Changes between Version 4 and Version 5 of doc/tec/topography

Timestamp:

Jun 6, 2017 1:05:06 PM (8 years ago)

Author:

suehring

Comment:

--

doc/tec/topography

@@ -1 +1 @@
 = Topography implementation =
 
-The topography implementation described in Sect. [wiki:/doc/tec/bc#Topography boundary conditions] allows the use of 2-D topography height data in PALM. Currently, the topography data has to be provided within a rastered ASCII file. After reading and mapping of these data to the
-horizontal grid in PALM, they can be directly incorporated into the standard loop structure of the Fortran code as lower vertical index
+Under construction
+
+With revision -r2232, the topography implementation is completely revised. 
+Starting from this revision, overhanging structures (e.g. bridges, ceilings or tunnels) are allowed, i.e. topography does not necessarily be surface-mounted. 
+
+The topography implementation described in Sect. [wiki:/doc/tec/bc#Topography boundary conditions] allows the use of 2-D topography height data (if no overhanging structures should be considered), or 3-D topography information. The topography input data has to be provided within a rastered ASCII or NetCDF file. Rastered 3-D topography consists of values either 1 (within topography) or 0 (outside topography), and has to be provided up to the level of the highest present topography structures. 
+After reading and mapping of topography data to the 3-D grid in PALM, flag arrays are set in order to mask grid boxes within the integration loops accordingly. 
+
+The prognostic terms are executed overall, i.e. also within topography. 
+Masking of topography is done via Fortran integer bit flags, where a certain bit position indicates whether the respective grid point belongs to an obstacle (bit is 0) or to the atmosphere (bit is 1). 
+Different bit positions indicate different masking on the staggered grid, e.g. on the u, v, w grid, equivalent to the former nzb_u/v/w_inner arrays.  Further, regions where special wall-bounded code is executed, is also masked via special bit flags. 
+A list of the respective bit flags and their meaning concerning the deprecated implementation via the nzb_u/v/w_inner arrays can be found in Table 1.  
+
+||='''Bit position'''  =||='''Meaning'''  =||
+|----------------
+{{{#!td  style="vertical-align:top"
+0
+}}}
+{{{#!td style="vertical-align:top"
+mask topography on scalar-grid ( former nzb_s_inner )
+}}}
+|----------------
+{{{#!td  style="vertical-align:top"
+1
+}}}
+{{{#!td style="vertical-align:top"
+mask topography on u-grid ( former nzb_u_inner )
+}}}
+|----------------
+{{{#!td  style="vertical-align:top"
+2
+}}}
+{{{#!td style="vertical-align:top"
+mask topography on v-grid ( former nzb_v_inner )
+}}}
+|----------------
+{{{#!td  style="vertical-align:top"
+3
+}}}
+{{{#!td style="vertical-align:top"
+mask topography on w-grid ( former nzb_w_inner )
+}}}
+|----------------
+{{{#!td  style="vertical-align:top"
+8
+}}}
+{{{#!td style="vertical-align:top"
+mask regions where surface-bounded code for u, v, w, and scalars is executed ( combines information of former nzb_s_outer, nzb_u_outer, nzb_v_outer and nzb_w_outer arrays )
+}}}
+|----------------
+{{{#!td  style="vertical-align:top"
+9
+}}}
+{{{#!td style="vertical-align:top"
+mask regions where model-top fluxes are applied ( former nzt_diff )
+}}}
+|----------------
+{{{#!td  style="vertical-align:top"
+12
+}}}
+{{{#!td style="vertical-align:top"
+flags upward-facing surfaces on scalar-grid
+}}}
+|----------------
+{{{#!td  style="vertical-align:top"
+13
+}}}
+{{{#!td style="vertical-align:top"
+flags downward-facing surfaces on scalar-grid
+}}}
+|----------------
+{{{#!td  style="vertical-align:top"
+14
+}}}
+{{{#!td style="vertical-align:top"
+flags upward-facing surfaces on u-grid
+}}}
+|----------------
+{{{#!td  style="vertical-align:top"
+15
+}}}
+{{{#!td style="vertical-align:top"
+flags downward-facing surfaces on u-grid
+}}}
+|----------------
+{{{#!td  style="vertical-align:top"
+17
+}}}
+{{{#!td style="vertical-align:top"
+flags upward-facing surfaces on v-grid
+}}}
+|----------------
+{{{#!td  style="vertical-align:top"
+17
+}}}
+{{{#!td style="vertical-align:top"
+flags downward-facing surfaces on v-grid
+}}}
+|----------------
+{{{#!td  style="vertical-align:top"
+18
+}}}
+{{{#!td style="vertical-align:top"
+flags upward-facing surfaces on w-grid
+}}}
+|----------------
+{{{#!td  style="vertical-align:top"
+19
+}}}
+{{{#!td style="vertical-align:top"
+flags downward-facing surfaces on w-grid
+}}}
+|----------------
+{{{#!td  style="vertical-align:top"
+20
+}}}
+{{{#!td style="vertical-align:top"
+mask topography on u-grid and one grid point above, used only for adding random perturbations ( former nzb_u_inner+1 )
+}}}
+|----------------
+{{{#!td  style="vertical-align:top"
+21
+}}}
+{{{#!td style="vertical-align:top"
+mask topography on v-grid and one grid point above, used only for adding random perturbations ( former nzb_v_inner+1 )
+}}}
+|----------------
+{{{#!td  style="vertical-align:top"
+24
+}}}
+{{{#!td style="vertical-align:top"
+mask regions where surface-bounded code is executed on scalar-grid ( former nzb_s_outer )
+}}}
+|----------------
+{{{#!td  style="vertical-align:top"
+25
+}}}
+{{{#!td style="vertical-align:top"
+mask regions where surface-bounded code is executed on scalar-grid ( former nzb_s_outer+1 )
+}}}
+|----------------
+{{{#!td  style="vertical-align:top"
+26
+}}}
+{{{#!td style="vertical-align:top"
+mask regions where surface-bounded code is executed on u-grid ( former nzb_u_outer )
+}}}
+|----------------
+{{{#!td  style="vertical-align:top"
+27
+}}}
+{{{#!td style="vertical-align:top"
+mask regions where surface-bounded code is executed on v-grid ( former nzb_v_outer )
+}}}
+|----------------
+{{{#!td  style="vertical-align:top"
+28
+}}}
+{{{#!td style="vertical-align:top"
+mask regions where surface-bounded code is executed on w-grid ( former nzb_w_outer )
+}}}
+|----------------
+{{{#!td  style="vertical-align:top"
+29
+}}}
+{{{#!td style="vertical-align:top"
+mask regions where surface-bounded code is executed on s-grid ( former nzb_diff_s_outer - 1 )
+}}}
+|----------------
+{{{#!td  style="vertical-align:top"
+30
+}}}
+{{{#!td style="vertical-align:top"
+mask regions where surface-bounded code is executed on s-grid ( former nzb_diff_s_outer )
+}}}
+|----------------
+
+
+The following  example illustrates the general realization of topography masking in the code using Bit flags:
+{{{
+#!Latex
+\begin{verbatim}
+   DO  k = nzb+1, nzt
+      tend(k,j,i) = tend(k,j,i) + …  &
+                       * MERGE( 1.0_wp, 0.0_wp, BTEST(wall_flags_0(k,j,i),0) )
+   ENDDO
+\end{verbatim}
+}}}
+Here, the intrinsic Fortran function `MERGE` returns 1 if `wall_flags_0(k,j,i)` at bit position zero is 1 (`BTEST` returns `.TRUE.`), else, `MERGE` would return 0 and no term is effectively added to the tend array.  
+A similar approach is also realized for the 5th-order advection-scheme  to degrade the order near surfaces.  
+
+
+
+ they can be directly incorporated into the standard loop structure of the Fortran code as lower vertical index
 for all integration loops. Therefore, PALM employs two 2-D height index arrays (e.g., '''''nzb_w_inner(j, i)''''' and '''''nzb_w_outer(j, i)''''' for the velocity component ''w'') to separate the domain into four regions based on the vertical index ''k'' (see Fig. 4 in Sect. [wiki:/doc/tec/bc#Topography boundary conditions]):
 {{{