align: top;">R</td> <td style="vertical-align: top;"><span style="font-style: italic;">50.0</span></td> <td>Height
345of a single building in m.<br> <br> <span style="font-weight: bold;">building_height</span> must
346be less than the height of the model domain. This parameter requires
347the use of&nbsp;<a href="#topography">topography</a>
348= <span style="font-style: italic;">'single_building'</span>.</td>
349</tr> <tr> <td style="vertical-align: top;"><span style="font-weight: bold;"><a name="building_length_x"></a>building_length_x</span></td>
350<td style="vertical-align: top;">R</td> <td style="vertical-align: top;"><span style="font-style: italic;">50.0</span></td> <td><span style="font-style: italic;"></span>Width of a single
351building in m.<br> <br>
352Currently, <span style="font-weight: bold;">building_length_x</span>
353must be at least <span style="font-style: italic;">3
354*&nbsp;</span><a style="font-style: italic;" href="#dx">dx</a> and no more than <span style="font-style: italic;">(&nbsp;</span><a style="font-style: italic;" href="#nx">nx</a><span style="font-style: italic;"> - 1 ) </span><span style="font-style: italic;"> * <a href="#dx">dx</a>
355</span><span style="font-style: italic;">- <a href="#building_wall_left">building_wall_left</a></span>.
356This parameter requires the use of&nbsp;<a href="#topography">topography</a>
357= <span style="font-style: italic;">'single_building'</span>.</td>
358</tr> <tr> <td style="vertical-align: top;"><span style="font-weight: bold;"><a name="building_length_y"></a>building_length_y</span></td>
359<td style="vertical-align: top;">R</td> <td style="vertical-align: top;"><span style="font-style: italic;">50.0</span></td> <td>Depth
360of a single building in m.<br> <br>
361Currently, <span style="font-weight: bold;">building_length_y</span>
362must be at least <span style="font-style: italic;">3
363*&nbsp;</span><a style="font-style: italic;" href="#dy">dy</a> and no more than <span style="font-style: italic;">(&nbsp;</span><a style="font-style: italic;" href="#ny">ny</a><span style="font-style: italic;"> - 1 )&nbsp;</span><span style="font-style: italic;"> * <a href="#dy">dy</a></span><span style="font-style: italic;"> - <a href="#building_wall_south">building_wall_south</a></span>. This parameter requires
364the use of&nbsp;<a href="#topography">topography</a>
365= <span style="font-style: italic;">'single_building'</span>.</td>
366</tr> <tr> <td style="vertical-align: top;"><span style="font-weight: bold;"><a name="building_wall_left"></a>building_wall_left</span></td>
367<td style="vertical-align: top;">R</td> <td style="vertical-align: top;"><span style="font-style: italic;">building centered in x-direction</span></td>
368<td>x-coordinate of the left building wall (distance between the
369left building wall and the left border of the model domain) in m.<br>
370<br>
371Currently, <span style="font-weight: bold;">building_wall_left</span>
372must be at least <span style="font-style: italic;">1
373*&nbsp;</span><a style="font-style: italic;" href="#dx">dx</a> and less than <span style="font-style: italic;">( <a href="#nx">nx</a>&nbsp;
374- 1 ) * <a href="#dx">dx</a> -&nbsp; <a href="#building_length_x">building_length_x</a></span>.
375This parameter requires the use of&nbsp;<a href="#topography">topography</a>
376= <span style="font-style: italic;">'single_building'</span>.<br>
377<br>
378The default value&nbsp;<span style="font-weight: bold;">building_wall_left</span>
379= <span style="font-style: italic;">( ( <a href="#nx">nx</a>&nbsp;+
3801 ) * <a href="#dx">dx</a> -&nbsp; <a href="#building_length_x">building_length_x</a> ) / 2</span>
381centers the building in x-direction. </td> </tr> <tr>
382<td style="vertical-align: top;"><span style="font-weight: bold;"><a name="building_wall_south"></a>building_wall_south</span></td>
383<td style="vertical-align: top;">R</td> <td style="vertical-align: top;"><span style="font-style: italic;"></span><span style="font-style: italic;">building centered in y-direction</span></td>
384<td>y-coordinate of the South building wall (distance between the
385South building wall and the South border of the model domain) in m.<br>
386<br>
387Currently, <span style="font-weight: bold;">building_wall_south</span>
388must be at least <span style="font-style: italic;">1
389*&nbsp;</span><a style="font-style: italic;" href="#dy">dy</a> and less than <span style="font-style: italic;">( <a href="#ny">ny</a>&nbsp;
390- 1 ) * <a href="#dy">dy</a> -&nbsp; <a href="#building_length_y">building_length_y</a></span>.
391This parameter requires the use of&nbsp;<a href="#topography">topography</a>
392= <span style="font-style: italic;">'single_building'</span>.<br>
393<br>
394The default value&nbsp;<span style="font-weight: bold;">building_wall_south</span>
395= <span style="font-style: italic;">( ( <a href="#ny">ny</a>&nbsp;+
3961 ) * <a href="#dy">dy</a> -&nbsp; <a href="#building_length_y">building_length_y</a> ) / 2</span>
397centers the building in y-direction. </td> </tr> <tr>
398<td style="vertical-align: top;"><span style="font-weight: bold;"><a name="cloud_droplets"></a>cloud_droplets</span><br>
399</td> <td style="vertical-align: top;">L<br> </td>
400<td style="vertical-align: top;"><span style="font-style: italic;">.F.</span><br> </td>
401<td style="vertical-align: top;">Parameter to switch on
402usage of cloud droplets.<br> <br>
403Cloud droplets require to use the particle package (<span style="font-weight: bold;">mrun</span>-option <span style="font-family: monospace;">-p particles</span>),
404so in this case a particle corresponds to a droplet. The droplet
405features (number of droplets, initial radius, etc.) can be steered with
406the&nbsp; respective particle parameters (see e.g. <a href="#chapter_4.2.html#radius">radius</a>).
407The real number of initial droplets in a grid cell is equal to the
408initial number of droplets (defined by the particle source parameters <span lang="en-GB"><font face="Thorndale, serif"> </font></span><a href="chapter_4.2.html#pst"><span lang="en-GB"><font face="Thorndale, serif">pst</font></span></a><span lang="en-GB"><font face="Thorndale, serif">, </font></span><a href="chapter_4.2.html#psl"><span lang="en-GB"><font face="Thorndale, serif">psl</font></span></a><span lang="en-GB"><font face="Thorndale, serif">, </font></span><a href="chapter_4.2.html#psr"><span lang="en-GB"><font face="Thorndale, serif">psr</font></span></a><span lang="en-GB"><font face="Thorndale, serif">, </font></span><a href="chapter_4.2.html#pss"><span lang="en-GB"><font face="Thorndale, serif">pss</font></span></a><span lang="en-GB"><font face="Thorndale, serif">, </font></span><a href="chapter_4.2.html#psn"><span lang="en-GB"><font face="Thorndale, serif">psn</font></span></a><span lang="en-GB"><font face="Thorndale, serif">, </font></span><a href="chapter_4.2.html#psb"><span lang="en-GB"><font face="Thorndale, serif">psb</font></span></a><span lang="en-GB"><font face="Thorndale, serif">, </font></span><a href="chapter_4.2.html#pdx"><span lang="en-GB"><font face="Thorndale, serif">pdx</font></span></a><span lang="en-GB"><font face="Thorndale, serif">, </font></span><a href="chapter_4.2.html#pdy"><span lang="en-GB"><font face="Thorndale, serif">pdy</font></span></a>
409<span lang="en-GB"><font face="Thorndale, serif">and
410</font></span><a href="chapter_4.2.html#pdz"><span lang="en-GB"><font face="Thorndale, serif">pdz</font></span></a><span lang="en-GB"></span><span lang="en-GB"></span>)
411times the <a href="#initial_weighting_factor">initial_weighting_factor</a>.<br>
412<br>
413In case of using cloud droplets, the default condensation scheme in
414PALM cannot be used, i.e. <a href="#cloud_physics">cloud_physics</a>
415must be set <span style="font-style: italic;">.F.</span>.<br>
416</td> </tr> <tr> <td style="vertical-align: top;"> <p><a name="cloud_physics"></a><b>cloud_physics</b></p>
417</td> <td style="vertical-align: top;">L<br> </td>
418<td style="vertical-align: top;"><span style="font-style: italic;">.F.</span></td> <td style="vertical-align: top;"> <p>Parameter to switch
419on the condensation scheme.&nbsp; </p>
420For <b>cloud_physics =</b> <span style="font-style: italic;">.TRUE.</span>, equations
421for the
422liquid water&nbsp;
423content and the liquid water potential temperature are solved instead
424of those for specific humidity and potential temperature. Note
425that a grid volume is assumed to be either completely saturated or
426completely
427unsaturated (0%-or-100%-scheme). A simple precipitation scheme can
428additionally be switched on with parameter <a href="#precipitation">precipitation</a>.
429Also cloud-top cooling by longwave radiation can be utilized (see <a href="#radiation">radiation</a>)<br> <b><br>
430cloud_physics =</b> <span style="font-style: italic;">.TRUE.
431</span>requires <a href="#moisture">moisture</a>
432=<span style="font-style: italic;"> .TRUE.</span> .<br>
433Detailed information about the condensation scheme is given in the
434description of the <a href="http://www.muk.uni-hannover.de/%7Eraasch/PALM-1/Dokumentationen/Cloud_physics/wolken.pdf">cloud
435physics module</a> (pdf-file, only in German).<br> <br>
436This condensation scheme is not allowed if cloud droplets are simulated
437explicitly (see <a href="#cloud_droplets">cloud_droplets</a>).<br>
438</td> </tr> <tr> <td style="vertical-align: top;"><span style="font-weight: bold;"><a name="conserve_volume_flow"></a>conserve_volume_flow</span></td>
439<td style="vertical-align: top;">L</td> <td style="vertical-align: top;"><span style="font-style: italic;">.F.</span></td> <td>Conservation
440of volume flow in x- and y-direction.<br> <br> <span style="font-weight: bold;">conserve_volume_flow</span>
441= <span style="font-style: italic;">.TRUE.</span>
442guarantees that the volume flow through the xz- or yz-cross-section of
443the total model domain remains constant (equal to the initial value at
444t=0) throughout the run.<br>
445</td> </tr> <tr> <td style="vertical-align: top;"> <p><a name="cut_spline_overshoot"></a><b>cut_spline_overshoot</b></p>
446</td> <td style="vertical-align: top;">L</td>
447<td style="vertical-align: top;"><span style="font-style: italic;">.T.</span></td> <td style="vertical-align: top;"> <p>Cuts off of
448so-called overshoots, which can occur with the
449upstream-spline scheme.&nbsp; </p> <p><font color="#000000">The cubic splines tend to overshoot in
450case of discontinuous changes of variables between neighbouring grid
451points.</font><font color="#ff0000"> </font><font color="#000000">This
452may lead to errors in calculating the advection tendency.</font>
453Choice
454of <b>cut_spline_overshoot</b> = <i>.TRUE.</i>
455(switched on by
456default)
457allows variable values not to exceed an interval defined by the
458respective adjacent grid points. This interval can be adjusted
459seperately for every prognostic variable (see initialization parameters
460<a href="#overshoot_limit_e">overshoot_limit_e</a>, <a href="#overshoot_limit_pt">overshoot_limit_pt</a>, <a href="#overshoot_limit_u">overshoot_limit_u</a>,
461etc.). This might be necessary in case that the
462default interval has a non-tolerable effect on the model
463results.&nbsp; </p> <p>Overshoots may also be removed
464using the parameters <a href="#ups_limit_e">ups_limit_e</a>,
465<a href="#ups_limit_pt">ups_limit_pt</a>,
466etc. as well as by applying a long-filter (see <a href="#long_filter_factor">long_filter_factor</a>).</p>
467</td> </tr> <tr> <td style="vertical-align: top;"> <p><a name="damp_level_1d"></a><b>damp_level_1d</b></p>
468</td> <td style="vertical-align: top;">R</td>
469<td style="vertical-align: top;"><span style="font-style: italic;">zu(nz+1)</span></td>
470<td style="vertical-align: top;"> <p>Height where
471the damping layer begins in the 1d-model
472(in m).&nbsp; </p> <p>This parameter is used to
473switch on a damping layer for the
4741d-model, which is generally needed for the damping of inertia
475oscillations. Damping is done by gradually increasing the value
476of the eddy diffusivities about 10% per vertical grid level
477(starting with the value at the height given by <b>damp_level_1d</b>,
478or possibly from the next grid pint above), i.e. K<sub>m</sub>(k+1)
479=
4801.1 * K<sub>m</sub>(k).
481The values of K<sub>m</sub> are limited to 10 m**2/s at
482maximum.&nbsp; <br>
483This parameter only comes into effect if the 1d-model is switched on
484for
485the initialization of the 3d-model using <a href="#initializing_actions">initializing_actions</a>
486= <span style="font-style: italic;">'set_1d-model_profiles'</span>.
487<br> </p> </td> </tr> <tr> <td style="vertical-align: top;"><a name="dissipation_1d"></a><span style="font-weight: bold;">dissipation_1d</span><br>
488</td> <td style="vertical-align: top;">C*20<br>
489</td> <td style="vertical-align: top;"><span style="font-style: italic;">'as_in_3d_</span><br style="font-style: italic;"> <span style="font-style: italic;">model'</span><br> </td>
490<td style="vertical-align: top;">Calculation method for
491the energy dissipation term in the TKE equation of the 1d-model.<br>
492<br>
493By default the dissipation is calculated as in the 3d-model using diss
494= (0.19 + 0.74 * l / l_grid) * e**1.5 / l.<br> <br>
495Setting <span style="font-weight: bold;">dissipation_1d</span>
496= <span style="font-style: italic;">'detering'</span>
497forces the dissipation to be calculated as diss = 0.064 * e**1.5 / l.<br>
498</td> </tr>
499<tr> <td style="vertical-align: top;"> <p><a name="dt"></a><b>dt</b></p> </td>
500<td style="vertical-align: top;">R</td> <td style="vertical-align: top;"><span style="font-style: italic;">variable</span></td>
501<td style="vertical-align: top;"> <p>Time step for
502the 3d-model (in s).&nbsp; </p> <p>By default, (i.e.
503if a Runge-Kutta scheme is used, see <a href="#timestep_scheme">timestep_scheme</a>)
504the value of the time step is calculating after each time step
505(following the time step criteria) and
506used for the next step.</p> <p>If the user assigns <b>dt</b>
507a value, then the time step is
508fixed to this value throughout the whole run (whether it fulfills the
509time step
510criteria or not). However, changes are allowed for restart runs,
511because <b>dt</b> can also be used as a <a href="chapter_4.2.html#dt_laufparameter">run
512parameter</a>.&nbsp; </p> <p>In case that the
513calculated time step meets the condition<br> </p> <ul>
514<p><b>dt</b> &lt; 0.00001 * <a href="chapter_4.2.html#dt_max">dt_max</a> (with dt_max
515= 20.0)</p> </ul> <p>the simulation will be
516aborted. Such situations usually arise
517in case of any numerical problem / instability which causes a
518non-realistic increase of the wind speed.&nbsp; </p> <p>A
519small time step due to a large mean horizontal windspeed
520speed may be enlarged by using a coordinate transformation (see <a href="#galilei_transformation">galilei_transformation</a>),
521in order to spare CPU time.<br> </p> <p>If the
522leapfrog timestep scheme is used (see <a href="#timestep_scheme">timestep_scheme</a>)
523a temporary time step value dt_new is calculated first, with dt_new = <a href="chapter_4.2.html#fcl_factor">cfl_factor</a>
524* dt_crit where dt_crit is the maximum timestep allowed by the CFL and
525diffusion condition. Next it is examined whether dt_new exceeds or
526falls below the
527value of the previous timestep by at
528least +5 % / -2%. If it is smaller, <span style="font-weight: bold;">dt</span>
529= dt_new is immediately used for the next timestep. If it is larger,
530then <span style="font-weight: bold;">dt </span>=
5311.02 * dt_prev
532(previous timestep) is used as the new timestep, however the time
533step is only increased if the last change of the time step is dated
534back at
535least 30 iterations. If dt_new is located in the interval mentioned
536above, then dt
537does not change at all. By doing so, permanent time step changes as
538well as large
539sudden changes (increases) in the time step are avoided.</p> </td>
540</tr> <tr> <td style="vertical-align: top;">
541<p><a name="dt_pr_1d"></a><b>dt_pr_1d</b></p>
542</td> <td style="vertical-align: top;">R</td>
543<td style="vertical-align: top;"><span style="font-style: italic;">9999999.9</span></td>
544<td style="vertical-align: top;"> <p>Temporal
545interval of vertical profile output of the 1D-model
546(in s).&nbsp; </p> <p>Data are written in ASCII
547format to file <a href="chapter_3.4.html#LIST_PROFIL_1D">LIST_PROFIL_1D</a>.
548This parameter is only in effect if the 1d-model has been switched on
549for the
550initialization of the 3d-model with <a href="#initializing_actions">initializing_actions</a>
551= <span style="font-style: italic;">'set_1d-model_profiles'</span>.</p>
552</td> </tr> <tr> <td style="vertical-align: top;"> <p><a name="dt_run_control_1d"></a><b>dt_run_control_1d</b></p>
553</td> <td style="vertical-align: top;">R</td>
554<td style="vertical-align: top;"><span style="font-style: italic;">60.0</span></td> <td style="vertical-align: top;"> <p>Temporal interval of
555runtime control output of the 1d-model
556(in s).&nbsp; </p> <p>Data are written in ASCII
557format to file <a href="chapter_3.4.html#RUN_CONTROL">RUN_CONTROL</a>.
558This parameter is only in effect if the 1d-model is switched on for the
559initialization of the 3d-model with <a href="#initializing_actions">initializing_actions</a>
560= <span style="font-style: italic;">'set_1d-model_profiles'</span>.</p>
561</td> </tr> <tr> <td style="vertical-align: top;"> <p><a name="dx"></a><b>dx</b></p>
562</td> <td style="vertical-align: top;">R</td>
563<td style="vertical-align: top;"><span style="font-style: italic;">1.0</span></td> <td style="vertical-align: top;"> <p>Horizontal grid
564spacing along the x-direction (in m).&nbsp; </p> <p>Along
565x-direction only a constant grid spacing is allowed.</p> </td>
566</tr> <tr> <td style="vertical-align: top;">
567<p><a name="dy"></a><b>dy</b></p>
568</td> <td style="vertical-align: top;">R</td>
569<td style="vertical-align: top;"><span style="font-style: italic;">1.0</span></td> <td style="vertical-align: top;"> <p>Horizontal grid
570spacing along the x-direction (in m).&nbsp; </p> <p>Along
571x-direction only a constant grid spacing is allowed.</p> </td>
572</tr> <tr> <td style="vertical-align: top;">
573<p><a name="dz"></a><b>dz</b></p>
574</td> <td style="vertical-align: top;">R</td>
575<td style="vertical-align: top;"><br> </td> <td style="vertical-align: top;"> <p>Vertical grid
576spacing (in m).&nbsp; </p> <p>This parameter must be
577assigned by the user, because no
578default value is given.<br> </p> <p>By default, the
579model uses constant grid spacing along z-direction, but it can be
580stretched using the parameters <a href="#dz_stretch_level">dz_stretch_level</a>
581and <a href="#dz_stretch_factor">dz_stretch_factor</a>.
582In case of stretching, a maximum allowed grid spacing can be given by <a href="#dz_max">dz_max</a>.<br> </p> <p>Assuming
583a constant <span style="font-weight: bold;">dz</span>,
584the scalar levels (zu) are calculated directly by:&nbsp; </p>
585<ul> <p>zu(0) = - dz * 0.5&nbsp; <br>
586zu(1) = dz * 0.5</p> </ul> <p>The w-levels lie
587half between them:&nbsp; </p> <ul> <p>zw(k) =
588( zu(k) + zu(k+1) ) * 0.5</p> </ul> </td> </tr>
589<tr><td style="vertical-align: top;"><a name="dz_max"></a><span style="font-weight: bold;">dz_max</span></td><td style="vertical-align: top;">R</td><td style="vertical-align: top;"><span style="font-style: italic;">9999999.9</span></td><td style="vertical-align: top;">Allowed maximum vertical grid
590spacing (in m).<br><br>If the vertical grid is stretched
591(see <a href="#dz_stretch_factor">dz_stretch_factor</a>
592and <a href="#dz_stretch_level">dz_stretch_level</a>),
593<span style="font-weight: bold;">dz_max</span> can
594be used to limit the vertical grid spacing.</td></tr><tr>
595<td style="vertical-align: top;"> <p><a name="dz_stretch_factor"></a><b>dz_stretch_factor</b></p>
596</td> <td style="vertical-align: top;">R</td>
597<td style="vertical-align: top;"><span style="font-style: italic;">1.08</span></td> <td style="vertical-align: top;"> <p>Stretch factor for a
598vertically stretched grid (see <a href="#dz_stretch_level">dz_stretch_level</a>).&nbsp;
599</p> <p>The stretch factor should not exceed a value of
600approx. 1.10 -
6011.12, otherwise the discretization errors due to the stretched grid not
602negligible any more. (refer Kalnay de Rivas)</p> </td> </tr>
603<tr> <td style="vertical-align: top;"> <p><a name="dz_stretch_level"></a><b>dz_stretch_level</b></p>
604</td> <td style="vertical-align: top;">R</td>
605<td style="vertical-align: top;"><span style="font-style: italic;">100000.0</span><br> </td>
606<td style="vertical-align: top;"> <p>Height level
607above which the grid is to be stretched
608vertically (in m).&nbsp; </p> <p>The vertical grid
609spacings <a href="#dz">dz</a>
610above this level are calculated as&nbsp; </p> <ul> <p><b>dz</b>(k+1)
611= <b>dz</b>(k) * <a href="#dz_stretch_factor">dz_stretch_factor</a></p>
612</ul> <p>and used as spacings for the scalar levels (zu).
613The
614w-levels are then defined as:&nbsp; </p> <ul> <p>zw(k)
615= ( zu(k) + zu(k+1) ) * 0.5</p> </ul> </td> </tr>
616<tr> <td style="vertical-align: top;"><span style="font-weight: bold;"><a name="e_min"></a>e_min</span></td>
617<td style="vertical-align: top;">R</td> <td style="vertical-align: top;"><span style="font-style: italic;">0.0</span></td> <td>Minimum
618subgrid-scale TKE in m<sup>2</sup>s<sup>-2</sup>.<br>
619<br>This
620option&nbsp;adds artificial viscosity to the flow by ensuring that
621the
622subgrid-scale TKE does not fall below the minimum threshold <span style="font-weight: bold;">e_min</span>.</td> </tr>
623<tr> <td style="vertical-align: top;"> <p><a name="end_time_1d"></a><b>end_time_1d</b></p>
624</td> <td style="vertical-align: top;">R</td>
625<td style="vertical-align: top;"><span style="font-style: italic;">864000.0</span><br> </td>
626<td style="vertical-align: top;"> <p>Time to be
627simulated for the 1d-model (in s).&nbsp; </p> <p>The
628default value corresponds to a simulated time of 10 days.
629Usually, after such a period the inertia oscillations have completely
630decayed and the solution of the 1d-model can be regarded as stationary
631(see <a href="#damp_level_1d">damp_level_1d</a>).
632This parameter is only in effect if the 1d-model is switched on for the
633initialization of the 3d-model with <a href="#initializing_actions">initializing_actions</a>
634= <span style="font-style: italic;">'set_1d-model_profiles'</span>.</p>
635</td> </tr> <tr> <td style="vertical-align: top;"> <p><a name="fft_method"></a><b>fft_method</b></p>
636</td> <td style="vertical-align: top;">C * 20</td>
637<td style="vertical-align: top;"><span style="font-style: italic;">'system-</span><br style="font-style: italic;"> <span style="font-style: italic;">specific'</span></td>
638<td style="vertical-align: top;"> <p>FFT-method to
639be used.<br> </p> <p><br>
640The fast fourier transformation (FFT) is used for solving the
641perturbation pressure equation with a direct method (see <a href="chapter_4.2.html#psolver">psolver</a>)
642and for calculating power spectra (see optional software packages,
643section <a href="chapter_4.2.html#spectra_package">4.2</a>).</p>
644<p><br>
645By default, system-specific, optimized routines from external
646vendor libraries are used. However, these are available only on certain
647computers and there are more or less severe restrictions concerning the
648number of gridpoints to be used with them.<br> </p> <p>There
649are two other PALM internal methods available on every
650machine (their respective source code is part of the PALM source code):</p>
651<p>1.: The <span style="font-weight: bold;">Temperton</span>-method
652from Clive Temperton (ECWMF) which is computationally very fast and
653switched on with <b>fft_method</b> = <span style="font-style: italic;">'temperton-algorithm'</span>.
654The number of horizontal gridpoints (nx+1, ny+1) to be used with this
655method must be composed of prime factors 2, 3 and 5.<br> </p>
6562.: The <span style="font-weight: bold;">Singleton</span>-method
657which is very slow but has no restrictions concerning the number of
658gridpoints to be used with, switched on with <b>fft_method</b>
659= <span style="font-style: italic;">'singleton-algorithm'</span>.
660</td> </tr> <tr> <td style="vertical-align: top;"> <p><a name="galilei_transformation"></a><b>galilei_transformation</b></p>
661</td> <td style="vertical-align: top;">L</td>
662<td style="vertical-align: top;"><i>.F.</i></td>
663<td style="vertical-align: top;">Application of a
664Galilei-transformation to the
665coordinate
666system of the model.<br><p>With <b>galilei_transformation</b>
667= <i>.T.,</i> a so-called
668Galilei-transformation is switched on which ensures that the coordinate
669system of the model is moved along with the geostrophical wind.
670Alternatively, the model domain can be moved along with the averaged
671horizontal wind (see <a href="#use_ug_for_galilei_tr">use_ug_for_galilei_tr</a>,
672this can and will naturally change in time). With this method,
673numerical inaccuracies of the Piascek - Williams - scheme (concerns in
674particular the momentum advection) are minimized. Beyond that, in the
675majority of cases the lower relative velocities in the moved system
676permit a larger time step (<a href="#dt">dt</a>).
677Switching the transformation on is only worthwhile if the geostrophical
678wind (ug, vg)
679and the averaged horizontal wind clearly deviate from the value 0. In
680each case, the distance the coordinate system has been moved is written
681to the file <a href="chapter_3.4.html#RUN_CONTROL">RUN_CONTROL</a>.&nbsp;
682</p> <p>Non-cyclic lateral boundary conditions (see <a href="#bc_lr">bc_lr</a>
683and <a href="#bc_ns">bc_ns</a>), the specification
684of a gestrophic
685wind that is not constant with height
686as well as e.g. stationary inhomogeneities at the bottom boundary do
687not allow the use of this transformation.</p> </td> </tr>
688<tr> <td style="vertical-align: top;"> <p><a name="grid_matching"></a><b>grid_matching</b></p>
689</td> <td style="vertical-align: top;">C * 6</td>
690<td style="vertical-align: top;"><span style="font-style: italic;">'match'</span></td> <td style="vertical-align: top;">Variable to adjust the
691subdomain
692sizes in parallel runs.<br> <br>
693For <b>grid_matching</b> = <span style="font-style: italic;">'strict'</span>,
694the subdomains are forced to have an identical
695size on all processors. In this case the processor numbers in the
696respective directions of the virtual processor net must fulfill certain
697divisor conditions concerning the grid point numbers in the three
698directions (see <a href="#nx">nx</a>, <a href="#ny">ny</a>
699and <a href="#nz">nz</a>).
700Advantage of this method is that all PEs bear the same computational
701load.<br> <br>
702There is no such restriction by default, because then smaller
703subdomains are allowed on those processors which
704form the right and/or north boundary of the virtual processor grid. On
705all other processors the subdomains are of same size. Whether smaller
706subdomains are actually used, depends on the number of processors and
707the grid point numbers used. Information about the respective settings
708are given in file <a href="file:///home/raasch/public_html/PALM_group/home/raasch/public_html/PALM_group/doc/app/chapter_3.4.html#RUN_CONTROL">RUN_CONTROL</a>.<br>
709<br>
710When using a multi-grid method for solving the Poisson equation (see <a href="http://www.muk.uni-hannover.de/%7Eraasch/PALM_group/doc/app/chapter_4.2.html#psolver">psolver</a>)
711only <b>grid_matching</b> = <span style="font-style: italic;">'strict'</span>
712is allowed.<br> <br> <b>Note:</b><br>
713In some cases for small processor numbers there may be a very bad load
714balancing among the
715processors which may reduce the performance of the code.</td> </tr>
716<tr> <td style="vertical-align: top;"><a name="inflow_disturbance_begin"></a><b>inflow_disturbance_<br>
717begin</b></td> <td style="vertical-align: top;">I</td>
718<td style="vertical-align: top;"><span style="font-style: italic;">MIN(10,</span><br style="font-style: italic;"> <span style="font-style: italic;">nx/2 or ny/2)</span></td>
719<td style="vertical-align: top;">Lower
720limit of the horizontal range for which random perturbations are to be
721imposed on the horizontal velocity field (gridpoints).<br> <br>
722If non-cyclic lateral boundary conditions are used (see <a href="#bc_lr">bc_lr</a>
723or <a href="#bc_ns">bc_ns</a>),
724this parameter gives the gridpoint number (counted horizontally from
725the inflow)&nbsp; from which on perturbations are imposed on the
726horizontal velocity field. Perturbations must be switched on with
727parameter <a href="chapter_4.2.html#create_disturbances">create_disturbances</a>.</td>
728</tr> <tr> <td style="vertical-align: top;"><a name="inflow_disturbance_end"></a><b>inflow_disturbance_<br>
729end</b></td> <td style="vertical-align: top;">I</td>
730<td style="vertical-align: top;"><span style="font-style: italic;">MIN(100,</span><br style="font-style: italic;"> <span style="font-style: italic;">3/4*nx or</span><br style="font-style: italic;"> <span style="font-style: italic;">3/4*ny)</span></td> <td style="vertical-align: top;">Upper
731limit of the horizontal range for which random perturbations are
732to be imposed on the horizontal velocity field (gridpoints).<br> <br>
733If non-cyclic lateral boundary conditions are used (see <a href="#bc_lr">bc_lr</a>
734or <a href="#bc_ns">bc_ns</a>),
735this parameter gives the gridpoint number (counted horizontally from
736the inflow)&nbsp; unto which perturbations are imposed on the
737horizontal
738velocity field. Perturbations must be switched on with parameter <a href="chapter_4.2.html#create_disturbances">create_disturbances</a>.</td>
739</tr> <tr> <td style="vertical-align: top;">
740<p><a name="initializing_actions"></a><b>initializing_actions</b></p>
741</td> <td style="vertical-align: top;">C * 100</td>
742<td style="vertical-align: top;"><br> </td> <td style="vertical-align: top;"> <p style="font-style: normal;">Initialization actions
743to be carried out.&nbsp; </p> <p style="font-style: normal;">This parameter does not have a
744default value and therefore must be assigned with each model run. For
745restart runs <b>initializing_actions</b> = <span style="font-style: italic;">'read_restart_data'</span>
746must be set. For the initial run of a job chain the following values
747are allowed:&nbsp; </p> <p style="font-style: normal;"><span style="font-style: italic;">'set_constant_profiles'</span>
748</p> <ul> <p>A horizontal wind profile consisting
749of linear sections (see <a href="#ug_surface">ug_surface</a>,
750<a href="#ug_vertical_gradient">ug_vertical_gradient</a>,
751<a href="#ug_vertical_gradient_level">ug_vertical_gradient_level</a>
752and <a href="#vg_surface">vg_surface</a>, <a href="#vg_vertical_gradient">vg_vertical_gradient</a>,
753<a href="#vg_vertical_gradient_level">vg_vertical_gradient_level</a>,
754respectively) as well as a vertical temperature (humidity) profile
755consisting of
756linear sections (see <a href="#pt_surface">pt_surface</a>,
757<a href="#pt_vertical_gradient">pt_vertical_gradient</a>,
758<a href="#q_surface">q_surface</a>
759and <a href="#q_vertical_gradient">q_vertical_gradient</a>)
760are assumed as initial profiles. The subgrid-scale TKE is set to 0 but K<sub>m</sub>
761and K<sub>h</sub> are set to very small values because
762otherwise no TKE
763would be generated.</p> </ul> <p style="font-style: italic;">'set_1d-model_profiles' </p>
764<ul> <p>The arrays of the 3d-model are initialized with
765the
766(stationary) solution of the 1d-model. These are the variables e, kh,
767km, u, v and with Prandtl layer switched on rif, us, usws, vsws. The
768temperature (humidity) profile consisting of linear sections is set as
769for 'set_constant_profiles' and assumed as constant in time within the
7701d-model. For steering of the 1d-model a set of parameters with suffix
771"_1d" (e.g. <a href="#end_time_1d">end_time_1d</a>,
772<a href="#damp_level_1d">damp_level_1d</a>)
773is available.</p> </ul> <p><span style="font-style: italic;">'by_user'</span></p><p style="margin-left: 40px;">The initialization of the arrays
774of the 3d-model is under complete control of the user and has to be
775done in routine <a href="chapter_3.5.1.html#user_init_3d_model">user_init_3d_model</a>
776of the user-interface.<span style="font-style: italic;"></span></p><p><span style="font-style: italic;">'initialize_vortex'</span>
777</p> <div style="margin-left: 40px;">The initial
778velocity field of the
7793d-model corresponds to a
780Rankine-vortex with vertical axis. This setting may be used to test
781advection schemes. Free-slip boundary conditions for u and v (see <a href="#bc_uv_b">bc_uv_b</a>, <a href="#bc_uv_t">bc_uv_t</a>)
782are necessary. In order not to distort the vortex, an initial
783horizontal wind profile constant
784with height is necessary (to be set by <b>initializing_actions</b>
785= <span style="font-style: italic;">'set_constant_profiles'</span>)
786and some other conditions have to be met (neutral stratification,
787diffusion must be
788switched off, see <a href="#km_constant">km_constant</a>).
789The center of the vortex is located at jc = (nx+1)/2. It
790extends from k = 0 to k = nz+1. Its radius is 8 * <a href="#dx">dx</a>
791and the exponentially decaying part ranges to 32 * <a href="#dx">dx</a>
792(see init_rankine.f90). </div> <p><span style="font-style: italic;">'initialize_ptanom'</span>
793</p> <ul> <p>A 2d-Gauss-like shape disturbance
794(x,y) is added to the
795initial temperature field with radius 10.0 * <a href="#dx">dx</a>
796and center at jc = (nx+1)/2. This may be used for tests of scalar
797advection schemes
798(see <a href="#scalar_advec">scalar_advec</a>).
799Such tests require a horizontal wind profile constant with hight and
800diffusion
801switched off (see <span style="font-style: italic;">'initialize_vortex'</span>).
802Additionally, the buoyancy term
803must be switched of in the equation of motion&nbsp; for w (this
804requires the user to comment out the call of <span style="font-family: monospace;">buoyancy</span> in the
805source code of <span style="font-family: monospace;">prognostic_equations.f90</span>).</p>
806</ul> <p style="font-style: normal;">Values may be
807combined, e.g. <b>initializing_actions</b> = <span style="font-style: italic;">'set_constant_profiles
808initialize_vortex'</span>, but the values of <span style="font-style: italic;">'set_constant_profiles'</span>,
809<span style="font-style: italic;">'set_1d-model_profiles'</span>
810, and <span style="font-style: italic;">'by_user'</span>
811must not be given at the same time.</p> <p style="font-style: italic;"> </p> </td> </tr>
812<tr> <td style="vertical-align: top;"> <p><a name="km_constant"></a><b>km_constant</b></p>
813</td> <td style="vertical-align: top;">R</td>
814<td style="vertical-align: top;"><i>variable<br>
815(computed from TKE)</i></td> <td style="vertical-align: top;"> <p>Constant eddy
816diffusivities are used (laminar
817simulations).&nbsp; </p> <p>If this parameter is
818specified, both in the 1d and in the
8193d-model constant values for the eddy diffusivities are used in
820space and time with K<sub>m</sub> = <b>km_constant</b>
821and K<sub>h</sub> = K<sub>m</sub> / <a href="chapter_4.2.html#prandtl_number">prandtl_number</a>.
822The prognostic equation for the subgrid-scale TKE is switched off.
823Constant eddy diffusivities are only allowed with the Prandtl layer (<a href="#prandtl_layer">prandtl_layer</a>)
824switched off.</p> </td> </tr> <tr> <td style="vertical-align: top;"> <p><a name="km_damp_max"></a><b>km_damp_max</b></p>
825</td> <td style="vertical-align: top;">R</td>
826<td style="vertical-align: top;"><span style="font-style: italic;">0.5*(dx
827or dy)</span></td> <td style="vertical-align: top;">Maximum
828diffusivity used for filtering the velocity field in the vicinity of
829the outflow (in m<sup>2</sup>/s).<br> <br>
830When using non-cyclic lateral boundaries (see <a href="#bc_lr">bc_lr</a>
831or <a href="#bc_ns">bc_ns</a>),
832a smoothing has to be applied to the
833velocity field in the vicinity of the outflow in order to suppress any
834reflections of outgoing disturbances. Smoothing is done by increasing
835the eddy diffusivity along the horizontal direction which is
836perpendicular to the outflow boundary. Only velocity components
837parallel to the outflow boundary are filtered (e.g. v and w, if the
838outflow is along x). Damping is applied from the bottom to the top of
839the domain.<br> <br>
840The horizontal range of the smoothing is controlled by <a href="#outflow_damping_width">outflow_damping_width</a>
841which defines the number of gridpoints (counted from the outflow
842boundary) from where on the smoothing is applied. Starting from that
843point, the eddy diffusivity is linearly increased (from zero to its
844maximum value given by <span style="font-weight: bold;">km_damp_max</span>)
845until half of the damping range width, from where it remains constant
846up to the outflow boundary. If at a certain grid point the eddy
847diffusivity calculated from the flow field is larger than as described
848above, it is used instead.<br> <br>
849The default value of <span style="font-weight: bold;">km_damp_max</span>
850has been empirically proven to be sufficient.</td> </tr> <tr>
851<td style="vertical-align: top;"> <p><a name="long_filter_factor"></a><b>long_filter_factor</b></p>
852</td> <td style="vertical-align: top;">R</td>
853<td style="vertical-align: top;"><i>0.0</i></td>
854<td style="vertical-align: top;"> <p>Filter factor
855for the so-called Long-filter.<br> </p> <p><br>
856This filter very efficiently
857eliminates 2-delta-waves sometimes cauesed by the upstream-spline
858scheme (see Mahrer and
859Pielke, 1978: Mon. Wea. Rev., 106, 818-830). It works in all three
860directions in space. A value of <b>long_filter_factor</b>
861= <i>0.01</i>
862sufficiently removes the small-scale waves without affecting the
863longer waves.<br> </p> <p>By default, the filter is
864switched off (= <i>0.0</i>).
865It is exclusively applied to the tendencies calculated by the
866upstream-spline scheme (see <a href="#momentum_advec">momentum_advec</a>
867and <a href="#scalar_advec">scalar_advec</a>),
868not to the prognostic variables themselves. At the bottom and top
869boundary of the model domain the filter effect for vertical
8702-delta-waves is reduced. There, the amplitude of these waves is only
871reduced by approx. 50%, otherwise by nearly 100%.&nbsp; <br>
872Filter factors with values &gt; <i>0.01</i> also
873reduce the amplitudes
874of waves with wavelengths longer than 2-delta (see the paper by Mahrer
875and
876Pielke, quoted above). </p> </td> </tr> <tr>
877<td style="vertical-align: top;"><a name="mixing_length_1d"></a><span style="font-weight: bold;">mixing_length_1d</span><br>
878</td> <td style="vertical-align: top;">C*20<br>
879</td> <td style="vertical-align: top;"><span style="font-style: italic;">'as_in_3d_</span><br style="font-style: italic;"> <span style="font-style: italic;">model'</span><br> </td>
880<td style="vertical-align: top;">Mixing length used in the
8811d-model.<br> <br>
882By default the mixing length is calculated as in the 3d-model (i.e. it
883depends on the grid spacing).<br> <br>
884By setting <span style="font-weight: bold;">mixing_length_1d</span>
885= <span style="font-style: italic;">'blackadar'</span>,
886the so-called Blackadar mixing length is used (l = kappa * z / ( 1 +
887kappa * z / lambda ) with the limiting value lambda = 2.7E-4 * u_g / f).<br>
888</td> </tr> <tr> <td style="vertical-align: top;"> <p><a name="moisture"></a><b>moisture</b></p>
889</td> <td style="vertical-align: top;">L</td>
890<td style="vertical-align: top;"><i>.F.</i></td>
891<td style="vertical-align: top;"> <p>Parameter to
892switch on the prognostic equation for specific
893humidity q.<br> </p> <p>The initial vertical
894profile of q can be set via parameters <a href="chapter_4.1.html#q_surface">q_surface</a>, <a href="chapter_4.1.html#q_vertical_gradient">q_vertical_gradient</a>
895and <a href="chapter_4.1.html#q_vertical_gradient_level">q_vertical_gradient_level</a>.&nbsp;
896Boundary conditions can be set via <a href="chapter_4.1.html#q_surface_initial_change">q_surface_initial_change</a>
897and <a href="chapter_4.1.html#surface_waterflux">surface_waterflux</a>.<br>
898</p>
899If the condensation scheme is switched on (<a href="chapter_4.1.html#cloud_physics">cloud_physics</a>
900= .TRUE.), q becomes the total liquid water content (sum of specific
901humidity and liquid water content).</td> </tr>
902<tr> <td style="vertical-align: top;"> <p><a name="momentum_advec"></a><b>momentum_advec</b></p>
903</td> <td style="vertical-align: top;">C * 10</td>
904<td style="vertical-align: top;"><i>'pw-scheme'</i></td>
905<td style="vertical-align: top;"> <p>Advection
906scheme to be used for the momentum equations.<br> <br>
907The user can choose between the following schemes:<br>
908&nbsp;<br> <br> <span style="font-style: italic;">'pw-scheme'</span><br>
909</p> <div style="margin-left: 40px;">The scheme of
910Piascek and
911Williams (1970, J. Comp. Phys., 6,
912392-405) with central differences in the form C3 is used.<br>
913If intermediate Euler-timesteps are carried out in case of <a href="#timestep_scheme">timestep_scheme</a>
914= <span style="font-style: italic;">'leapfrog+euler'</span>
915the
916advection scheme is - for the Euler-timestep - automatically switched
917to an upstream-scheme.<br> </div> <p> </p> <p><span style="font-style: italic;">'ups-scheme'</span><br>
918</p> <div style="margin-left: 40px;">The
919upstream-spline scheme is
920used
921(see Mahrer and Pielke,
9221978: Mon. Wea. Rev., 106, 818-830). In opposite to the
923Piascek-Williams scheme, this is characterized by much better numerical
924features (less numerical diffusion, better preservation of flow
925structures, e.g. vortices), but computationally it is much more
926expensive. In
927addition, the use of the Euler-timestep scheme is mandatory (<a href="#timestep_scheme">timestep_scheme</a>
928= <span style="font-style: italic;">'</span><i>euler'</i>),
929i.e. the
930timestep accuracy is only of first order.
931For this reason the advection of scalar variables (see <a href="#scalar_advec">scalar_advec</a>)
932should then also be carried out with the upstream-spline scheme,
933because otherwise the scalar variables would
934be subject to large numerical diffusion due to the upstream
935scheme.&nbsp; </div> <p style="margin-left: 40px;">Since
936the cubic splines used tend
937to overshoot under
938certain circumstances, this effect must be adjusted by suitable
939filtering and smoothing (see <a href="#cut_spline_overshoot">cut_spline_overshoot</a>,
940<a href="#long_filter_factor">long_filter_factor</a>,
941<a href="#ups_limit_pt">ups_limit_pt</a>, <a href="#ups_limit_u">ups_limit_u</a>, <a href="#ups_limit_v">ups_limit_v</a>, <a href="#ups_limit_w">ups_limit_w</a>).
942This is always neccessary for runs with stable stratification,
943even if this stratification appears only in parts of the model domain.<br>
944</p> <div style="margin-left: 40px;">With stable
945stratification the
946upstream-spline scheme also
947produces gravity waves with large amplitude, which must be
948suitably damped (see <a href="chapter_4.2.html#rayleigh_damping_factor">rayleigh_damping_factor</a>).<br>
949<br> <span style="font-weight: bold;">Important: </span>The&nbsp;
950upstream-spline scheme is not implemented for humidity and passive
951scalars (see <a href="#moisture">moisture</a>
952and <a href="#passive_scalar">passive_scalar</a>)
953and requires the use of a 2d-domain-decomposition. The last conditions
954severely restricts code optimization on several machines leading to
955very long execution times! The scheme is also not allowed for
956non-cyclic lateral boundary conditions (see <a href="#bc_lr">bc_lr</a>
957and <a href="#bc_ns">bc_ns</a>).</div> </td>
958</tr> <tr> <td style="vertical-align: top;"><a name="netcdf_precision"></a><span style="font-weight: bold;">netcdf_precision</span><br>
959</td> <td style="vertical-align: top;">C*20<br>
960(10)<br> </td> <td style="vertical-align: top;"><span style="font-style: italic;">single preci-</span><br style="font-style: italic;"> <span style="font-style: italic;">sion for all</span><br style="font-style: italic;"> <span style="font-style: italic;">output quan-</span><br style="font-style: italic;"> <span style="font-style: italic;">tities</span><br> </td>
961<td style="vertical-align: top;">Defines the accuracy of
962the NetCDF output.<br> <br>
963By default, all NetCDF output data (see <a href="chapter_4.2.html#data_output_format">data_output_format</a>)
964have single precision&nbsp; (4 byte) accuracy. Double precision (8
965byte) can be choosen alternatively.<br>
966Accuracy for the different output data (cross sections, 3d-volume data,
967spectra, etc.) can be set independently.<br> <span style="font-style: italic;">'&lt;out&gt;_NF90_REAL4'</span>
968(single precision) or <span style="font-style: italic;">'&lt;out&gt;_NF90_REAL8'</span>
969(double precision) are the two principally allowed values for <span style="font-weight: bold;">netcdf_precision</span>,
970where the string <span style="font-style: italic;">'&lt;out&gt;'
971</span>can be chosen out of the following list:<br> <br>
972<table style="text-align: left; width: 284px; height: 234px;" border="1" cellpadding="2" cellspacing="2"> <tbody>
973<tr> <td style="vertical-align: top;"><span style="font-style: italic;">'xy'</span><br> </td>
974<td style="vertical-align: top;">horizontal cross section<br>
975</td> </tr> <tr> <td style="vertical-align: top;"><span style="font-style: italic;">'xz'</span><br> </td>
976<td style="vertical-align: top;">vertical (xz) cross
977section<br> </td> </tr> <tr> <td style="vertical-align: top;"><span style="font-style: italic;">'yz'</span><br> </td>
978<td style="vertical-align: top;">vertical (yz) cross
979section<br> </td> </tr> <tr> <td style="vertical-align: top;"><span style="font-style: italic;">'2d'</span><br> </td>
980<td style="vertical-align: top;">all cross sections<br>
981</td> </tr> <tr> <td style="vertical-align: top;"><span style="font-style: italic;">'3d'</span><br> </td>
982<td style="vertical-align: top;">volume data<br> </td>
983</tr> <tr> <td style="vertical-align: top;"><span style="font-style: italic;">'pr'</span><br> </td>
984<td style="vertical-align: top;">vertical profiles<br>
985</td> </tr> <tr> <td style="vertical-align: top;"><span style="font-style: italic;">'ts'</span><br> </td>
986<td style="vertical-align: top;">time series, particle
987time series<br> </td> </tr> <tr> <td style="vertical-align: top;"><span style="font-style: italic;">'sp'</span><br> </td>
988<td style="vertical-align: top;">spectra<br> </td>
989</tr> <tr> <td style="vertical-align: top;"><span style="font-style: italic;">'prt'</span><br> </td>
990<td style="vertical-align: top;">particles<br> </td>
991</tr> <tr> <td style="vertical-align: top;"><span style="font-style: italic;">'all'</span><br> </td>
992<td style="vertical-align: top;">all output quantities<br>
993</td> </tr> </tbody> </table> <br> <span style="font-weight: bold;">Example:</span><br>
994If all cross section data and the particle data shall be output in
995double precision and all other quantities in single precision, then <span style="font-weight: bold;">netcdf_precision</span> = <span style="font-style: italic;">'2d_NF90_REAL8'</span>, <span style="font-style: italic;">'prt_NF90_REAL8'</span>
996has to be assigned.<br> </td> </tr>
997<tr> <td style="vertical-align: top;"> <p><a name="npex"></a><b>npex</b></p> </td>
998<td style="vertical-align: top;">I</td> <td style="vertical-align: top;"><br> </td> <td style="vertical-align: top;"> <p>Number of processors
999along x-direction of the virtual
1000processor
1001net.&nbsp; </p> <p>For parallel runs, the total
1002number of processors to be used
1003is given by
1004the <span style="font-weight: bold;">mrun</span>
1005option <a href="http://www.muk.uni-hannover.de/software/mrun_beschreibung.html#Opt-X">-X</a>.
1006By default, depending on the type of the parallel computer, PALM
1007generates a 1d processor
1008net (domain decomposition along x, <span style="font-weight: bold;">npey</span>
1009= <span style="font-style: italic;">1</span>) or a
10102d-net (this is
1011favored on machines with fast communication network). In case of a
10122d-net, it is tried to make it more or less square-shaped. If, for
1013example, 16 processors are assigned (-X 16), a 4 * 4 processor net is
1014generated (<span style="font-weight: bold;">npex</span>
1015= <span style="font-style: italic;">4</span>, <span style="font-weight: bold;">npey</span>
1016= <span style="font-style: italic;">4</span>).
1017This choice is optimal for square total domains (<a href="#nx">nx</a>
1018= <a href="#ny">ny</a>),
1019since then the number of ghost points at the lateral boundarys of
1020the subdomains is minimal. If <span style="font-weight: bold;">nx</span>
1021nd <span style="font-weight: bold;">ny</span>
1022differ extremely, the
1023processor net should be manually adjusted using adequate values for <span style="font-weight: bold;">npex</span> and <span style="font-weight: bold;">npey</span>.&nbsp; </p>
1024<p><b>Important:</b> The value of <span style="font-weight: bold;">npex</span> * <span style="font-weight: bold;">npey</span> must exactly
1025correspond to the
1026value assigned by the <span style="font-weight: bold;">mrun</span>-option
1027<tt>-X</tt>.
1028Otherwise the model run will abort with a corresponding error
1029message.&nbsp; <br>
1030Additionally, the specification of <span style="font-weight: bold;">npex</span>
1031and <span style="font-weight: bold;">npey</span>
1032may of course
1033override the default setting for the domain decomposition (1d or 2d)
1034which may have a significant (negative) effect on the code performance.
1035</p> </td> </tr> <tr> <td style="vertical-align: top;"> <p><a name="npey"></a><b>npey</b></p>
1036</td> <td style="vertical-align: top;">I</td>
1037<td style="vertical-align: top;"><br> </td> <td style="vertical-align: top;"> <p>Number of processors
1038along y-direction of the virtual
1039processor
1040net.&nbsp; </p> <p>For further information see <a href="#npex">npex</a>.</p> </td> </tr>
1041<tr> <td style="vertical-align: top;"> <p><a name="nsor_ini"></a><b>nsor_ini</b></p>
1042</td> <td style="vertical-align: top;">I</td>
1043<td style="vertical-align: top;"><i>100</i></td>
1044<td style="vertical-align: top;"> <p>Initial number
1045of iterations with the SOR algorithm.&nbsp; </p> <p>This
1046parameter is only effective if the SOR algorithm was
1047selected as the pressure solver scheme (<a href="chapter_4.2.html#psolver">psolver</a>
1048= <span style="font-style: italic;">'sor'</span>)
1049and specifies the
1050number of initial iterations of the SOR
1051scheme (at t = 0). The number of subsequent iterations at the following
1052timesteps is determined
1053with the parameter <a href="#nsor">nsor</a>.
1054Usually <b>nsor</b> &lt; <b>nsor_ini</b>,
1055since in each case
1056subsequent calls to <a href="chapter_4.2.html#psolver">psolver</a>
1057use the solution of the previous call as initial value. Suitable
1058test runs should determine whether sufficient convergence of the
1059solution is obtained with the default value and if necessary the value
1060of <b>nsor_ini</b> should be changed.</p> </td>
1061</tr> <tr> <td style="vertical-align: top;">
1062<p><a name="nx"></a><b>nx</b></p>
1063</td> <td style="vertical-align: top;">I</td>
1064<td style="vertical-align: top;"><br> </td> <td style="vertical-align: top;"> <p>Number of grid
1065points in x-direction.&nbsp; </p> <p>A value for this
1066parameter must be assigned. Since the lower
1067array bound in PALM
1068starts with i = 0, the actual number of grid points is equal to <b>nx+1</b>.
1069In case of cyclic boundary conditions along x, the domain size is (<b>nx+1</b>)*
1070<a href="#dx">dx</a>.</p> <p>For
1071parallel runs, in case of <a href="#grid_matching">grid_matching</a>
1072= <span style="font-style: italic;">'strict'</span>,
1073<b>nx+1</b> must
1074be an integral multiple
1075of the processor numbers (see <a href="#npex">npex</a>
1076and <a href="#npey">npey</a>)
1077along x- as well as along y-direction (due to data
1078transposition restrictions).</p> </td> </tr> <tr>
1079<td style="vertical-align: top;"> <p><a name="ny"></a><b>ny</b></p>
1080</td> <td style="vertical-align: top;">I</td>
1081<td style="vertical-align: top;"><br> </td> <td style="vertical-align: top;"> <p>Number of grid
1082points in y-direction.&nbsp; </p> <p>A value for this
1083parameter must be assigned. Since the lower
1084array bound in PALM starts with i = 0, the actual number of grid points
1085is equal to <b>ny+1</b>. In case of cyclic boundary
1086conditions along
1087y, the domain size is (<b>ny+1</b>) * <a href="#dy">dy</a>.</p>
1088<p>For parallel runs, in case of <a href="#grid_matching">grid_matching</a>
1089= <span style="font-style: italic;">'strict'</span>,
1090<b>ny+1</b> must
1091be an integral multiple
1092of the processor numbers (see <a href="#npex">npex</a>
1093and <a href="#npey">npey</a>)&nbsp;
1094along y- as well as along x-direction (due to data
1095transposition restrictions).</p> </td> </tr> <tr>
1096<td style="vertical-align: top;"> <p><a name="nz"></a><b>nz</b></p>
1097</td> <td style="vertical-align: top;">I</td>
1098<td style="vertical-align: top;"><br> </td> <td style="vertical-align: top;"> <p>Number of grid
1099points in z-direction.&nbsp; </p> <p>A value for this
1100parameter must be assigned. Since the lower
1101array bound in PALM
1102starts with k = 0 and since one additional grid point is added at the
1103top boundary (k = <b>nz+1</b>), the actual number of grid
1104points is <b>nz+2</b>.
1105However, the prognostic equations are only solved up to <b>nz</b>
1106(u,
1107v)
1108or up to <b>nz-1</b> (w, scalar quantities). The top
1109boundary for u
1110and v is at k = <b>nz+1</b> (u, v) while at k = <b>nz</b>
1111for all
1112other quantities.&nbsp; </p> <p>For parallel
1113runs,&nbsp; in case of <a href="#grid_matching">grid_matching</a>
1114= <span style="font-style: italic;">'strict'</span>,
1115<b>nz</b> must
1116be an integral multiple of
1117the number of processors in x-direction (due to data transposition
1118restrictions).</p> </td> </tr> <tr> <td style="vertical-align: top;"> <p><a name="omega"></a><b>omega</b></p>
1119</td> <td style="vertical-align: top;">R</td>
1120<td style="vertical-align: top;"><i>7.29212E-5</i></td>
1121<td style="vertical-align: top;"> <p>Angular
1122velocity of the rotating system (in rad s<sup>-1</sup>).&nbsp;
1123</p> <p>The angular velocity of the earth is set by
1124default. The
1125values
1126of the Coriolis parameters are calculated as:&nbsp; </p> <ul>
1127<p>f = 2.0 * <b>omega</b> * sin(<a href="#phi">phi</a>)&nbsp;
1128<br>f* = 2.0 * <b>omega</b> * cos(<a href="#phi">phi</a>)</p>
1129</ul> </td> </tr> <tr> <td style="vertical-align: top;"> <p><a name="outflow_damping_width"></a><b>outflow_damping_width</b></p>
1130</td> <td style="vertical-align: top;">I</td>
1131<td style="vertical-align: top;"><span style="font-style: italic;">MIN(20,
1132nx/2</span> or <span style="font-style: italic;">ny/2)</span></td>
1133<td style="vertical-align: top;">Width of
1134the damping range in the vicinity of the outflow (gridpoints).<br>
1135<br>
1136When using non-cyclic lateral boundaries (see <a href="chapter_4.1.html#bc_lr">bc_lr</a>
1137or <a href="chapter_4.1.html#bc_ns">bc_ns</a>),
1138a smoothing has to be applied to the
1139velocity field in the vicinity of the outflow in order to suppress any
1140reflections of outgoing disturbances. This parameter controlls the
1141horizontal range to which the smoothing is applied. The range is given
1142in gridpoints counted from the respective outflow boundary. For further
1143details about the smoothing see parameter <a href="chapter_4.1.html#km_damp_max">km_damp_max</a>,
1144which defines the magnitude of the damping.</td> </tr>
1145<tr> <td style="vertical-align: top;"> <p><a name="overshoot_limit_e"></a><b>overshoot_limit_e</b></p>
1146</td> <td style="vertical-align: top;">R</td>
1147<td style="vertical-align: top;"><i>0.0</i></td>
1148<td style="vertical-align: top;"> <p>Allowed limit
1149for the overshooting of subgrid-scale TKE in
1150case that the upstream-spline scheme is switched on (in m<sup>2</sup>/s<sup>2</sup>).&nbsp;
1151</p> <p>By deafult, if cut-off of overshoots is switched
1152on for the
1153upstream-spline scheme (see <a href="#cut_spline_overshoot">cut_spline_overshoot</a>),
1154no overshoots are permitted at all. If <b>overshoot_limit_e</b>
1155is given a non-zero value, overshoots with the respective
1156amplitude (both upward and downward) are allowed.&nbsp; </p>
1157<p>Only positive values are allowed for <b>overshoot_limit_e</b>.</p>
1158</td> </tr> <tr> <td style="vertical-align: top;"> <p><a name="overshoot_limit_pt"></a><b>overshoot_limit_pt</b></p>
1159</td> <td style="vertical-align: top;">R</td>
1160<td style="vertical-align: top;"><i>0.0</i></td>
1161<td style="vertical-align: top;"> <p>Allowed limit
1162for the overshooting of potential temperature in
1163case that the upstream-spline scheme is switched on (in K).&nbsp; </p>
1164<p>For further information see <a href="#overshoot_limit_e">overshoot_limit_e</a>.&nbsp;
1165</p> <p>Only positive values are allowed for <b>overshoot_limit_pt</b>.</p>
1166</td> </tr> <tr> <td style="vertical-align: top;"> <p><a name="overshoot_limit_u"></a><b>overshoot_limit_u</b></p>
1167</td> <td style="vertical-align: top;">R</td>
1168<td style="vertical-align: top;"><i>0.0</i></td>
1169<td style="vertical-align: top;">Allowed limit for the
1170overshooting of
1171the u-component of velocity in case that the upstream-spline scheme is
1172switched on (in m/s). <p>For further information see <a href="#overshoot_limit_e">overshoot_limit_e</a>.&nbsp;
1173</p> <p>Only positive values are allowed for <b>overshoot_limit_u</b>.</p>
1174</td> </tr> <tr> <td style="vertical-align: top;"> <p><a name="overshoot_limit_v"></a><b>overshoot_limit_v</b></p>
1175</td> <td style="vertical-align: top;">R</td>
1176<td style="vertical-align: top;"><i>0.0</i></td>
1177<td style="vertical-align: top;"> <p>Allowed limit
1178for the overshooting of the v-component of
1179velocity in case that the upstream-spline scheme is switched on
1180(in m/s).&nbsp; </p> <p>For further information see <a href="#overshoot_limit_e">overshoot_limit_e</a>.&nbsp;
1181</p> <p>Only positive values are allowed for <b>overshoot_limit_v</b>.</p>
1182</td> </tr> <tr> <td style="vertical-align: top;"> <p><a name="overshoot_limit_w"></a><b>overshoot_limit_w</b></p>
1183</td> <td style="vertical-align: top;">R</td>
1184<td style="vertical-align: top;"><i>0.0</i></td>
1185<td style="vertical-align: top;"> <p>Allowed limit
1186for the overshooting of the w-component of
1187velocity in case that the upstream-spline scheme is switched on
1188(in m/s).&nbsp; </p> <p>For further information see <a href="#overshoot_limit_e">overshoot_limit_e</a>.&nbsp;
1189</p> <p>Only positive values are permitted for <b>overshoot_limit_w</b>.</p>
1190</td> </tr> <tr> <td style="vertical-align: top;"> <p><a name="passive_scalar"></a><b>passive_scalar</b></p>
1191</td> <td style="vertical-align: top;">L</td>
1192<td style="vertical-align: top;"><i>.F.</i></td>
1193<td style="vertical-align: top;"> <p>Parameter to
1194switch on the prognostic equation for a passive
1195scalar. <br> </p> <p>The initial vertical profile
1196of s can be set via parameters <a href="#s_surface">s_surface</a>,
1197<a href="#s_vertical_gradient">s_vertical_gradient</a>
1198and&nbsp; <a href="#s_vertical_gradient_level">s_vertical_gradient_level</a>.
1199Boundary conditions can be set via <a href="#s_surface_initial_change">s_surface_initial_change</a>
1200and <a href="#surface_scalarflux">surface_scalarflux</a>.&nbsp;
1201</p> <p><b>Note:</b> <br>
1202With <span style="font-weight: bold;">passive_scalar</span>
1203switched
1204on, the simultaneous use of humidity (see <a href="#moisture">moisture</a>)
1205is impossible.</p> </td> </tr> <tr> <td style="vertical-align: top;"> <p><a name="phi"></a><b>phi</b></p>
1206</td> <td style="vertical-align: top;">R</td>
1207<td style="vertical-align: top;"><i>55.0</i></td>
1208<td style="vertical-align: top;"> <p>Geographical
1209latitude (in degrees).&nbsp; </p> <p>The value of
1210this parameter determines the value of the
1211Coriolis parameters f and f*, provided that the angular velocity (see <a href="#omega">omega</a>)
1212is non-zero.</p> </td> </tr> <tr> <td style="vertical-align: top;"> <p><a name="prandtl_layer"></a><b>prandtl_layer</b></p>
1213</td> <td style="vertical-align: top;">L</td>
1214<td style="vertical-align: top;"><i>.T.</i></td>
1215<td style="vertical-align: top;"> <p>Parameter to
1216switch on a Prandtl layer.&nbsp; </p> <p>By default,
1217a Prandtl layer is switched on at the bottom
1218boundary between z = 0 and z = 0.5 * <a href="#dz">dz</a>
1219(the first computational grid point above ground for u, v and the
1220scalar quantities).
1221In this case, at the bottom boundary, free-slip conditions for u and v
1222(see <a href="#bc_uv_b">bc_uv_b</a>)
1223are not allowed. Likewise, laminar
1224simulations with constant eddy diffusivities (<a href="#km_constant">km_constant</a>)
1225are forbidden.&nbsp; </p> <p>With Prandtl-layer
1226switched off, the TKE boundary condition <a href="#bc_e_b">bc_e_b</a>
1227= '<i>(u*)**2+neumann'</i> must not be used and is
1228automatically
1229changed to <i>'neumann'</i> if necessary.&nbsp; Also,
1230the pressure
1231boundary condition <a href="#bc_p_b">bc_p_b</a>
1232= <i>'neumann+inhomo'</i>&nbsp; is not allowed. </p>
1233<p>The roughness length is declared via the parameter <a href="#roughness_length">roughness_length</a>.</p>
1234</td> </tr> <tr> <td style="vertical-align: top;"> <p><a name="precipitation"></a><b>precipitation</b></p>
1235</td> <td style="vertical-align: top;">L</td>
1236<td style="vertical-align: top;"><span style="font-style: italic;">.F.</span></td> <td style="vertical-align: top;"> <p>Parameter to switch
1237on the precipitation scheme.<br> </p> <p>For
1238precipitation processes PALM uses a simplified Kessler
1239scheme. This scheme only considers the
1240so-called autoconversion, that means the generation of rain water by
1241coagulation of cloud drops among themselves. Precipitation begins and
1242is immediately removed from the flow as soon as the liquid water
1243content exceeds the critical value of 0.5 g/kg.</p> </td> </tr>
1244<tr><td style="vertical-align: top;"><a name="pt_reference"></a><span style="font-weight: bold;">pt_reference</span></td><td style="vertical-align: top;">R</td><td style="vertical-align: top;"><span style="font-style: italic;">use horizontal average as
1245refrence</span></td><td style="vertical-align: top;">Reference
1246temperature to be used in all buoyancy terms (in K).<br><br>By
1247default, the instantaneous horizontal average over the total model
1248domain is used.</td></tr><tr> <td style="vertical-align: top;"> <p><a name="pt_surface"></a><b>pt_surface</b></p>
1249</td> <td style="vertical-align: top;">R</td>
1250<td style="vertical-align: top;"><i>300.0</i></td>
1251<td style="vertical-align: top;"> <p>Surface
1252potential temperature (in K).&nbsp; </p> <p>This
1253parameter assigns the value of the potential temperature
1254pt at the surface (k=0)<b>.</b> Starting from this value,
1255the
1256initial vertical temperature profile is constructed with <a href="#pt_vertical_gradient">pt_vertical_gradient</a>
1257and <a href="#pt_vertical_gradient_level">pt_vertical_gradient_level
1258</a>.
1259This profile is also used for the 1d-model as a stationary profile.</p>
1260</td> </tr> <tr> <td style="vertical-align: top;"> <p><a name="pt_surface_initial_change"></a><b>pt_surface_initial</b>
1261<br> <b>_change</b></p> </td> <td style="vertical-align: top;">R</td> <td style="vertical-align: top;"><span style="font-style: italic;">0.0</span><br> </td>
1262<td style="vertical-align: top;"> <p>Change in
1263surface temperature to be made at the beginning of
1264the 3d run
1265(in K).&nbsp; </p> <p>If <b>pt_surface_initial_change</b>
1266is set to a non-zero
1267value, the near surface sensible heat flux is not allowed to be given
1268simultaneously (see <a href="#surface_heatflux">surface_heatflux</a>).</p>
1269</td> </tr> <tr> <td style="vertical-align: top;"> <p><a name="pt_vertical_gradient"></a><b>pt_vertical_gradient</b></p>
1270</td> <td style="vertical-align: top;">R (10)</td>
1271<td style="vertical-align: top;"><i>10 * 0.0</i></td>
1272<td style="vertical-align: top;"> <p>Temperature
1273gradient(s) of the initial temperature profile (in
1274K
1275/ 100 m).&nbsp; </p> <p>This temperature gradient
1276holds starting from the height&nbsp;
1277level defined by <a href="#pt_vertical_gradient_level">pt_vertical_gradient_level</a>
1278(precisely: for all uv levels k where zu(k) &gt;
1279pt_vertical_gradient_level,
1280pt_init(k) is set: pt_init(k) = pt_init(k-1) + dzu(k) * <b>pt_vertical_gradient</b>)
1281up to the top boundary or up to the next height level defined
1282by <a href="#pt_vertical_gradient_level">pt_vertical_gradient_level</a>.
1283A total of 10 different gradients for 11 height intervals (10 intervals
1284if <a href="#pt_vertical_gradient_level">pt_vertical_gradient_level</a>(1)
1285= <i>0.0</i>) can be assigned. The surface temperature is
1286assigned via <a href="#pt_surface">pt_surface</a>.&nbsp;
1287</p> <p>Example:&nbsp; </p> <ul> <p><b>pt_vertical_gradient</b>
1288= <i>1.0</i>, <i>0.5</i>,&nbsp; <br>
1289<b>pt_vertical_gradient_level</b> = <i>500.0</i>,
1290<i>1000.0</i>,</p> </ul> <p>That
1291defines the temperature profile to be neutrally
1292stratified
1293up to z = 500.0 m with a temperature given by <a href="#pt_surface">pt_surface</a>.
1294For 500.0 m &lt; z &lt;= 1000.0 m the temperature gradient is
12951.0 K /
1296100 m and for z &gt; 1000.0 m up to the top boundary it is
12970.5 K / 100 m (it is assumed that the assigned height levels correspond
1298with uv levels). </p> </td> </tr> <tr> <td style="vertical-align: top;"> <p><a name="pt_vertical_gradient_level"></a><b>pt_vertical_gradient</b>
1299<br> <b>_level</b></p> </td> <td style="vertical-align: top;">R (10)</td> <td style="vertical-align: top;"> <p><i>10 *</i>&nbsp;
1300<span style="font-style: italic;">0.0</span><br>
1301</p> </td> <td style="vertical-align: top;">
1302<p>Height level from which on the temperature gradient defined by
1303<a href="#pt_vertical_gradient">pt_vertical_gradient</a>
1304is effective (in m).&nbsp; </p> <p>The height levels
1305are to be assigned in ascending order. The
1306default values result in a neutral stratification regardless of the
1307values of <a href="#pt_vertical_gradient">pt_vertical_gradient</a>
1308(unless the top boundary of the model is higher than 100000.0 m).
1309For the piecewise construction of temperature profiles see <a href="#pt_vertical_gradient">pt_vertical_gradient</a>.</p>
1310</td> </tr> <tr> <td style="vertical-align: top;"> <p><a name="q_surface"></a><b>q_surface</b></p>
1311</td> <td style="vertical-align: top;">R</td>
1312<td style="vertical-align: top;"><i>0.0</i></td>
1313<td style="vertical-align: top;"> <p>Surface
1314specific humidity / total water content (kg/kg).&nbsp; </p> <p>This
1315parameter assigns the value of the specific humidity q at
1316the surface (k=0).&nbsp; Starting from this value, the initial
1317humidity
1318profile is constructed with&nbsp; <a href="#q_vertical_gradient">q_vertical_gradient</a>
1319and <a href="#q_vertical_gradient_level">q_vertical_gradient_level</a>.
1320This profile is also used for the 1d-model as a stationary profile.</p>
1321</td> </tr> <tr> <td style="vertical-align: top;"> <p><a name="q_surface_initial_change"></a><b>q_surface_initial</b>
1322<br> <b>_change</b></p> </td> <td style="vertical-align: top;">R<br> </td> <td style="vertical-align: top;"><i>0.0</i></td>
1323<td style="vertical-align: top;"> <p>Change in
1324surface specific humidity / total water content to
1325be made at the beginning
1326of the 3d run (kg/kg).&nbsp; </p> <p>If <b>q_surface_initial_change</b><i>
1327</i>is set to a
1328non-zero value the
1329near surface latent heat flux (water flux) is not allowed to be given
1330simultaneously (see <a href="#surface_waterflux">surface_waterflux</a>).</p>
1331</td> </tr> <tr> <td style="vertical-align: top;"> <p><a name="q_vertical_gradient"></a><b>q_vertical_gradient</b></p>
1332</td> <td style="vertical-align: top;">R (10)</td>
1333<td style="vertical-align: top;"><i>10 * 0.0</i></td>
1334<td style="vertical-align: top;"> <p>Humidity
1335gradient(s) of the initial humidity profile
1336(in 1/100 m).&nbsp; </p> <p>This humidity gradient
1337holds starting from the height
1338level&nbsp; defined by <a href="#q_vertical_gradient_level">q_vertical_gradient_level</a>
1339(precisely: for all uv levels k, where zu(k) &gt;
1340q_vertical_gradient_level,
1341q_init(k) is set: q_init(k) = q_init(k-1) + dzu(k) * <b>q_vertical_gradient</b>)
1342up to the top boundary or up to the next height level defined
1343by <a href="#q_vertical_gradient_level">q_vertical_gradient_level</a>.
1344A total of 10 different gradients for 11 height intervals (10 intervals
1345if <a href="#q_vertical_gradient_level">q_vertical_gradient_level</a>(1)
1346= <i>0.0</i>) can be asigned. The surface humidity is
1347assigned
1348via <a href="#q_surface">q_surface</a>. </p>
1349<p>Example:&nbsp; </p> <ul> <p><b>q_vertical_gradient</b>
1350= <i>0.001</i>, <i>0.0005</i>,&nbsp; <br>
1351<b>q_vertical_gradient_level</b> = <i>500.0</i>,
1352<i>1000.0</i>,</p> </ul>
1353That defines the humidity to be constant with height up to z =
1354500.0
1355m with a
1356value given by <a href="#q_surface">q_surface</a>.
1357For 500.0 m &lt; z &lt;= 1000.0 m the humidity gradient is
13580.001 / 100
1359m and for z &gt; 1000.0 m up to the top boundary it is
13600.0005 / 100 m (it is assumed that the assigned height levels
1361correspond with uv
1362levels). </td> </tr> <tr> <td style="vertical-align: top;"> <p><a name="q_vertical_gradient_level"></a><b>q_vertical_gradient</b>
1363<br> <b>_level</b></p> </td> <td style="vertical-align: top;">R (10)</td> <td style="vertical-align: top;"> <p><i>10 *</i>&nbsp;
1364<i>0.0</i></p> </td> <td style="vertical-align: top;"> <p>Height level from
1365which on the moisture gradient defined by <a href="#q_vertical_gradient">q_vertical_gradient</a>
1366is effective (in m).&nbsp; </p> <p>The height levels
1367are to be assigned in ascending order. The
1368default values result in a humidity constant with height regardless of
1369the values of <a href="#q_vertical_gradient">q_vertical_gradient</a>
1370(unless the top boundary of the model is higher than 100000.0 m). For
1371the piecewise construction of humidity profiles see <a href="#q_vertical_gradient">q_vertical_gradient</a>.</p>
1372</td> </tr> <tr> <td style="vertical-align: top;"> <p><a name="radiation"></a><b>radiation</b></p>
1373</td> <td style="vertical-align: top;">L</td>
1374<td style="vertical-align: top;"><i>.F.</i></td>
1375<td style="vertical-align: top;"> <p>Parameter to
1376switch on longwave radiation cooling at
1377cloud-tops.&nbsp; </p> <p>Long-wave radiation
1378processes are parameterized by the
1379effective emissivity, which considers only the absorption and emission
1380of long-wave radiation at cloud droplets. The radiation scheme can be
1381used only with <a href="#cloud_physics">cloud_physics</a>
1382= .TRUE. .</p> </td> </tr> <tr> <td style="vertical-align: top;"> <p><a name="random_generator"></a><b>random_generator</b></p>
1383</td> <td style="vertical-align: top;">C * 20</td>
1384<td style="vertical-align: top;"> <p><i>'numerical</i><br>
1385<i>recipes'</i></p> </td> <td style="vertical-align: top;"> <p>Random number
1386generator to be used for creating uniformly
1387distributed random numbers. <br> </p> <p>It is
1388used if random perturbations are to be imposed on the
1389velocity field or on the surface heat flux field (see <a href="chapter_4.2.html#create_disturbances">create_disturbances</a>
1390and <a href="chapter_4.2.html#random_heatflux">random_heatflux</a>).
1391By default, the "Numerical Recipes" random number generator is used.
1392This one provides exactly the same order of random numbers on all
1393different machines and should be used in particular for comparison runs.<br>
1394<br>
1395Besides, a system-specific generator is available ( <b>random_generator</b>
1396= <i>'system-specific')</i> which should particularly be
1397used for runs
1398on vector parallel computers (NEC), because the default generator
1399cannot be vectorized and therefore significantly drops down the code
1400performance on these machines.<br> </p> <span style="font-weight: bold;">Note:</span><br>
1401Results from two otherwise identical model runs will not be comparable
1402one-to-one if they used different random number generators.</td> </tr>
1403<tr> <td style="vertical-align: top;"> <p><a name="random_heatflux"></a><b>random_heatflux</b></p>
1404</td> <td style="vertical-align: top;">L</td>
1405<td style="vertical-align: top;"><i>.F.</i></td>
1406<td style="vertical-align: top;"> <p>Parameter to
1407impose random perturbations on the internal two-dimensional near
1408surface heat flux field <span style="font-style: italic;">shf</span>.
1409<br> </p>If a near surface heat flux is used as bottom
1410boundary
1411condition (see <a href="#surface_heatflux">surface_heatflux</a>),
1412it is by default assumed to be horizontally homogeneous. Random
1413perturbations can be imposed on the internal
1414two-dimensional&nbsp;heat flux field <span style="font-style: italic;">shf</span> by assigning <b>random_heatflux</b>
1415= <i>.T.</i>. The disturbed heat flux field is calculated
1416by
1417multiplying the
1418values at each mesh point with a normally distributed random number
1419with a mean value and standard deviation of 1. This is repeated after
1420every timestep.<br> <br>
1421In case of a non-flat <a href="#topography">topography</a>,&nbsp;assigning
1422<b>random_heatflux</b>
1423= <i>.T.</i> imposes random perturbations on the
1424combined&nbsp;heat
1425flux field <span style="font-style: italic;">shf</span>
1426composed of <a href="#surface_heatflux">surface_heatflux</a>
1427at the bottom surface and <a href="#wall_heatflux">wall_heatflux(0)</a>
1428at the topography top face.</td> </tr> <tr> <td style="vertical-align: top;"> <p><a name="rif_max"></a><b>rif_max</b></p>
1429</td> <td style="vertical-align: top;">R</td>
1430<td style="vertical-align: top;"><i>1.0</i></td>
1431<td style="vertical-align: top;"> <p>Upper limit of
1432the flux-Richardson number.&nbsp; </p> <p>With the
1433Prandtl layer switched on (see <a href="#prandtl_layer">prandtl_layer</a>),
1434flux-Richardson numbers (rif) are calculated for z=z<sub>p</sub>
1435(k=1)
1436in the 3d-model (in the 1d model for all heights). Their values in
1437particular determine the
1438values of the friction velocity (1d- and 3d-model) and the values of
1439the eddy diffusivity (1d-model). With small wind velocities at the
1440Prandtl layer top or small vertical wind shears in the 1d-model, rif
1441can take up unrealistic large values. They are limited by an upper (<span style="font-weight: bold;">rif_max</span>) and lower
1442limit (see <a href="#rif_min">rif_min</a>)
1443for the flux-Richardson number. The condition <b>rif_max</b>
1444&gt; <b>rif_min</b>
1445must be met.</p> </td> </tr> <tr> <td style="vertical-align: top;"> <p><a name="rif_min"></a><b>rif_min</b></p>
1446</td> <td style="vertical-align: top;">R</td>
1447<td style="vertical-align: top;"><i>- 5.0</i></td>
1448<td style="vertical-align: top;"> <p>Lower limit of
1449the flux-Richardson number.&nbsp; </p> <p>For further
1450explanations see <a href="#rif_max">rif_max</a>.
1451The condition <b>rif_max</b> &gt; <b>rif_min </b>must
1452be met.</p> </td> </tr> <tr> <td style="vertical-align: top;"> <p><a name="roughness_length"></a><b>roughness_length</b></p>
1453</td> <td style="vertical-align: top;">R</td>
1454<td style="vertical-align: top;"><i>0.1</i></td>
1455<td style="vertical-align: top;"> <p>Roughness
1456length (in m).&nbsp; </p> <p>This parameter is
1457effective only in case that a Prandtl layer
1458is switched
1459on (see <a href="#prandtl_layer">prandtl_layer</a>).</p>
1460</td> </tr> <tr> <td style="vertical-align: top;"> <p><a name="scalar_advec"></a><b>scalar_advec</b></p>
1461</td> <td style="vertical-align: top;">C * 10</td>
1462<td style="vertical-align: top;"><i>'pw-scheme'</i></td>
1463<td style="vertical-align: top;"> <p>Advection
1464scheme to be used for the scalar quantities.&nbsp; </p> <p>The
1465user can choose between the following schemes:<br> </p> <p><span style="font-style: italic;">'pw-scheme'</span><br>
1466</p> <div style="margin-left: 40px;">The scheme of
1467Piascek and
1468Williams (1970, J. Comp. Phys., 6,
1469392-405) with central differences in the form C3 is used.<br>
1470If intermediate Euler-timesteps are carried out in case of <a href="#timestep_scheme">timestep_scheme</a>
1471= <span style="font-style: italic;">'leapfrog+euler'</span>
1472the
1473advection scheme is - for the Euler-timestep - automatically switched
1474to an upstream-scheme. <br> </div> <br> <p><span style="font-style: italic;">'bc-scheme'</span><br>
1475</p> <div style="margin-left: 40px;">The Bott
1476scheme modified by
1477Chlond (1994, Mon.
1478Wea. Rev., 122, 111-125). This is a conservative monotonous scheme with
1479very small numerical diffusion and therefore very good conservation of
1480scalar flow features. The scheme however, is computationally very
1481expensive both because it is expensive itself and because it does (so
1482far) not allow specific code optimizations (e.g. cache optimization).
1483Choice of this
1484scheme forces the Euler timestep scheme to be used for the scalar
1485quantities. For output of horizontally averaged
1486profiles of the resolved / total heat flux, <a href="chapter_4.2.html#data_output_pr">data_output_pr</a>
1487= <i>'w*pt*BC'</i> / <i>'wptBC' </i>should
1488be used, instead of the
1489standard profiles (<span style="font-style: italic;">'w*pt*'</span>
1490and <span style="font-style: italic;">'wpt'</span>)
1491because these are
1492too inaccurate with this scheme. However, for subdomain analysis (see <a href="#statistic_regions">statistic_regions</a>)
1493exactly the reverse holds: here <i>'w*pt*BC'</i> and <i>'wptBC'</i>
1494show very large errors and should not be used.<br> <br>
1495This scheme is not allowed for non-cyclic lateral boundary conditions
1496(see <a href="#bc_lr">bc_lr</a>
1497and <a href="#bc_ns">bc_ns</a>).<br> <br>
1498</div> <span style="font-style: italic;">'ups-scheme'</span><br>
1499<p style="margin-left: 40px;">The upstream-spline-scheme
1500is used
1501(see Mahrer and Pielke,
15021978: Mon. Wea. Rev., 106, 818-830). In opposite to the Piascek
1503Williams scheme, this is characterized by much better numerical
1504features (less numerical diffusion, better preservation of flux
1505structures, e.g. vortices), but computationally it is much more
1506expensive. In
1507addition, the use of the Euler-timestep scheme is mandatory (<a href="#timestep_scheme">timestep_scheme</a>
1508= <span style="font-style: italic;">'</span><i>euler'</i>),
1509i.e. the
1510timestep accuracy is only first order. For this reason the advection of
1511momentum (see <a href="#momentum_advec">momentum_advec</a>)
1512should then also be carried out with the upstream-spline scheme,
1513because otherwise the momentum would
1514be subject to large numerical diffusion due to the upstream
1515scheme.&nbsp; </p> <p style="margin-left: 40px;">Since
1516the cubic splines used tend
1517to overshoot under
1518certain circumstances, this effect must be adjusted by suitable
1519filtering and smoothing (see <a href="#cut_spline_overshoot">cut_spline_overshoot</a>,
1520<a href="#long_filter_factor">long_filter_factor</a>,
1521<a href="#ups_limit_pt">ups_limit_pt</a>, <a href="#ups_limit_u">ups_limit_u</a>, <a href="#ups_limit_v">ups_limit_v</a>, <a href="#ups_limit_w">ups_limit_w</a>).
1522This is always neccesssary for runs with stable stratification,
1523even if this stratification appears only in parts of the model
1524domain.&nbsp; </p> <p style="margin-left: 40px;">With
1525stable stratification the
1526upstream-upline scheme also produces gravity waves with large
1527amplitude, which must be
1528suitably damped (see <a href="chapter_4.2.html#rayleigh_damping_factor">rayleigh_damping_factor</a>).<br>
1529</p> <p style="margin-left: 40px;"><span style="font-weight: bold;">Important: </span>The&nbsp;
1530upstream-spline scheme is not implemented for humidity and passive
1531scalars (see <a href="#moisture">moisture</a>
1532and <a href="#passive_scalar">passive_scalar</a>)
1533and requires the use of a 2d-domain-decomposition. The last conditions
1534severely restricts code optimization on several machines leading to
1535very long execution times! This scheme is also not allowed for
1536non-cyclic lateral boundary conditions (see <a href="#bc_lr">bc_lr</a>
1537and <a href="#bc_ns">bc_ns</a>).</p><br>A
1538differing advection scheme can be choosed for the subgrid-scale TKE
1539using parameter <a href="chapter_4.1.html#use_upstream_for_tke">use_upstream_for_tke</a>.</td>
1540</tr> <tr> <td style="vertical-align: top;">
1541<p><a name="statistic_regions"></a><b>statistic_regions</b></p>
1542</td> <td style="vertical-align: top;">I</td>
1543<td style="vertical-align: top;"><i>0</i></td>
1544<td style="vertical-align: top;"> <p>Number of
1545additional user-defined subdomains for which
1546statistical analysis
1547and corresponding output (profiles, time series) shall be
1548made.&nbsp; </p> <p>By default, vertical profiles and
1549other statistical quantities
1550are calculated as horizontal and/or volume average of the total model
1551domain. Beyond that, these calculations can also be carried out for
1552subdomains which can be defined using the field <a href="chapter_3.5.3.html">rmask </a>within the
1553user-defined software
1554(see <a href="chapter_3.5.3.html">chapter
15553.5.3</a>). The number of these subdomains is determined with the
1556parameter <b>statistic_regions</b>. Maximum 9 additional
1557subdomains
1558are allowed. The parameter <a href="chapter_4.3.html#region">region</a>
1559can be used to assigned names (identifier) to these subdomains which
1560are then used in the headers
1561of the output files and plots.</p><p>If the default NetCDF
1562output format is selected (see parameter <a href="chapter_4.2.html#data_output_format">data_output_format</a>),
1563data for the total domain and all defined subdomains are output to the
1564same file(s) (<a href="chapter_3.4.html#DATA_1D_PR_NETCDF">DATA_1D_PR_NETCDF</a>,
1565<a href="chapter_3.4.html#DATA_1D_TS_NETCDF">DATA_1D_TS_NETCDF</a>).
1566In case of <span style="font-weight: bold;">statistic_regions</span>
1567&gt; <span style="font-style: italic;">0</span>,
1568data on the file for the different domains can be distinguished by a
1569suffix which is appended to the quantity names. Suffix 0 means data for
1570the total domain, suffix 1 means data for subdomain 1, etc.</p><p>In
1571case of <span style="font-weight: bold;">data_output_format</span>
1572= <span style="font-style: italic;">'profil'</span>,
1573individual local files for profiles (<a href="chapter_3.4.html#PLOT1D_DATA">PLOT1D_DATA</a>)&nbsp;are
1574created for each subdomain. The individual subdomain files differ by
1575their name (the
1576number of the respective subdomain is attached, e.g.
1577PLOT1D_DATA_1). In this case the name of the file with the data of
1578the total domain is PLOT1D_DATA_0. If no subdomains
1579are declared (<b>statistic_regions</b> = <i>0</i>),
1580the name
1581PLOT1D_DATA is used (this must be considered in the
1582respective file connection statements of the <span style="font-weight: bold;">mrun</span> configuration
1583file).</p> </td> </tr> <tr> <td style="vertical-align: top;"> <p><a name="surface_heatflux"></a><b>surface_heatflux</b></p>
1584</td> <td style="vertical-align: top;">R</td>
1585<td style="vertical-align: top;"><span style="font-style: italic;">no prescribed<br>
1586heatflux<br> </span></td> <td style="vertical-align: top;"> <p>Kinematic sensible
1587heat flux at the bottom surface (in K m/s).&nbsp; </p> <p>If
1588a value is assigned to this parameter, the internal two-dimensional
1589surface heat flux field <span style="font-style: italic;">shf</span>
1590is initialized with the value of <span style="font-weight: bold;">surface_heatflux</span>&nbsp;as
1591bottom (horizontally homogeneous) boundary condition for the
1592temperature equation. This additionally requires that a Neumann
1593condition must be used for the potential temperature (see <a href="#bc_pt_b">bc_pt_b</a>),
1594because otherwise the resolved scale may contribute to
1595the surface flux so that a constant value cannot be guaranteed. Also,
1596changes of the
1597surface temperature (see <a href="#pt_surface_initial_change">pt_surface_initial_change</a>)
1598are not allowed. The parameter <a href="#random_heatflux">random_heatflux</a>
1599can be used to impose random perturbations on the (homogeneous) surface
1600heat
1601flux field <span style="font-style: italic;">shf</span>.&nbsp;</p>
1602<p>
1603In case of a non-flat <a href="#topography">topography</a>,&nbsp;the
1604internal two-dimensional&nbsp;surface heat
1605flux field <span style="font-style: italic;">shf</span>
1606is initialized with the value of <span style="font-weight: bold;">surface_heatflux</span>
1607at the bottom surface and <a href="#wall_heatflux">wall_heatflux(0)</a>
1608at the topography top face.&nbsp;The parameter<a href="#random_heatflux"> random_heatflux</a>
1609can be used to impose random perturbations on this combined surface
1610heat
1611flux field <span style="font-style: italic;">shf</span>.&nbsp;
1612</p> <p>If no surface heat flux is assigned, <span style="font-style: italic;">shf</span> is calculated
1613at each timestep by u<sub>*</sub> * theta<sub>*</sub>
1614(of course only with <a href="#prandtl_layer">prandtl_layer</a>
1615switched on). Here, u<sub>*</sub>
1616and theta<sub>*</sub> are calculated from the Prandtl law
1617assuming
1618logarithmic wind and temperature
1619profiles between k=0 and k=1. In this case a Dirichlet condition (see <a href="#bc_pt_b">bc_pt_b</a>)
1620must be used as bottom boundary condition for the potential temperature.</p><p>See
1621also <a href="#top_heatflux">top_heatflux</a>.</p>
1622</td> </tr> <tr> <td style="vertical-align: top;"> <p><a name="surface_pressure"></a><b>surface_pressure</b></p>
1623</td> <td style="vertical-align: top;">R</td>
1624<td style="vertical-align: top;"><i>1013.25</i></td>
1625<td style="vertical-align: top;"> <p>Atmospheric
1626pressure at the surface (in hPa).&nbsp; </p>
1627Starting from this surface value, the vertical pressure
1628profile is calculated once at the beginning of the run assuming a
1629neutrally stratified
1630atmosphere. This is needed for
1631converting between the liquid water potential temperature and the
1632potential temperature (see <a href="#cloud_physics">cloud_physics</a><span style="text-decoration: underline;"></span>).</td>
1633</tr> <tr> <td style="vertical-align: top;">
1634<p><a name="surface_scalarflux"></a><b>surface_scalarflux</b></p>
1635</td> <td style="vertical-align: top;">R</td>
1636<td style="vertical-align: top;"><i>0.0</i></td>
1637<td style="vertical-align: top;"> <p>Scalar flux at
1638the surface (in kg/(m<sup>2</sup> s)).&nbsp; </p>
1639<p>If a non-zero value is assigned to this parameter, the
1640respective scalar flux value is used
1641as bottom (horizontally homogeneous) boundary condition for the scalar
1642concentration equation.&nbsp;This additionally requires that a
1643Neumann
1644condition must be used for the scalar concentration&nbsp;(see <a href="#bc_s_b">bc_s_b</a>),
1645because otherwise the resolved scale may contribute to
1646the surface flux so that a constant value cannot be guaranteed. Also,
1647changes of the
1648surface scalar concentration (see <a href="#s_surface_initial_change">s_surface_initial_change</a>)
1649are not allowed. <br> </p> <p>If no surface scalar
1650flux is assigned (<b>surface_scalarflux</b>
1651= <i>0.0</i>),
1652it is calculated at each timestep by u<sub>*</sub> * s<sub>*</sub>
1653(of course only with Prandtl layer switched on). Here, s<sub>*</sub>
1654is calculated from the Prandtl law assuming a logarithmic scalar
1655concentration
1656profile between k=0 and k=1. In this case a Dirichlet condition (see <a href="#bc_s_b">bc_s_b</a>)
1657must be used as bottom boundary condition for the scalar concentration.</p>
1658</td> </tr> <tr> <td style="vertical-align: top;"> <p><a name="surface_waterflux"></a><b>surface_waterflux</b></p>
1659</td> <td style="vertical-align: top;">R</td>
1660<td style="vertical-align: top;"><i>0.0</i></td>
1661<td style="vertical-align: top;"> <p>Kinematic
1662water flux near the surface (in m/s).&nbsp; </p> <p>If
1663a non-zero value is assigned to this parameter, the
1664respective water flux value is used
1665as bottom (horizontally homogeneous) boundary condition for the
1666humidity equation. This additionally requires that a Neumann
1667condition must be used for the specific humidity / total water content
1668(see <a href="#bc_q_b">bc_q_b</a>),
1669because otherwise the resolved scale may contribute to
1670the surface flux so that a constant value cannot be guaranteed. Also,
1671changes of the
1672surface humidity (see <a href="#q_surface_initial_change">q_surface_initial_change</a>)
1673are not allowed.<br> </p> <p>If no surface water
1674flux is assigned (<b>surface_waterflux</b>
1675= <i>0.0</i>),
1676it is calculated at each timestep by u<sub>*</sub> * q<sub>*</sub>
1677(of course only with Prandtl layer switched on). Here, q<sub>*</sub>
1678is calculated from the Prandtl law assuming a logarithmic temperature
1679profile between k=0 and k=1. In this case a Dirichlet condition (see <a href="#bc_q_b">bc_q_b</a>)
1680must be used as the bottom boundary condition for the humidity.</p>
1681</td> </tr> <tr> <td style="vertical-align: top;"> <p><a name="s_surface"></a><b>s_surface</b></p>
1682</td> <td style="vertical-align: top;">R</td>
1683<td style="vertical-align: top;"><i>0.0</i></td>
1684<td style="vertical-align: top;"> <p>Surface value
1685of the passive scalar (in kg/m<sup>3</sup>).&nbsp;<br>
1686</p>
1687This parameter assigns the value of the passive scalar s at
1688the surface (k=0)<b>.</b> Starting from this value, the
1689initial vertical scalar concentration profile is constructed with<a href="#s_vertical_gradient">
1690s_vertical_gradient</a> and <a href="#s_vertical_gradient_level">s_vertical_gradient_level</a>.</td>
1691</tr> <tr> <td style="vertical-align: top;">
1692<p><a name="s_surface_initial_change"></a><b>s_surface_initial</b>
1693<br> <b>_change</b></p> </td> <td style="vertical-align: top;">R</td> <td style="vertical-align: top;"><i>0.0</i></td>
1694<td style="vertical-align: top;"> <p>Change in
1695surface scalar concentration to be made at the
1696beginning of the 3d run (in kg/m<sup>3</sup>).&nbsp; </p>
1697<p>If <b>s_surface_initial_change</b><i>&nbsp;</i>is
1698set to a
1699non-zero
1700value, the near surface scalar flux is not allowed to be given
1701simultaneously (see <a href="#surface_scalarflux">surface_scalarflux</a>).</p>
1702</td> </tr> <tr> <td style="vertical-align: top;"> <p><a name="s_vertical_gradient"></a><b>s_vertical_gradient</b></p>
1703</td> <td style="vertical-align: top;">R (10)</td>
1704<td style="vertical-align: top;"><i>10 * 0</i><i>.0</i></td>
1705<td style="vertical-align: top;"> <p>Scalar
1706concentration gradient(s) of the initial scalar
1707concentration profile (in kg/m<sup>3 </sup>/
1708100 m).&nbsp; </p> <p>The scalar gradient holds
1709starting from the height level
1710defined by <a href="#s_vertical_gradient_level">s_vertical_gradient_level
1711</a>(precisely: for all uv levels k, where zu(k) &gt;
1712s_vertical_gradient_level, s_init(k) is set: s_init(k) = s_init(k-1) +
1713dzu(k) * <b>s_vertical_gradient</b>) up to the top
1714boundary or up to
1715the next height level defined by <a href="#s_vertical_gradient_level">s_vertical_gradient_level</a>.
1716A total of 10 different gradients for 11 height intervals (10 intervals
1717if <a href="#s_vertical_gradient_level">s_vertical_gradient_level</a>(1)
1718= <i>0.0</i>) can be assigned. The surface scalar value is
1719assigned
1720via <a href="#s_surface">s_surface</a>.<br> </p>
1721<p>Example:&nbsp; </p> <ul> <p><b>s_vertical_gradient</b>
1722= <i>0.1</i>, <i>0.05</i>,&nbsp; <br>
1723<b>s_vertical_gradient_level</b> = <i>500.0</i>,
1724<i>1000.0</i>,</p> </ul> <p>That
1725defines the scalar concentration to be constant with
1726height up to z = 500.0 m with a value given by <a href="#s_surface">s_surface</a>.
1727For 500.0 m &lt; z &lt;= 1000.0 m the scalar gradient is 0.1
1728kg/m<sup>3 </sup>/ 100 m and for z &gt; 1000.0 m up to
1729the top
1730boundary it is 0.05 kg/m<sup>3 </sup>/ 100 m (it is
1731assumed that the
1732assigned height levels
1733correspond with uv
1734levels).</p> </td> </tr> <tr> <td style="vertical-align: top;"> <p><a name="s_vertical_gradient_level"></a><b>s_vertical_gradient_</b>
1735<br> <b>level</b></p> </td> <td style="vertical-align: top;">R (10)</td> <td style="vertical-align: top;"> <p><i>10 *</i>
1736<i>0.0</i></p> </td> <td style="vertical-align: top;"> <p>Height level from
1737which on the scalar gradient defined by <a href="#s_vertical_gradient">s_vertical_gradient</a>
1738is effective (in m).&nbsp; </p> <p>The height levels
1739are to be assigned in ascending order. The
1740default values result in a scalar concentration constant with height
1741regardless of the values of <a href="#s_vertical_gradient">s_vertical_gradient</a>
1742(unless the top boundary of the model is higher than 100000.0 m). For
1743the
1744piecewise construction of scalar concentration profiles see <a href="#s_vertical_gradient">s_vertical_gradient</a>.</p>
1745</td> </tr> <tr> <td style="vertical-align: top;"> <p><a name="timestep_scheme"></a><b>timestep_scheme</b></p>
1746</td> <td style="vertical-align: top;">C * 20</td>
1747<td style="vertical-align: top;"> <p><i>'runge</i><br>
1748<i>kutta-3'</i></p> </td> <td style="vertical-align: top;"> <p>Time step scheme to
1749be used for the integration of the prognostic
1750variables.&nbsp; </p> <p>The user can choose between
1751the following schemes:<br> </p> <p><span style="font-style: italic;">'runge-kutta-3'</span><br>
1752</p> <div style="margin-left: 40px;">Third order
1753Runge-Kutta scheme.<br>
1754This scheme requires the use of <a href="#momentum_advec">momentum_advec</a>
1755= <a href="#scalar_advec">scalar_advec</a>
1756= '<i>pw-scheme'</i>. Please refer to the&nbsp;<a href="../tec/numerik.heiko/zeitschrittverfahren.pdf">documentation
1757on PALM's time integration schemes&nbsp;(28p., in German)</a>
1758fur further details.<br> </div> <p><span style="font-style: italic;">'runge-kutta-2'</span><br>
1759</p> <div style="margin-left: 40px;">Second order
1760Runge-Kutta scheme.<br>
1761For special features see <b>timestep_scheme</b> = '<i>runge-kutta-3'</i>.<br>
1762</div> <br> <span style="font-style: italic;"><span style="font-style: italic;">'leapfrog'</span><br>
1763<br> </span> <div style="margin-left: 40px;">Second
1764order leapfrog scheme.<br>
1765Although this scheme requires a constant timestep (because it is
1766centered in time),&nbsp; is even applied in case of changes in
1767timestep. Therefore, only small
1768changes of the timestep are allowed (see <a href="#dt">dt</a>).
1769However, an Euler timestep is always used as the first timestep of an
1770initiali run. When using the Bott-Chlond scheme for scalar advection
1771(see <a href="#scalar_advec">scalar_advec</a>),
1772the prognostic equation for potential temperature will be calculated
1773with the Euler scheme, although the leapfrog scheme is switched
1774on.&nbsp; <br>
1775The leapfrog scheme must not be used together with the upstream-spline
1776scheme for calculating the advection (see <a href="#scalar_advec">scalar_advec</a>
1777= '<i>ups-scheme'</i> and <a href="#momentum_advec">momentum_advec</a>
1778= '<i>ups-scheme'</i>).<br> </div> <br>
1779<span style="font-style: italic;">'</span><span style="font-style: italic;"><span style="font-style: italic;">leapfrog+euler'</span><br>
1780<br> </span> <div style="margin-left: 40px;">The
1781leapfrog scheme is used, but
1782after each change of a timestep an Euler timestep is carried out.
1783Although this method is theoretically correct (because the pure
1784leapfrog method does not allow timestep changes), the divergence of the
1785velocity field (after applying the pressure solver) may be
1786significantly larger than with <span style="font-style: italic;">'leapfrog'</span>.<br>
1787</div> <br> <span style="font-style: italic;">'euler'</span><br>
1788<br> <div style="margin-left: 40px;">First order
1789Euler scheme.&nbsp; <br>
1790The Euler scheme must be used when treating the advection terms with
1791the upstream-spline scheme (see <a href="#scalar_advec">scalar_advec</a>
1792= <span style="font-style: italic;">'ups-scheme'</span>
1793and <a href="#momentum_advec">momentum_advec</a>
1794= <span style="font-style: italic;">'ups-scheme'</span>).</div>
1795<br><br>A differing timestep scheme can be choosed for the
1796subgrid-scale TKE using parameter <a href="#use_upstream_for_tke">use_upstream_for_tke</a>.<br>
1797</td> </tr> <tr> <td style="text-align: left; vertical-align: top;"><span style="font-weight: bold;"><a name="topography"></a></span><span style="font-weight: bold;">topography</span></td>
1798<td style="vertical-align: top;">C * 40</td> <td style="vertical-align: top;"><span style="font-style: italic;">'flat'</span></td> <td>
1799<p>Topography mode.&nbsp; </p> <p>The user can
1800choose between the following modes:<br> </p> <p><span style="font-style: italic;">'flat'</span><br> </p>
1801<div style="margin-left: 40px;">Flat surface.</div> <p><span style="font-style: italic;">'single_building'</span><br>
1802</p> <div style="margin-left: 40px;">Flow
1803around&nbsp;a single rectangular building mounted on a flat surface.<br>
1804The building size and location can be specified with the parameters <a href="#building_height">building_height</a>, <a href="#building_length_x">building_length_x</a>, <a href="#building_length_y">building_length_y</a>, <a href="#building_wall_left">building_wall_left</a> and <a href="#building_wall_south">building_wall_south</a>.</div>
1805<span style="font-style: italic;"></span> <p><span style="font-style: italic;">'read_from_file'</span><br>
1806</p> <div style="margin-left: 40px;">Flow around
1807arbitrary topography.<br>
1808This mode requires the input file <a href="chapter_3.4.html#TOPOGRAPHY_DATA">TOPOGRAPHY_DATA</a><font color="#000000">. This file contains </font><font color="#000000"><font color="#000000">the&nbsp;</font></font><font color="#000000">arbitrary topography </font><font color="#000000"><font color="#000000">height
1809information</font></font><font color="#000000">
1810in m. These data&nbsp;<span style="font-style: italic;"></span>must
1811exactly match the horizontal grid.</font> </div> <span style="font-style: italic;"><br> </span><font color="#000000">
1812Alternatively, the user may add code to the user interface subroutine <a href="chapter_3.5.1.html#user_init_grid">user_init_grid</a>
1813to allow further topography modes.<br> <br>
1814All non-flat <span style="font-weight: bold;">topography</span>
1815modes </font>require the use of <a href="#momentum_advec">momentum_advec</a>
1816= <a href="#scalar_advec">scalar_advec</a>
1817= '<i>pw-scheme'</i>, <a href="chapter_4.2.html#psolver">psolver</a>
1818= <i>'poisfft'</i> or '<i>poisfft_hybrid'</i>,
1819<i>&nbsp;</i><a href="#alpha_surface">alpha_surface</a>
1820= 0.0, <a href="#bc_lr">bc_lr</a> = <a href="#bc_ns">bc_ns</a> = <span style="font-style: italic;">'cyclic'</span>,&nbsp;<a style="" href="#galilei_transformation">galilei_transformation</a>
1821= <span style="font-style: italic;">.F.</span>,&nbsp;<a href="#cloud_physics">cloud_physics&nbsp;</a> = <span style="font-style: italic;">.F.</span>,&nbsp; <a href="#cloud_droplets">cloud_droplets</a> = <span style="font-style: italic;">.F.</span>,&nbsp; <a href="#moisture">moisture</a> = <span style="font-style: italic;">.F.</span>, and <a href="#prandtl_layer">prandtl_layer</a> = .T..<br>
1822<font color="#000000"><br>
1823Note that an inclined model domain requires the use of <span style="font-weight: bold;">topography</span> = <span style="font-style: italic;">'flat'</span> and a
1824nonzero </font><a href="#alpha_surface">alpha_surface</a>.</td>
1825</tr> <tr><td style="vertical-align: top;"><a name="top_heatflux"></a><span style="font-weight: bold;">top_heatflux</span></td><td style="vertical-align: top;">R</td><td style="vertical-align: top;"><span style="font-style: italic;">no prescribed<br>
1826heatflux</span></td><td style="vertical-align: top;"><p>Kinematic
1827sensible heat flux at the top boundary (in K m/s).&nbsp; </p>
1828<p>If a value is assigned to this parameter, the internal
1829two-dimensional surface heat flux field <span style="font-family: monospace;">tswst</span> is
1830initialized with the value of <span style="font-weight: bold;">top_heatflux</span>&nbsp;as
1831top (horizontally homogeneous) boundary condition for the
1832temperature equation. This additionally requires that a Neumann
1833condition must be used for the potential temperature (see <a href="chapter_4.1.html#bc_pt_t">bc_pt_t</a>),
1834because otherwise the resolved scale may contribute to
1835the top flux so that a constant value cannot be guaranteed.<span style="font-style: italic;"></span>&nbsp;</p>
1836<p><span style="font-weight: bold;">Note:</span><br>The
1837application of a top heat flux additionally requires the setting of
1838initial parameter <a href="#use_top_fluxes">use_top_fluxes</a>
1839= .T..<span style="font-style: italic;"></span><span style="font-weight: bold;"></span> </p><p>No
1840Prandtl-layer is available at the top boundary so far.</p><p>See
1841also <a href="#surface_heatflux">surface_heatflux</a>.</p>
1842</td></tr><tr> <td style="vertical-align: top;">
1843<p><a name="ug_surface"></a><span style="font-weight: bold;">ug_surface</span></p>
1844</td> <td style="vertical-align: top;">R<br> </td>
1845<td style="vertical-align: top;"><span style="font-style: italic;">0.0</span><br> </td>
1846<td style="vertical-align: top;">u-component of the
1847geostrophic
1848wind at the surface (in m/s).<br> <br>
1849This parameter assigns the value of the u-component of the geostrophic
1850wind (ug) at the surface (k=0). Starting from this value, the initial
1851vertical profile of the <br>
1852u-component of the geostrophic wind is constructed with <a href="#ug_vertical_gradient">ug_vertical_gradient</a>
1853and <a href="#ug_vertical_gradient_level">ug_vertical_gradient_level</a>.
1854The
1855profile constructed in that way is used for creating the initial
1856vertical velocity profile of the 3d-model. Either it is applied, as it
1857has been specified by the user (<a href="#initializing_actions">initializing_actions</a>
1858= 'set_constant_profiles') or it is used for calculating a stationary
1859boundary layer wind profile (<a href="#initializing_actions">initializing_actions</a>
1860= 'set_1d-model_profiles'). If ug is constant with height (i.e. ug(k)=<span style="font-weight: bold;">ug_surface</span>)
1861and&nbsp; has a large
1862value, it is recommended to use a Galilei-transformation of the
1863coordinate system, if possible (see <a href="#galilei_transformation">galilei_transformation</a>),
1864in order to obtain larger time steps.<br> </td> </tr>
1865<tr> <td style="vertical-align: top;"> <p><a name="ug_vertical_gradient"></a><span style="font-weight: bold;">ug_vertical_gradient</span></p>
1866</td> <td style="vertical-align: top;">R(10)<br>
1867</td> <td style="vertical-align: top;"><span style="font-style: italic;">10
1868* 0.0</span><br> </td> <td style="vertical-align: top;">Gradient(s) of the initial
1869profile of the&nbsp; u-component of the geostrophic wind (in
18701/100s).<br> <br>
1871The gradient holds starting from the height level defined by <a href="#ug_vertical_gradient_level">ug_vertical_gradient_level</a>
1872(precisely: for all uv levels k where zu(k) &gt; <a href="#ug_vertical_gradient_level">ug_vertical_gradient_level</a>,
1873ug(k) is set: ug(k) = ug(k-1) + dzu(k) * <span style="font-weight: bold;">ug_vertical_gradient</span>)
1874up to the top
1875boundary or up to the next height level defined by <a href="#ug_vertical_gradient_level">ug_vertical_gradient_level</a>.
1876A
1877total of 10 different gradients for 11 height intervals (10
1878intervals&nbsp; if <a href="#ug_vertical_gradient_level">ug_vertical_gradient_level</a>(1)
1879= 0.0) can be assigned. The surface geostrophic wind is assigned by <a href="#ug_surface">ug_surface</a>. <br> </td>
1880</tr> <tr> <td style="vertical-align: top;">
1881<p><a name="ug_vertical_gradient_level"></a><span style="font-weight: bold;">ug_vertical_gradient_level</span></p>
1882</td> <td style="vertical-align: top;">R(10)<br>
1883</td> <td style="vertical-align: top;"><span style="font-style: italic;">10
1884* 0.0</span><br> </td> <td style="vertical-align: top;">Height level from which on the
1885gradient defined by <a href="#ug_vertical_gradient">ug_vertical_gradient</a>
1886is effective (in m).<br> <br>
1887The height levels are to be assigned in ascending order. For the
1888piecewise construction of a profile of the u-component of the
1889geostrophic wind component (ug) see <a href="#ug_vertical_gradient">ug_vertical_gradient</a>.<br>
1890</td> </tr> <tr> <td style="vertical-align: top;"> <p><a name="ups_limit_e"></a><b>ups_limit_e</b></p>
1891</td> <td style="vertical-align: top;">R</td>
1892<td style="vertical-align: top;"><i>0.0</i></td>
1893<td style="vertical-align: top;"> <p>Subgrid-scale
1894turbulent kinetic energy difference used as
1895criterion for applying the upstream scheme when upstream-spline
1896advection is switched on (in m<sup>2</sup>/s<sup>2</sup>).
1897&nbsp; </p> <p>This variable steers the appropriate
1898treatment of the
1899advection of the subgrid-scale turbulent kinetic energy in case that
1900the uptream-spline scheme is used . For further information see <a href="#ups_limit_pt">ups_limit_pt</a>.&nbsp; </p>
1901<p>Only positive values are allowed for <b>ups_limit_e</b>.
1902</p> </td> </tr> <tr> <td style="vertical-align: top;"> <p><a name="ups_limit_pt"></a><b>ups_limit_pt</b></p>
1903</td> <td style="vertical-align: top;">R</td>
1904<td style="vertical-align: top;"><i>0.0</i></td>
1905<td style="vertical-align: top;"> <p>Temperature
1906difference used as criterion for applying&nbsp;
1907the upstream scheme when upstream-spline advection&nbsp; is
1908switched on
1909(in K).&nbsp; </p> <p>This criterion is used if the
1910upstream-spline scheme is
1911switched on (see <a href="#scalar_advec">scalar_advec</a>).<br>
1912If, for a given gridpoint, the absolute temperature difference with
1913respect to the upstream
1914grid point is smaller than the value given for <b>ups_limit_pt</b>,
1915the upstream scheme is used for this gridpoint (by default, the
1916upstream-spline scheme is always used). Reason: in case of a very small
1917upstream gradient, the advection should cause only a very small
1918tendency. However, in such situations the upstream-spline scheme may
1919give wrong tendencies at a
1920grid point due to spline overshooting, if simultaneously the downstream
1921gradient is very large. In such cases it may be more reasonable to use
1922the upstream scheme. The numerical diffusion caused by the upstream
1923schme remains small as long as the upstream gradients are small.<br>
1924</p> <p>The percentage of grid points for which the
1925upstream
1926scheme is actually used, can be output as a time series with respect to
1927the
1928three directions in space with run parameter (see <a href="chapter_4.2.html#dt_dots">dt_dots</a>, the
1929timeseries names in the NetCDF file are <i>'splptx'</i>, <i>'splpty'</i>,
1930<i>'splptz'</i>). The percentage
1931of gridpoints&nbsp; should stay below a certain limit, however, it
1932is
1933not possible to give
1934a general limit, since it depends on the respective flow.&nbsp; </p>
1935<p>Only positive values are permitted for <b>ups_limit_pt</b>.<br>
1936</p>
1937A more effective control of
1938the &ldquo;overshoots&rdquo; can be achieved with parameter <a href="#cut_spline_overshoot">cut_spline_overshoot</a>.
1939</td> </tr> <tr> <td style="vertical-align: top;"> <p><a name="ups_limit_u"></a><b>ups_limit_u</b></p>
1940</td> <td style="vertical-align: top;">R</td>
1941<td style="vertical-align: top;"><i>0.0</i></td>
1942<td style="vertical-align: top;"> <p>Velocity
1943difference (u-component) used as criterion for
1944applying the upstream scheme
1945when upstream-spline advection is switched on (in m/s).&nbsp; </p>
1946<p>This variable steers the appropriate treatment of the
1947advection of the u-velocity-component in case that the upstream-spline
1948scheme is used. For further
1949information see <a href="#ups_limit_pt">ups_limit_pt</a>.&nbsp;
1950</p> <p>Only positive values are permitted for <b>ups_limit_u</b>.</p>
1951</td> </tr> <tr> <td style="vertical-align: top;"> <p><a name="ups_limit_v"></a><b>ups_limit_v</b></p>
1952</td> <td style="vertical-align: top;">R</td>
1953<td style="vertical-align: top;"><i>0.0</i></td>
1954<td style="vertical-align: top;"> <p>Velocity
1955difference (v-component) used as criterion for
1956applying the upstream scheme
1957when upstream-spline advection is switched on (in m/s).&nbsp; </p>
1958<p>This variable steers the appropriate treatment of the
1959advection of the v-velocity-component in case that the upstream-spline
1960scheme is used. For further
1961information see <a href="#ups_limit_pt">ups_limit_pt</a>.&nbsp;
1962</p> <p>Only positive values are permitted for <b>ups_limit_v</b>.</p>
1963</td> </tr> <tr> <td style="vertical-align: top;"> <p><a name="ups_limit_w"></a><b>ups_limit_w</b></p>
1964</td> <td style="vertical-align: top;">R</td>
1965<td style="vertical-align: top;"><i>0.0</i></td>
1966<td style="vertical-align: top;"> <p>Velocity
1967difference (w-component) used as criterion for
1968applying the upstream scheme
1969when upstream-spline advection is switched on (in m/s).&nbsp; </p>
1970<p>This variable steers the appropriate treatment of the
1971advection of the w-velocity-component in case that the upstream-spline
1972scheme is used. For further
1973information see <a href="#ups_limit_pt">ups_limit_pt</a>.&nbsp;
1974</p> <p>Only positive values are permitted for <b>ups_limit_w</b>.</p>
1975</td> </tr> <tr> <td style="vertical-align: top;"> <p><a name="use_surface_fluxes"></a><b>use_surface_fluxes</b></p>
1976</td> <td style="vertical-align: top;">L</td>
1977<td style="vertical-align: top;"><i>.F.</i></td>
1978<td style="vertical-align: top;"> <p>Parameter to
1979steer the treatment of the subgrid-scale vertical
1980fluxes within the diffusion terms at k=1 (bottom boundary).<br> </p>
1981<p>By default, the near-surface subgrid-scale fluxes are
1982parameterized (like in the remaining model domain) using the gradient
1983approach. If <b>use_surface_fluxes</b>
1984= <i>.TRUE.</i>, the user-assigned surface fluxes are used
1985instead
1986(see <a href="#surface_heatflux">surface_heatflux</a>,
1987<a href="#surface_waterflux">surface_waterflux</a>
1988and <a href="#surface_scalarflux">surface_scalarflux</a>)
1989<span style="font-weight: bold;">or</span> the
1990surface fluxes are
1991calculated via the Prandtl layer relation (depends on the bottom
1992boundary conditions, see <a href="#bc_pt_b">bc_pt_b</a>,
1993<a href="#bc_q_b">bc_q_b</a>
1994and <a href="#bc_s_b">bc_s_b</a>).<br> </p>
1995<p><b>use_surface_fluxes</b>
1996is automatically set <i>.TRUE.</i>, if a Prandtl layer is
1997used (see <a href="#prandtl_layer">prandtl_layer</a>).&nbsp;
1998</p> <p>The user may prescribe the surface fluxes at the
1999bottom
2000boundary without using a Prandtl layer by setting <span style="font-weight: bold;">use_surface_fluxes</span> =
2001<span style="font-style: italic;">.T.</span> and <span style="font-weight: bold;">prandtl_layer</span> = <span style="font-style: italic;">.F.</span>. If , in this
2002case, the
2003momentum flux (u<sub>*</sub><sup>2</sup>)
2004should also be prescribed,
2005the user must assign an appropriate value within the user-defined code.</p>
2006</td> </tr> <tr><td style="vertical-align: top;"><a name="use_top_fluxes"></a><span style="font-weight: bold;">use_top_fluxes</span></td><td style="vertical-align: top;">L</td><td style="vertical-align: top;"><span style="font-style: italic;">.F.</span></td><td style="vertical-align: top;"> <p>Parameter to steer
2007the treatment of the subgrid-scale vertical
2008fluxes within the diffusion terms at k=nz (top boundary).</p><p>By
2009default, the fluxes at nz are calculated using the gradient approach.
2010If <b>use_top_fluxes</b>
2011= <i>.TRUE.</i>, the user-assigned top fluxes are used
2012instead
2013(see <a href="chapter_4.1.html#top_heatflux">top_heatflux</a>).</p><p>Currently,
2014only a value for the sensible heatflux can be assigned. In case of <span style="font-weight: bold;">use_top_fluxes</span> = <span style="font-style: italic;">.TRUE.</span>, the latent
2015heat flux at the top will be automatically set to zero.</p></td></tr><tr>
2016<td style="vertical-align: top;"> <p><a name="use_ug_for_galilei_tr"></a><b>use_ug_for_galilei_tr</b></p>
2017</td> <td style="vertical-align: top;">L</td>
2018<td style="vertical-align: top;"><i>.T.</i></td>
2019<td style="vertical-align: top;"> <p>Switch to
2020determine the translation velocity in case that a
2021Galilean transformation is used.<br> </p> <p>In
2022case of a Galilean transformation (see <a href="#galilei_transformation">galilei_transformation</a>),
2023<b>use_ug_for_galilei_tr</b>
2024= <i>.T.</i>&nbsp; ensures
2025that the coordinate system is translated with the geostrophic windspeed.<br>
2026</p> <p>Alternatively, with <b>use_ug_for_galilei_tr</b>
2027= <i>.F</i>.,
2028the
2029geostrophic wind can be replaced as translation speed by the (volume)
2030averaged velocity. However, in this case the user must be aware of fast
2031growing gravity waves, so this
2032choice is usually not recommended!</p> </td> </tr> <tr><td align="left" valign="top"><a name="use_upstream_for_tke"></a><span style="font-weight: bold;">use_upstream_for_tke</span></td><td align="left" valign="top">L</td><td align="left" valign="top"><span style="font-style: italic;">.F.</span></td><td align="left" valign="top">Parameter to choose the
2033advection/timestep scheme to be used for the subgrid-scale TKE.<br><br>By
2034default, the advection scheme and the timestep scheme to be used for
2035the subgrid-scale TKE are set by the initialization parameters <a href="#scalar_advec">scalar_advec</a> and <a href="#timestep_scheme">timestep_scheme</a>,
2036respectively. <span style="font-weight: bold;">use_upstream_for_tke</span>
2037= <span style="font-style: italic;">.T.</span>
2038forces the Euler-scheme and the upstream-scheme to be used as timestep
2039scheme and advection scheme, respectively. By these methods, the strong
2040(artificial) near-surface vertical gradients of the subgrid-scale TKE
2041are significantly reduced. This is required when subgrid-scale
2042velocities are used for advection of particles (see particle package
2043parameter <a href="chapter_4.2.html#use_sgs_for_particles">use_sgs_for_particles</a>).</td></tr><tr>
2044<td style="vertical-align: top;"> <p><a name="vg_surface"></a><span style="font-weight: bold;">vg_surface</span></p>
2045</td> <td style="vertical-align: top;">R<br> </td>
2046<td style="vertical-align: top;"><span style="font-style: italic;">0.0</span><br> </td>
2047<td style="vertical-align: top;">v-component of the
2048geostrophic
2049wind at the surface (in m/s).<br> <br>
2050This parameter assigns the value of the v-component of the geostrophic
2051wind (vg) at the surface (k=0). Starting from this value, the initial
2052vertical profile of the <br>
2053v-component of the geostrophic wind is constructed with <a href="#vg_vertical_gradient">vg_vertical_gradient</a>
2054and <a href="#vg_vertical_gradient_level">vg_vertical_gradient_level</a>.
2055The
2056profile
2057constructed in that way is used for creating the initial vertical
2058velocity profile of the 3d-model. Either it is applied, as it has been
2059specified by the user (<a href="#initializing_actions">initializing_actions</a>
2060= 'set_constant_profiles')
2061or it is used for calculating a stationary boundary layer wind profile
2062(<a href="#initializing_actions">initializing_actions</a>
2063=
2064'set_1d-model_profiles'). If vg is constant
2065with height (i.e. vg(k)=<span style="font-weight: bold;">vg_surface</span>)
2066and&nbsp; has a large value, it is
2067recommended to use a Galilei-transformation of the coordinate system,
2068if possible (see <a href="#galilei_transformation">galilei_transformation</a>),
2069in order to obtain larger
2070time steps.</td> </tr> <tr> <td style="vertical-align: top;"> <p><a name="vg_vertical_gradient"></a><span style="font-weight: bold;">vg_vertical_gradient</span></p>
2071</td> <td style="vertical-align: top;">R(10)<br>
2072</td> <td style="vertical-align: top;"><span style="font-style: italic;">10
2073* 0.0</span><br> </td> <td style="vertical-align: top;">Gradient(s) of the initial
2074profile of the&nbsp; v-component of the geostrophic wind (in
20751/100s).<br> <br>
2076The gradient holds starting from the height level defined by <a href="#vg_vertical_gradient_level">vg_vertical_gradient_level</a>
2077(precisely: for all uv levels k where zu(k)
2078&gt; <a href="#vg_vertical_gradient_level">vg_vertical_gradient_level</a>,
2079vg(k) is set: vg(k) = vg(k-1) + dzu(k)
2080* <span style="font-weight: bold;">vg_vertical_gradient</span>)
2081up to
2082the top boundary or up to the next height
2083level defined by <a href="#vg_vertical_gradient_level">vg_vertical_gradient_level</a>.
2084A total of 10 different
2085gradients for 11 height intervals (10 intervals&nbsp; if <a href="#vg_vertical_gradient_level">vg_vertical_gradient_level</a>(1)
2086=
20870.0) can be assigned. The surface
2088geostrophic wind is assigned by <a href="#vg_surface">vg_surface</a>.</td>
2089</tr> <tr> <td style="vertical-align: top;">
2090<p><a name="vg_vertical_gradient_level"></a><span style="font-weight: bold;">vg_vertical_gradient_level</span></p>
2091</td> <td style="vertical-align: top;">R(10)<br>
2092</td> <td style="vertical-align: top;"><span style="font-style: italic;">10
2093* 0.0</span><br> </td> <td style="vertical-align: top;">Height level from which on the
2094gradient defined by <a href="#vg_vertical_gradient">vg_vertical_gradient</a>
2095is effective (in m).<br> <br>
2096The height levels are to be assigned in ascending order. For the
2097piecewise construction of a profile of the v-component of the
2098geostrophic wind component (vg) see <a href="#vg_vertical_gradient">vg_vertical_gradient</a>.</td>
2099</tr> <tr> <td style="vertical-align: top;">
2100<p><a name="wall_adjustment"></a><b>wall_adjustment</b></p>
2101</td> <td style="vertical-align: top;">L</td>
2102<td style="vertical-align: top;"><i>.T.</i></td>
2103<td style="vertical-align: top;"> <p>Parameter to
2104restrict the mixing length in the vicinity of the
2105bottom
2106boundary.&nbsp; </p> <p>With <b>wall_adjustment</b>
2107= <i>.TRUE., </i>the mixing
2108length is limited to a maximum of&nbsp; 1.8 * z. This condition
2109typically affects only the
2110first grid points above the bottom boundary.</p> </td> </tr>
2111<tr> <td style="vertical-align: top;"><span style="font-weight: bold;"><a name="wall_heatflux"></a>wall_heatflux</span></td>
2112<td style="vertical-align: top;">R(5)</td> <td style="vertical-align: top;"><span style="font-style: italic;">5 * 0.0</span></td> <td>Prescribed
2113kinematic sensible heat flux in W m<sup>-2</sup>
2114at the five topography faces:<br> <br> <div style="margin-left: 40px;"><span style="font-weight: bold;">wall_heatflux(0)&nbsp;&nbsp;
2115&nbsp;</span>top face<br> <span style="font-weight: bold;">wall_heatflux(1)&nbsp;&nbsp;&nbsp;
2116</span>left face<br> <span style="font-weight: bold;">wall_heatflux(2)&nbsp;&nbsp;&nbsp;
2117</span>right face<br> <span style="font-weight: bold;">wall_heatflux(3)&nbsp;&nbsp;&nbsp;
2118</span>south face<br> <span style="font-weight: bold;">wall_heatflux(4)&nbsp;&nbsp;&nbsp;
2119</span>north face</div> <br>
2120This parameter applies only in case of a non-flat <a href="#topography">topography</a>.&nbsp;The
2121parameter <a href="#random_heatflux">random_heatflux</a>
2122can be used to impose random perturbations on the internal
2123two-dimensional surface heat
2124flux field <span style="font-style: italic;">shf</span>
2125that is composed of <a href="#surface_heatflux">surface_heatflux</a>
2126at the bottom surface and <span style="font-weight: bold;">wall_heatflux(0)</span>
2127at the topography top face.&nbsp;</td> </tr> </tbody>
2128</table><br>
2129<p style="line-height: 100%;"><br><font color="#000080"><font color="#000080"><a href="chapter_4.0.html"><font color="#000080"><img name="Grafik1" src="left.gif" align="bottom" border="2" height="32" width="32"></font></a><a href="index.html"><font color="#000080"><img name="Grafik2" src="up.gif" align="bottom" border="2" height="32" width="32"></font></a><a href="chapter_4.2.html"><font color="#000080"><img name="Grafik3" src="right.gif" align="bottom" border="2" height="32" width="32"></font></a></font></font></p>
2130<p style="line-height: 100%;"><i>Last
2131change:&nbsp;</i> $Id: chapter_4.1.html 61 2007-03-12 05:42:06Z raasch $ </p>
2132<br><br>
2133</body></html>
Note: See TracBrowser for help on using the repository browser.