source: palm/trunk/SCRIPTS/palm_csd @ 4815

#!/usr/bin/env python3
# -*- coding: utf-8 -*-
#--------------------------------------------------------------------------------#
# This file is part of the PALM model system.
#
# PALM is free software: you can redistribute it and/or modify it under the terms
# of the GNU General Public License as published by the Free Software Foundation,
# either version 3 of the License, or (at your option) any later version.
#
# PALM is distributed in the hope that it will be useful, but WITHOUT ANY
# WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS FOR
# A PARTICULAR PURPOSE.  See the GNU General Public License for more details.
#
# You should have received a copy of the GNU General Public License along with
# PALM. If not, see <http://www.gnu.org/licenses/>.
#
# Copyright 1997-2020  Leibniz Universitaet Hannover
#--------------------------------------------------------------------------------#
#
# Current revisions:
# -----------------
#
#
# Former revisions:
# -----------------
# $Id: palm_csd 4794 2020-11-24 10:40:20Z gronemeier $
# Removed debug output; assign negative ids for tree patches
#
# 4793 2020-11-24 09:48:21Z maronga
# Added output of tree_type (species or classification). Added tree_id for
# vegetation patches. Allow for overlapping vegetation with buildings
#
# 4749 2020-10-16 11:21:27Z maronga
# Bugfix: processing of patch height faulty for multiple domains
#
# 4746 2020-10-15 17:16:30Z maronga
# Add support for backyard high vegetation, which is commonly classified as 
# short grass
# 
# 4490 2020-04-10 05:32:45Z maronga
# Bugfix: call of canopy generator was not made if tree heights were only
# available via patch_height, but not in tree_height
# 
# 4481 2020-03-31 18:55:54Z maronga
# Bugfix: green roofs were assigned to all available roofs when switched on
# 
# 4149 2019-08-08 14:02:18Z suehring
# Bugfix, nbuilding_pars was 1 element to small 
# 
# 3955 2019-05-07 09:55:25Z maronga
# Bugfix in preparaing of green roofs
# Bugfix: remove LAI for bare soils
# 
# 3859 2019-04-03 20:30:31Z maronga
# Bugfix: wrong variable naming for 'y'
# 
# 3773 2019-03-01 08:56:57Z maronga
# Unspecificed changes
# 
# 3726 2019-02-07 18:22:49Z maronga
# Removed some more bugs
# 
# 3688 2019-01-22 10:44:20Z maronga
# Some unspecified bugfixes
# 
# 3668 2019-01-14 12:49:24Z maronga
# Various improvements and bugfixes
# 
# 3629 2018-12-13 12:18:54Z maronga
# Added canopy generator calls. Some improvements
# 
# 3567 2018-11-27 13:59:21Z maronga
# Initial revisions
#
# Description:
# ------------
# Processing tool for creating PIDS conform static drivers from rastered NetCDF 
# input
#
# @Author Bjoern Maronga (maronga@muk.uni-hannover.de)
#
# @todo Make input files optional
# @todo Allow for ASCII input of terrain height and building height
# @todo Modularize reading config file
# @todo Convert to object-oriented treatment (buidings, trees)
# @todo Automatically shift child domains so that their origin lies intersects
#       a edge note of the parent grid
#------------------------------------------------------------------------------#

from palm_csd_files.palm_csd_netcdf_interface import *
from palm_csd_files.palm_csd_tools import *
from palm_csd_files.palm_csd_canopy_generator import *
import numpy as np

    
def read_config_file():
   
   import configparser
   from math import floor
   import numpy as np
   import os
   import subprocess as sub
   import sys

#  Check if configuration files exists and quit otherwise
   input_config = ".csd.config"
   for i in range(1,len(sys.argv)): 
      input_config = str(sys.argv[i])
        
   config = configparser.RawConfigParser(allow_no_value=True)

   if ( os.path.isfile(input_config) == False ):
      print ("Error. No configuration file " + input_config + " found.")
      raise SystemExit   

   config.read(input_config)


#  Definition of settings

   global settings_bridge_width
   global settings_lai_roof_intensive
   global settings_lai_roof_extensive
   global settings_season
   global settings_lai_low_default
   global settings_lai_high_default
   global settings_patch_height_default
   global settings_lai_alpha
   global settings_lai_beta
   global settings_veg_type_below_trees
   global ndomains

#  Definition of global configuration parameters
   global global_acronym
   global global_angle
   global global_author
   global global_campaign
   global global_comment
   global global_contact
   global global_data_content  
   global global_dependencies
   global global_institution
   global global_keywords
   global global_location
   global global_palm_version
   global global_references
   global global_site
   global global_source

   global  path_out
   global filename_out
   global version_out


#  Definition of domain parameters
   global domain_names
   global domain_px
   global domain_x0
   global domain_y0
   global domain_x1
   global domain_y1
   global domain_nx
   global domain_ny
   global domain_dz
   global domain_3d
   global domain_high_vegetation
   global domain_ip
   global domain_za
   global domain_parent
   global domain_green_roofs
   global domain_street_trees
   global domain_canopy_patches
   global domain_overhanging_trees

#  Definition of input data parameters
   global input_names
   global input_px
  
        
   global input_file_x
   global input_file_y 
   global input_file_x_UTM
   global input_file_y_UTM   
   global input_file_lat
   global input_file_lon  
   global input_file_zt
   global input_file_buildings_2d
   global input_file_bridges_2d 
   global input_file_building_id
   global input_file_bridges_id   
   global input_file_building_type
   global input_file_building_type
   global input_file_lai  
   global input_file_vegetation_type
   global input_file_vegetation_height
   global input_file_pavement_type
   global input_file_water_type
   global input_file_street_type
   global input_file_street_crossings
   global input_file_soil_type
   global input_file_vegetation_on_roofs
   global input_file_tree_crown_diameter
   global input_file_tree_height
   global input_file_tree_trunk_diameter
   global input_file_tree_type
   global input_file_patch_height

   global zt_all
   global zt_min

   settings_bridge_width = 3.0
   settings_lai_roof_intensive = 2.5
   settings_lai_roof_extensive = 0.8
   settings_season = "summer"
   settings_lai_high_default = 6.0
   settings_lai_low_default = 1.0
   settings_patch_height_default = 10.0
   settings_lai_alpha = 5.0
   settings_lai_beta = 3.0
   settings_veg_type_below_trees = 3
   ndomains = 0

   global_acronym = ""
   global_angle = ""
   global_author = ""
   global_campaign = ""
   global_comment = ""
   global_contact = ""
   global_data_content = ""
   global_dependencies = ""
   global_institution = ""
   global_keywords = ""
   global_location = ""
   global_palm_version = 6.0
   global_references = ""
   global_site = ""
   global_source = ""

   path_out = ""
   version_out = 1
   filename_out = "default"
   
   domain_names = []
   domain_px = []
   domain_x0 = []
   domain_y0 = []
   domain_x1 = []
   domain_y1 = []
   domain_nx = []
   domain_ny = []
   domain_dz = []
   domain_3d = []
   domain_high_vegetation = []
   domain_ip = []
   domain_za = []
   domain_parent = []
   domain_green_roofs = []
   domain_street_trees = []
   domain_canopy_patches = []
   domain_overhanging_trees = []

   zt_min = 0.0
   zt_all = []

   input_names = []
   input_px = []

   input_file_x = []
   input_file_y = []  
   input_file_x_UTM = []
   input_file_y_UTM = []  
   input_file_lat = []
   input_file_lon = []  
   
   input_file_zt = []
   input_file_buildings_2d = []
   input_file_bridges_2d = []
   input_file_building_id = []
   input_file_bridges_id = []   
   input_file_building_type = []
   input_file_lai = []
   input_file_vegetation_type = []
   input_file_vegetation_height = []   
   input_file_pavement_type = []
   input_file_water_type = []
   input_file_street_type = []  
   input_file_street_crossings = []     
   input_file_soil_type = []
   input_file_vegetation_on_roofs = []
   input_file_tree_crown_diameter = []
   input_file_tree_height = []
   input_file_tree_trunk_diameter = []
   input_file_tree_type = []
   input_file_patch_height = []
   
# Load all user parameters from config file
   for i in range(0,len(config.sections())):

      read_tmp = config.sections()[i]
      
      if ( read_tmp == 'global' ):
         global_acronym = config.get(read_tmp, 'acronym')
         global_angle = float(config.get(read_tmp, 'rotation_angle'))
         global_author = config.get(read_tmp, 'author')
         global_campaign = config.get(read_tmp, 'campaign')
         global_comment = config.get(read_tmp, 'comment')
         global_contact = config.get(read_tmp, 'contact_person')
         global_data_content = config.get(read_tmp, 'data_content')   
         global_dependencies = config.get(read_tmp, 'dependencies')
         global_institution = config.get(read_tmp, 'institution')
         global_keywords = config.get(read_tmp, 'keywords')
         global_location = config.get(read_tmp, 'location')
         global_palm_version = float(config.get(read_tmp, 'palm_version'))
         global_references = config.get(read_tmp, 'references')
         global_site = config.get(read_tmp, 'site')
         global_source = config.get(read_tmp, 'source')


      if ( read_tmp == 'settings' ):
         settings_lai_roof_intensive = config.get(read_tmp, 'lai_roof_intensive')
         settings_lai_roof_extensive = config.get(read_tmp, 'lai_roof_extensive') 
         settings_bridge_width = float(config.get(read_tmp, 'bridge_width'))   
         settings_season = config.get(read_tmp, 'season')   
         settings_lai_high_default = float(config.get(read_tmp, 'lai_high_vegetation_default'))
         settings_lai_low_default = float(config.get(read_tmp, 'lai_low_vegetation_default'))         
         settings_patch_height_default = float(config.get(read_tmp, 'patch_height_default'))  
         settings_lai_alpha = float(config.get(read_tmp, 'lai_alpha')) 
         settings_lai_beta = float(config.get(read_tmp, 'lai_beta')) 
         settings_veg_type_below_trees = config.get(read_tmp, 'vegetation_type_below_trees')  

      if ( read_tmp == 'output' ):
         path_out = config.get(read_tmp, 'path')         
         filename_out = config.get(read_tmp, 'file_out')
         version_out = float(config.get(read_tmp, 'version'))

      if ( read_tmp.split("_")[0] == 'domain' ):
         ndomains = ndomains + 1
         domain_names.append(read_tmp.split("_")[1])
         domain_px.append(float(config.get(read_tmp, 'pixel_size')))
         domain_nx.append(int(config.get(read_tmp, 'nx')))
         domain_ny.append(int(config.get(read_tmp, 'ny')))
         domain_dz.append(float(config.get(read_tmp, 'dz')))       
         domain_3d.append(config.getboolean(read_tmp, 'buildings_3d'))
         domain_high_vegetation.append(config.getboolean(read_tmp, 'allow_high_vegetation'))
         domain_canopy_patches.append(config.getboolean(read_tmp, 'generate_vegetation_patches'))       
         domain_ip.append(config.getboolean(read_tmp, 'interpolate_terrain'))
         domain_za.append(config.getboolean(read_tmp, 'use_palm_z_axis')) 
         if domain_ip[ndomains-1] and not domain_za[ndomains-1]:
            domain_za[ndomains-1] = True
            print("+++ Overwrite user setting for use_palm_z_axis")
         
         domain_parent.append(config.get(read_tmp, 'domain_parent'))

         domain_x0.append(int(floor(float(config.get(read_tmp, 'origin_x'))/float(config.get(read_tmp, 'pixel_size')))))
         domain_y0.append(int(floor(float(config.get(read_tmp, 'origin_y'))/float(config.get(read_tmp, 'pixel_size')))))       
         domain_x1.append(int(floor(float(config.get(read_tmp, 'origin_x'))/float(config.get(read_tmp, 'pixel_size'))) + int(config.get(read_tmp, 'nx'))))
         domain_y1.append(int(floor(float(config.get(read_tmp, 'origin_y'))/float(config.get(read_tmp, 'pixel_size'))) + int(config.get(read_tmp, 'ny'))))
         domain_green_roofs.append(config.getboolean(read_tmp, 'vegetation_on_roofs')) 
         domain_street_trees.append(config.getboolean(read_tmp, 'street_trees')) 
         domain_overhanging_trees.append(config.getboolean(read_tmp, 'overhanging_trees')) 
                 
      if ( read_tmp.split("_")[0] == 'input' ):
         input_names.append(read_tmp.split("_")[1])
         input_px.append(float(config.get(read_tmp, 'pixel_size')))
         input_file_x.append(config.get(read_tmp, 'path') + "/" + config.get(read_tmp, 'file_x'))
         input_file_y.append(config.get(read_tmp, 'path') + "/" + config.get(read_tmp, 'file_y')) 
         input_file_lat.append(config.get(read_tmp, 'path') + "/" + config.get(read_tmp, 'file_lat'))
         input_file_lon.append(config.get(read_tmp, 'path') + "/" + config.get(read_tmp, 'file_lon'))   
         input_file_x_UTM.append(config.get(read_tmp, 'path') + "/" + config.get(read_tmp, 'file_x_UTM'))
         input_file_y_UTM.append(config.get(read_tmp, 'path') + "/" + config.get(read_tmp, 'file_y_UTM'))  
         input_file_zt.append(config.get(read_tmp, 'path') + "/" + config.get(read_tmp, 'file_zt'))
         input_file_buildings_2d.append(config.get(read_tmp, 'path') + "/" + config.get(read_tmp, 'file_buildings_2d'))
         input_file_bridges_2d.append(config.get(read_tmp, 'path') + "/" + config.get(read_tmp, 'file_bridges_2d'))
         input_file_building_id.append(config.get(read_tmp, 'path') + "/" + config.get(read_tmp, 'file_building_id'))  
         input_file_bridges_id.append(config.get(read_tmp, 'path') + "/" + config.get(read_tmp, 'file_bridges_id')) 
         input_file_building_type.append(config.get(read_tmp, 'path') + "/" + config.get(read_tmp, 'file_building_type'))
         input_file_lai.append(config.get(read_tmp, 'path') + "/" + config.get(read_tmp, 'file_lai'))         
         input_file_vegetation_type.append(config.get(read_tmp, 'path') + "/" + config.get(read_tmp, 'file_vegetation_type')) 
         input_file_vegetation_height.append(config.get(read_tmp, 'path') + "/" + config.get(read_tmp, 'file_vegetation_height'))          
         input_file_pavement_type.append(config.get(read_tmp, 'path') + "/" + config.get(read_tmp, 'file_pavement_type'))    
         input_file_water_type.append(config.get(read_tmp, 'path') + "/" + config.get(read_tmp, 'file_water_type'))
         input_file_street_type.append(config.get(read_tmp, 'path') + "/" + config.get(read_tmp, 'file_street_type'))    
         input_file_street_crossings.append(config.get(read_tmp, 'path') + "/" + config.get(read_tmp, 'file_street_crossings')) 
         input_file_patch_height.append(config.get(read_tmp, 'path') + "/" + config.get(read_tmp, 'file_patch_height')) 
         
         tmp = config.get(read_tmp, 'file_tree_crown_diameter')
         if tmp is not None:
            input_file_tree_crown_diameter.append(config.get(read_tmp, 'path') + "/" + tmp)  
         else:
            input_file_tree_crown_diameter.append(None)
         input_file_tree_height.append(config.get(read_tmp, 'path') + "/" + config.get(read_tmp, 'file_tree_height'))  
         input_file_tree_trunk_diameter.append(config.get(read_tmp, 'path') + "/" + config.get(read_tmp, 'file_tree_trunk_diameter'))  
         input_file_tree_type.append(config.get(read_tmp, 'path') + "/" + config.get(read_tmp, 'file_tree_type'))  
         input_file_vegetation_on_roofs.append(config.get(read_tmp, 'path') + "/" + config.get(read_tmp, 'file_vegetation_on_roofs'))           
         #input_file_soil_type.append(config.get(read_tmp, 'path') + "/" + config.get(read_tmp, 'file_soil_type'))    
   return 0


############################################################

# Start of main program
datatypes = {
   "x": "f4",  
   "y": "f4", 
   "z": "f4", 
   "lat": "f4",  
   "lon": "f4",   
   "E_UTM": "f4",  
   "N_UTM": "f4", 
   "zt": "f4",
   "buildings_2d": "f4",
   "buildings_3d": "b",
   "bridges_2d": "f4",
   "building_id": "i",
   "bridges_id": "i",
   "building_type": "b",
   "nsurface_fraction": "i",
   "vegetation_type": "b",
   "vegetation_height": "f4",
   "pavement_type": "b", 
   "water_type": "b", 
   "street_type": "b",  
   "street_crossings": "b",     
   "soil_type": "b",
   "surface_fraction": "f4",
   "building_pars": "f4",
   "vegetation_pars": "f4",
   "tree_data": "f4",
   "tree_id": "i", 
   "tree_type": "f4",
   "tree_types": "b",  
   "nbuilding_pars": "i",
   "nvegetation_pars": "i",
   "zlad": "f4"
   }

fillvalues = {
   "lat": float(-9999.0),
   "lon": float(-9999.0),
   "E_UTM": float(-9999.0),
   "N_UTM": float(-9999.0),   
   "zt": float(-9999.0),
   "buildings_2d": float(-9999.0),
   "buildings_3d": np.byte(-127),
   "bridges_2d": float(-9999.0),
   "building_id": int(-9999),
   "bridges_id": int(-9999),
   "building_type": np.byte(-127),
   "nsurface_fraction": int(-9999),
   "vegetation_type": np.byte(-127),
   "vegetation_height": float(-9999.0),
   "pavement_type": np.byte(-127),
   "water_type": np.byte(-127),
   "street_type": np.byte(-127), 
   "street_crossings": np.byte(-127),   
   "soil_type": np.byte(-127),
   "surface_fraction": float(-9999.0),
   "building_pars": float(-9999.0),
   "vegetation_pars": float(-9999.0),
   "tree_data": float(-9999.0),
   "tree_id": int(-9999),
   "tree_type": float(-9999.0),
   "tree_types": np.byte(-127)
   }

defaultvalues = {
   "lat": float(-9999.0),
   "lon": float(-9999.0),
   "E_UTM": float(-9999.0),
   "N_UTM": float(-9999.0),  
   "zt": float(0.0),
   "buildings_2d": float(0.0),
   "buildings_3d": 0,   
   "bridges_2d": float(0.0),
   "building_id": int(0),
   "bridges_id": int(0),
   "building_type": 1,
   "nsurface_fraction": int(-9999),
   "vegetation_type": 3,
   "vegetation_height": float(-9999.0),
   "pavement_type": 1, 
   "water_type": 1, 
   "street_type": 1, 
   "street_crossings": 0,    
   "soil_type": 1,
   "surface_fraction": float(0.0),
   "buildings_pars": float(-9999.0),
   "tree_data": float(-9999.0),
   "tree_type": np.byte(-127),
   "vegetation_pars": float(-9999.0)   
   }
  
# Read configuration file and set parameters accordingly
read_config_file()


filename = []
ii = []
ii_parent = []
# Define indices and filenames for all domains and create netCDF files
for i in range(0,ndomains):

#  Calculate indices and input files
   ii.append(input_px.index(domain_px[i]))
   filename.append(path_out + "/" + filename_out + "_" + domain_names[i])
   if domain_parent[i] is not None:
      ii_parent.append(input_px.index(domain_px[domain_names.index(domain_parent[i])]))
   else:
      ii_parent.append(None)


   x_UTM = nc_read_from_file_2d(input_file_x[ii[i]], "Band1", domain_x0[i], domain_x0[i]+1, domain_y0[i], domain_y0[i]+1)
   y_UTM = nc_read_from_file_2d(input_file_y[ii[i]], "Band1", domain_x0[i], domain_x0[i]+1, domain_y0[i], domain_y0[i]+1)
   lat = nc_read_from_file_2d(input_file_lat[ii[i]], "Band1", domain_x0[i], domain_x0[i]+1, domain_y0[i], domain_y0[i]+1)
   lon = nc_read_from_file_2d(input_file_lon[ii[i]], "Band1", domain_x0[i], domain_x0[i]+1, domain_y0[i], domain_y0[i]+1)  
   
#  Calculate position of origin
   x_UTM_origin = float(x_UTM[0,0]) - 0.5 * (float(x_UTM[0,1]) - float(x_UTM[0,0]))
   y_UTM_origin = float(y_UTM[0,0]) - 0.5 * (float(y_UTM[1,0]) - float(y_UTM[0,0]))  
   x_origin = float(lon[0,0]) - 0.5 * (float(lon[0,1]) - float(lon[0,0]))
   y_origin = float(lat[0,0]) - 0.5 * (float(lat[1,0]) - float(lat[0,0]))   
   
#  Create NetCDF output file and set global attributes
   nc_create_file(filename[i])
   nc_write_global_attributes(filename[i],x_UTM_origin,y_UTM_origin,y_origin,x_origin,"",global_acronym,global_angle,global_author,global_campaign,global_comment,global_contact,global_data_content,global_dependencies,global_institution,global_keywords,global_location,global_palm_version,global_references,global_site,global_source,version_out)

   del x_UTM, y_UTM, lat, lon
   
# Process terrain height
for i in range(0,ndomains):

#  Read and write terrain height (zt)
   zt = nc_read_from_file_2d(input_file_zt[ii[i]], 'Band1', domain_x0[i], domain_x1[i], domain_y0[i], domain_y1[i])

#  Final step: add zt array to the global array   
   zt_all.append(zt)
   del zt
 
#  Calculate the global (all domains) minimum of the terrain height. This value will be substracted for all
#  data sets
zt_min = min(zt_all[0].flatten())  
for i in range(0,ndomains):
   zt_min = min(zt_min,min(zt_all[i].flatten()))
   
del zt_all[:]

print( "Shift terrain heights by -" + str(zt_min))
for i in range(0,ndomains):

#  Read and write terrain height (zt)
   zt = nc_read_from_file_2d(input_file_zt[ii[i]], 'Band1', domain_x0[i], domain_x1[i], domain_y0[i], domain_y1[i])
   x = nc_read_from_file_1d(input_file_x[ii[i]], "x", domain_x0[i], domain_x1[i])
   y = nc_read_from_file_1d(input_file_y[ii[i]], "y", domain_y0[i], domain_y1[i])

 
   zt = zt - zt_min
   
   nc_write_global_attribute(filename[i], 'origin_z', float(zt_min))

#  If necessary, interpolate parent domain terrain height on child domain grid and blend the two
   if domain_ip[i]:
      parent_id = domain_names.index(domain_parent[i])
      tmp_x0 = int( domain_x0[i] * domain_px[i] / domain_px[parent_id] ) - 1
      tmp_y0 = int( domain_y0[i] * domain_px[i] / domain_px[parent_id] ) - 1
      tmp_x1 = int( domain_x1[i] * domain_px[i] / domain_px[parent_id] ) + 1
      tmp_y1 = int( domain_y1[i] * domain_px[i] / domain_px[parent_id] ) + 1

      tmp_x = nc_read_from_file_1d(input_file_x[ii_parent[i]], "x", tmp_x0, tmp_x1)
      tmp_y = nc_read_from_file_1d(input_file_y[ii_parent[i]], "y", tmp_y0, tmp_y1)

      zt_parent = nc_read_from_file_2d(input_file_zt[ii_parent[i]], 'Band1', tmp_x0, tmp_x1, tmp_y0, tmp_y1)

      zt_parent = zt_parent - zt_min

#     Interpolate array and bring to PALM grid of child domain
      zt_ip = interpolate_2d(zt_parent,tmp_x,tmp_y,x,y)
      zt_ip = bring_to_palm_grid(zt_ip,x,y,domain_dz[parent_id])
      
      
#     Shift the child terrain height according to the parent mean terrain height 
      print("shifting: -" + str(np.mean(zt)) + " +" + str(np.mean(zt_ip)))
      #zt = zt - np.min(zt) + np.min(zt_ip)
      zt = zt - np.mean(zt) + np.mean(zt_ip)
  
  
#     Blend over the parent and child terrain height within a radius of 50 px  
      zt = blend_array_2d(zt,zt_ip,50)
     # zt = zt_ip
   
#  Final step: add zt array to the global array   

   zt_all.append(zt)
   del zt
 

# Read and shift x and y coordinates, shift terrain height according to its minimum value and write all data
# to file  
for i in range(0,ndomains):   
#  Read horizontal grid variables from zt file and write them to output file
   x = nc_read_from_file_1d(input_file_x[ii[i]], "x", domain_x0[i], domain_x1[i])
   y = nc_read_from_file_1d(input_file_y[ii[i]], "y", domain_y0[i], domain_y1[i])  
   x = x - min(x.flatten()) + domain_px[i]/2.0
   y = y - min(y.flatten()) + domain_px[i]/2.0  
   nc_write_dimension(filename[i], 'x', x, datatypes["x"])
   nc_write_dimension(filename[i], 'y', y, datatypes["y"])  
   nc_write_attribute(filename[i], 'x', 'long_name', 'x')
   nc_write_attribute(filename[i], 'x', 'standard_name','projection_x_coordinate')   
   nc_write_attribute(filename[i], 'x', 'units', 'm')
   nc_write_attribute(filename[i], 'y', 'long_name', 'y')
   nc_write_attribute(filename[i], 'y', 'standard_name', 'projection_y_coordinate')   
   nc_write_attribute(filename[i], 'y', 'units', 'm')

   lat = nc_read_from_file_2d(input_file_lat[ii[i]], "Band1", domain_x0[i], domain_x1[i], domain_y0[i], domain_y1[i])
   lon = nc_read_from_file_2d(input_file_lon[ii[i]], "Band1", domain_x0[i], domain_x1[i], domain_y0[i], domain_y1[i])  

   nc_write_to_file_2d(filename[i], 'lat', lat, datatypes["lat"],'y','x',fillvalues["lat"])
   nc_write_to_file_2d(filename[i], 'lon', lon, datatypes["lon"],'y','x',fillvalues["lon"])  
   
   nc_write_attribute(filename[i], 'lat', 'long_name', 'latitude')
   nc_write_attribute(filename[i], 'lat', 'standard_name','latitude')   
   nc_write_attribute(filename[i], 'lat', 'units', 'degrees_north')
 
   nc_write_attribute(filename[i], 'lon', 'long_name', 'longitude')
   nc_write_attribute(filename[i], 'lon', 'standard_name','longitude')   
   nc_write_attribute(filename[i], 'lon', 'units', 'degrees_east')  
   
   x_UTM = nc_read_from_file_2d(input_file_x_UTM[ii[i]], "Band1", domain_x0[i], domain_x1[i], domain_y0[i], domain_y1[i])
   y_UTM = nc_read_from_file_2d(input_file_y_UTM[ii[i]], "Band1", domain_x0[i], domain_x1[i], domain_y0[i], domain_y1[i])  

   
   nc_write_to_file_2d(filename[i], 'E_UTM', x_UTM, datatypes["E_UTM"],'y','x',fillvalues["E_UTM"])
   nc_write_to_file_2d(filename[i], 'N_UTM', y_UTM, datatypes["N_UTM"],'y','x',fillvalues["N_UTM"])  
   
   nc_write_attribute(filename[i], 'E_UTM', 'long_name', 'easting')
   nc_write_attribute(filename[i], 'E_UTM', 'standard_name','projection_x_coorindate')   
   nc_write_attribute(filename[i], 'E_UTM', 'units', 'm')
 
   nc_write_attribute(filename[i], 'N_UTM', 'long_name', 'northing')
   nc_write_attribute(filename[i], 'N_UTM', 'standard_name','projection_y_coorindate')   
   nc_write_attribute(filename[i], 'N_UTM', 'units', 'm')  
  
   nc_write_crs(filename[i])
   
  
  
#  If necessary, bring terrain height to PALM's vertical grid. This is either forced by the user or implicitly
#  by using interpolation for a child domain        
   if domain_za[i]:
      zt_all[i] = bring_to_palm_grid(zt_all[i],x,y,domain_dz[i])

   nc_write_to_file_2d(filename[i], 'zt', zt_all[i], datatypes["zt"],'y','x',fillvalues["zt"])
   nc_write_attribute(filename[i], 'zt', 'long_name', 'orography')
   nc_write_attribute(filename[i], 'zt', 'units', 'm')
   nc_write_attribute(filename[i], 'zt', 'res_orig', domain_px[i]) 
   nc_write_attribute(filename[i], 'zt', 'coordinates', 'E_UTM N_UTM lon lat') 
   nc_write_attribute(filename[i], 'zt', 'grid_mapping', 'E_UTM N_UTM lon lat')   

del zt_all


#  Process building height, id, and type
for i in range(0,ndomains):  
   buildings_2d = nc_read_from_file_2d(input_file_buildings_2d[ii[i]], 'Band1', domain_x0[i], domain_x1[i], domain_y0[i], domain_y1[i])

   building_id = nc_read_from_file_2d(input_file_building_id[ii[i]], 'Band1', domain_x0[i], domain_x1[i], domain_y0[i], domain_y1[i])  
 
   building_type = nc_read_from_file_2d(input_file_building_type[ii[i]], 'Band1', domain_x0[i], domain_x1[i], domain_y0[i], domain_y1[i])  
   building_type[building_type >= 254] = fillvalues["building_type"]
   building_type = np.where(building_type < 1,defaultvalues["building_type"],building_type)
 
   check = check_arrays_2(buildings_2d,building_id,fillvalues["buildings_2d"],fillvalues["building_id"])
   if not check:
      buildings_2d = np.where(building_id != fillvalues["building_id"],buildings_2d,fillvalues["buildings_2d"])
      building_id = np.where(buildings_2d == fillvalues["buildings_2d"],fillvalues["building_id"],building_id)
      print("Data check #1 " + str(check_arrays_2(buildings_2d,building_id,fillvalues["buildings_2d"],fillvalues["building_id"])))

   check = check_arrays_2(buildings_2d,building_type,fillvalues["buildings_2d"],fillvalues["building_type"])
   if not check:
      building_type = np.where(buildings_2d == fillvalues["buildings_2d"],fillvalues["building_type"],building_type)
      building_type = np.where((building_type == fillvalues["building_type"]) & (buildings_2d != fillvalues["buildings_2d"]),defaultvalues["building_type"],building_type)
      print("Data check #2 " + str(check_arrays_2(buildings_2d,building_type,fillvalues["buildings_2d"],fillvalues["building_type"])))
 

   nc_write_to_file_2d(filename[i], 'buildings_2d', buildings_2d, datatypes["buildings_2d"],'y','x',fillvalues["buildings_2d"])
   nc_write_attribute(filename[i], 'buildings_2d', 'long_name', 'buildings')
   nc_write_attribute(filename[i], 'buildings_2d', 'units', 'm')
   nc_write_attribute(filename[i], 'buildings_2d', 'res_orig', domain_px[i])  
   nc_write_attribute(filename[i], 'buildings_2d', 'lod', 1) 
   nc_write_attribute(filename[i], 'buildings_2d', 'coordinates', 'E_UTM N_UTM lon lat') 
   nc_write_attribute(filename[i], 'buildings_2d', 'grid_mapping', 'E_UTM N_UTM lon lat')   
   
   nc_write_to_file_2d(filename[i], 'building_id', building_id, datatypes["building_id"],'y','x',fillvalues["building_id"])
   nc_write_attribute(filename[i], 'building_id', 'long_name', 'building id')
   nc_write_attribute(filename[i], 'building_id', 'units', '')
   nc_write_attribute(filename[i], 'building_id', 'res _orig', domain_px[i]) 
   nc_write_attribute(filename[i], 'building_id', 'coordinates', 'E_UTM N_UTM lon lat') 
   nc_write_attribute(filename[i], 'building_id', 'grid_mapping', 'E_UTM N_UTM lon lat')   
   
   nc_write_to_file_2d(filename[i], 'building_type', building_type, datatypes["building_type"],'y','x',fillvalues["building_type"])
   nc_write_attribute(filename[i], 'building_type', 'long_name', 'building type')
   nc_write_attribute(filename[i], 'building_type', 'units', '')
   nc_write_attribute(filename[i], 'building_type', 'res_orig', domain_px[i])  
   nc_write_attribute(filename[i], 'building_type', 'coordinates', 'E_UTM N_UTM lon lat') 
   nc_write_attribute(filename[i], 'building_type', 'grid_mapping', 'E_UTM N_UTM lon lat')   
 
del buildings_2d
del building_id
del building_type

# Create 3d buildings if necessary. In that course, read bridge objects and add them to building layer
for i in range(0,ndomains): 
   
   if domain_3d[i]:
      x = nc_read_from_file_2d_all(filename[i], 'x')
      y = nc_read_from_file_2d_all(filename[i], 'y')  
      buildings_2d = nc_read_from_file_2d_all(filename[i], 'buildings_2d') 
      building_id = nc_read_from_file_2d_all(filename[i], 'building_id') 
      
      bridges_2d = nc_read_from_file_2d(input_file_bridges_2d[ii[i]], 'Band1', domain_x0[i], domain_x1[i], domain_y0[i], domain_y1[i])
      bridges_id = nc_read_from_file_2d(input_file_bridges_id[ii[i]], 'Band1', domain_x0[i], domain_x1[i], domain_y0[i], domain_y1[i])
      
      bridges_2d = np.where(bridges_2d == 0.0,fillvalues["bridges_2d"],bridges_2d)
      building_id = np.where(bridges_2d == fillvalues["bridges_2d"],building_id,bridges_id)
      
      if np.any(buildings_2d != fillvalues["buildings_2d"]):
         buildings_3d, z = make_3d_from_2d(buildings_2d,x,y,domain_dz[i])
         if np.any(bridges_2d != fillvalues["bridges_2d"]):
            buildings_3d = make_3d_from_bridges_2d(buildings_3d,bridges_2d,x,y,domain_dz[i],settings_bridge_width,fillvalues["bridges_2d"])
         else:
            print("Skipping creation of 3D bridges (no bridges in domain)")
            
            
         nc_write_dimension(filename[i], 'z', z, datatypes["z"])  
         nc_write_attribute(filename[i], 'z', 'long_name', 'z')  
         nc_write_attribute(filename[i], 'z', 'units', 'm')
         
         nc_overwrite_to_file_2d(filename[i], 'building_id', building_id)  
 
         nc_write_to_file_3d(filename[i], 'buildings_3d', buildings_3d, datatypes["buildings_3d"],'z','y','x',fillvalues["buildings_3d"])    
         nc_write_attribute(filename[i], 'buildings_3d', 'long_name', 'buildings 3d')
         nc_write_attribute(filename[i], 'buildings_3d', 'units', '')
         nc_write_attribute(filename[i], 'buildings_3d', 'res_orig', domain_px[i])  
         nc_write_attribute(filename[i], 'buildings_3d', 'lod', 2)  
         
         del buildings_3d
         
      else:
         print("Skipping creation of 3D buildings (no buildings in domain)")


      del bridges_2d, bridges_id, building_id, buildings_2d



# Read vegetation type, water_type, pavement_type, soil_type and make fields consistent
for i in range(0,ndomains):  

   building_type = nc_read_from_file_2d_all(filename[i], 'building_type') 
   
   vegetation_type = nc_read_from_file_2d(input_file_vegetation_type[ii[i]], 'Band1', domain_x0[i], domain_x1[i], domain_y0[i], domain_y1[i])   
   vegetation_type[vegetation_type == 255] = fillvalues["vegetation_type"]
   vegetation_type = np.where((vegetation_type < 1) & (vegetation_type != fillvalues["vegetation_type"]),defaultvalues["vegetation_type"],vegetation_type)
   
   pavement_type = nc_read_from_file_2d(input_file_pavement_type[ii[i]], 'Band1', domain_x0[i], domain_x1[i], domain_y0[i], domain_y1[i])   
   pavement_type[pavement_type == 255] = fillvalues["pavement_type"]
   pavement_type = np.where((pavement_type < 1) & (pavement_type != fillvalues["pavement_type"]),defaultvalues["pavement_type"],pavement_type)

   water_type = nc_read_from_file_2d(input_file_water_type[ii[i]], 'Band1', domain_x0[i], domain_x1[i], domain_y0[i], domain_y1[i])   
   water_type[water_type == 255] = fillvalues["water_type"]
   water_type = np.where((water_type < 1) & (water_type != fillvalues["water_type"]),defaultvalues["water_type"],water_type)
  
#  to do: replace by real soil input data
   soil_type = nc_read_from_file_2d(input_file_vegetation_type[ii[i]], 'Band1', domain_x0[i], domain_x1[i], domain_y0[i], domain_y1[i])  
   soil_type[soil_type == 255] = fillvalues["soil_type"]
   soil_type = np.where((soil_type < 1) & (soil_type != fillvalues["soil_type"]),defaultvalues["soil_type"],soil_type)

#  Make arrays consistent
#  #1 Set vegetation type to missing for pixel where a pavement type is set
   vegetation_type = np.where((vegetation_type != fillvalues["vegetation_type"]) & (pavement_type != fillvalues["pavement_type"]),fillvalues["vegetation_type"],vegetation_type)

#  #2 Set vegetation type to missing for pixel where a building type is set
   vegetation_type = np.where((vegetation_type != fillvalues["vegetation_type"]) & (building_type != fillvalues["building_type"]) ,fillvalues["vegetation_type"],vegetation_type)

#  #3 Set vegetation type to missing for pixel where a building type is set
   vegetation_type = np.where((vegetation_type != fillvalues["vegetation_type"]) & (water_type != fillvalues["water_type"]),fillvalues["vegetation_type"],vegetation_type)   

#  #4 Remove pavement for pixels with buildings 
   pavement_type = np.where((pavement_type != fillvalues["pavement_type"]) & (building_type != fillvalues["building_type"]),fillvalues["pavement_type"],pavement_type)  

#  #5 Remove pavement for pixels with water.
   pavement_type = np.where((pavement_type != fillvalues["pavement_type"]) & (water_type != fillvalues["water_type"]),fillvalues["pavement_type"],pavement_type) 
   
#  #6 Remove water for pixels with buildings
   water_type = np.where((water_type != fillvalues["water_type"]) & (building_type != fillvalues["building_type"]),fillvalues["water_type"],water_type) 

#  Correct vegetation_type when a vegetation height is available and is indicative of low vegeetation
   vegetation_height = nc_read_from_file_2d(input_file_vegetation_height[ii[i]], 'Band1', domain_x0[i], domain_x1[i], domain_y0[i], domain_y1[i])   
         
   vegetation_type = np.where((vegetation_height != fillvalues["vegetation_height"]) & (vegetation_height == 0.0) & ((vegetation_type == 4) | (vegetation_type == 5) | (vegetation_type == 6) |(vegetation_type == 7) | (vegetation_type == 17) | (vegetation_type == 18)), 3, vegetation_type)
   vegetation_height = np.where((vegetation_height != fillvalues["vegetation_height"]) & (vegetation_height == 0.0) & ((vegetation_type == 4) | (vegetation_type == 5) | (vegetation_type == 6) |(vegetation_type == 7) | (vegetation_type == 17) | (vegetation_type == 18)), fillvalues["vegetation_height"],vegetation_height)
  
#  Check for consistency and fill empty fields with default vegetation type
   consistency_array, test = check_consistency_4(vegetation_type,building_type,pavement_type,water_type,fillvalues["vegetation_type"],fillvalues["building_type"],fillvalues["pavement_type"],fillvalues["water_type"])
   
   if test:
      vegetation_type = np.where(consistency_array == 0,defaultvalues["vegetation_type"],vegetation_type)
      consistency_array, test = check_consistency_4(vegetation_type,building_type,pavement_type,water_type,fillvalues["vegetation_type"],fillvalues["building_type"],fillvalues["pavement_type"],fillvalues["water_type"])

#  #7 to be removed: set default soil type everywhere
   soil_type = np.where((vegetation_type != fillvalues["vegetation_type"]) | (pavement_type != fillvalues["pavement_type"]),defaultvalues["soil_type"],fillvalues["soil_type"]) 
        
   
#  Check for consistency and fill empty fields with default vegetation type
   consistency_array, test = check_consistency_3(vegetation_type,pavement_type,soil_type,fillvalues["vegetation_type"],fillvalues["pavement_type"],fillvalues["soil_type"])
       
#  Create surface_fraction array
   x = nc_read_from_file_2d_all(filename[i], 'x')
   y = nc_read_from_file_2d_all(filename[i], 'y')  
   nsurface_fraction = np.arange(0,3)
   surface_fraction = np.ones((len(nsurface_fraction),len(y),len(x)))
   
   surface_fraction[0,:,:] = np.where(vegetation_type != fillvalues["vegetation_type"], 1.0, 0.0)
   surface_fraction[1,:,:] = np.where(pavement_type != fillvalues["pavement_type"], 1.0, 0.0)
   surface_fraction[2,:,:] = np.where(water_type != fillvalues["water_type"], 1.0, 0.0)
   
   nc_write_dimension(filename[i], 'nsurface_fraction', nsurface_fraction, datatypes["nsurface_fraction"]) 
   nc_write_to_file_3d(filename[i], 'surface_fraction', surface_fraction, datatypes["surface_fraction"],'nsurface_fraction','y','x',fillvalues["surface_fraction"])     
   nc_write_attribute(filename[i], 'surface_fraction', 'long_name', 'surface fraction')
   nc_write_attribute(filename[i], 'surface_fraction', 'units', '')
   nc_write_attribute(filename[i], 'surface_fraction', 'res_orig', domain_px[i])     
   del surface_fraction   

   nc_write_to_file_2d(filename[i], 'vegetation_type', vegetation_type, datatypes["vegetation_type"],'y','x',fillvalues["vegetation_type"])     
   nc_write_attribute(filename[i], 'vegetation_type', 'long_name', 'vegetation type')
   nc_write_attribute(filename[i], 'vegetation_type', 'units', '')
   nc_write_attribute(filename[i], 'vegetation_type', 'res_orig', domain_px[i])  
   nc_write_attribute(filename[i], 'vegetation_type', 'coordinates', 'E_UTM N_UTM lon lat') 
   nc_write_attribute(filename[i], 'vegetation_type', 'grid_mapping', 'E_UTM N_UTM lon lat')   
   del vegetation_type

   nc_write_to_file_2d(filename[i], 'pavement_type', pavement_type, datatypes["pavement_type"],'y','x',fillvalues["pavement_type"])    
   nc_write_attribute(filename[i], 'pavement_type', 'long_name', 'pavement type')
   nc_write_attribute(filename[i], 'pavement_type', 'units', '')
   nc_write_attribute(filename[i], 'pavement_type', 'res_orig', domain_px[i])  
   nc_write_attribute(filename[i], 'pavement_type', 'coordinates', 'E_UTM N_UTM lon lat') 
   nc_write_attribute(filename[i], 'pavement_type', 'grid_mapping', 'E_UTM N_UTM lon lat')   
   del pavement_type

   nc_write_to_file_2d(filename[i], 'water_type', water_type, datatypes["water_type"],'y','x',fillvalues["water_type"]) 
   nc_write_attribute(filename[i], 'water_type', 'long_name', 'water type')
   nc_write_attribute(filename[i], 'water_type', 'units', '')
   nc_write_attribute(filename[i], 'water_type', 'res_orig', domain_px[i]) 
   nc_write_attribute(filename[i], 'water_type', 'coordinates', 'E_UTM N_UTM lon lat') 
   nc_write_attribute(filename[i], 'water_type', 'grid_mapping', 'E_UTM N_UTM lon lat')   
   del water_type

   nc_write_to_file_2d(filename[i], 'soil_type', soil_type, datatypes["soil_type"],'y','x',fillvalues["soil_type"])  
   nc_write_attribute(filename[i], 'soil_type', 'long_name', 'soil type')
   nc_write_attribute(filename[i], 'soil_type', 'units', '')
   nc_write_attribute(filename[i], 'soil_type', 'res_orig', domain_px[i])   
   nc_write_attribute(filename[i], 'soil_type', 'coordinates', 'E_UTM N_UTM lon lat') 
   nc_write_attribute(filename[i], 'soil_type', 'grid_mapping', 'E_UTM N_UTM lon lat')   
   del soil_type

   del x
   del y

#  pixels with bridges get building_type = 7 = bridge. This does not change the _type setting for the under-bridge
#  area NOTE: when bridges are present the consistency check will fail at the moment
   if domain_3d[i]:
      if np.any(building_type != fillvalues["building_type"]):
 
         bridges_2d = nc_read_from_file_2d(input_file_bridges_2d[ii[i]], 'Band1', domain_x0[i], domain_x1[i], domain_y0[i], domain_y1[i])         
         bridges_2d = np.where(bridges_2d == 0.0,fillvalues["bridges_2d"],bridges_2d)
         building_type = np.where(bridges_2d != fillvalues["bridges_2d"],7,building_type)
         nc_overwrite_to_file_2d(filename[i], 'building_type', building_type)  
 
         del building_type  
         del bridges_2d

# Read/write street type and street crossings
for i in range(0,ndomains):  

   street_type = nc_read_from_file_2d(input_file_street_type[ii[i]], 'Band1', domain_x0[i], domain_x1[i], domain_y0[i], domain_y1[i])   
   street_type[street_type == 255] = fillvalues["street_type"]
   street_type = np.where((street_type < 1) & (street_type != fillvalues["street_type"]),defaultvalues["street_type"],street_type)

   pavement_type = nc_read_from_file_2d_all(filename[i], 'pavement_type') 
   street_type = np.where((pavement_type == fillvalues["pavement_type"]),fillvalues["street_type"],street_type) 

   nc_write_to_file_2d(filename[i], 'street_type', street_type, datatypes["street_type"],'y','x',fillvalues["street_type"]) 
   nc_write_attribute(filename[i], 'street_type', 'long_name', 'street type')
   nc_write_attribute(filename[i], 'street_type', 'units', '')
   nc_write_attribute(filename[i], 'street_type', 'res_orig', domain_px[i])  
   nc_write_attribute(filename[i], 'street_type', 'coordinates', 'E_UTM N_UTM lon lat') 
   nc_write_attribute(filename[i], 'street_type', 'grid_mapping', 'E_UTM N_UTM lon lat')   
   del street_type
   
   street_crossings = nc_read_from_file_2d(input_file_street_crossings[ii[i]], 'Band1', domain_x0[i], domain_x1[i], domain_y0[i], domain_y1[i])   
   street_crossings[street_crossings == 255] = fillvalues["street_crossings"]
   street_crossings = np.where((street_crossings < 1) & (street_crossings != fillvalues["street_crossings"]),defaultvalues["street_crossings"],street_crossings)
   
   nc_write_to_file_2d(filename[i], 'street_crossing', street_crossings, datatypes["street_crossings"],'y','x',fillvalues["street_crossings"]) 
   nc_write_attribute(filename[i], 'street_crossing', 'long_name', 'street crossings')
   nc_write_attribute(filename[i], 'street_crossing', 'units', '')
   nc_write_attribute(filename[i], 'street_crossing', 'res_orig', domain_px[i])  
   nc_write_attribute(filename[i], 'street_crossing', 'coordinates', 'E_UTM N_UTM lon lat') 
   nc_write_attribute(filename[i], 'street_crossing', 'grid_mapping', 'E_UTM N_UTM lon lat')   
   del street_crossings


# Read/write vegetation on roofs
for i in range(0,ndomains):  
   if domain_green_roofs[i]: 
      green_roofs = nc_read_from_file_2d(input_file_vegetation_on_roofs[ii[i]], 'Band1', domain_x0[i], domain_x1[i], domain_y0[i], domain_y1[i])   
      buildings_2d = nc_read_from_file_2d_all(filename[i], 'buildings_2d') 


      x = nc_read_from_file_2d_all(filename[i], 'x')
      y = nc_read_from_file_2d_all(filename[i], 'y')  
      nbuilding_pars = np.arange(0,47)
      building_pars = np.ones((len(nbuilding_pars),len(y),len(x)))
      building_pars[:,:,:] = fillvalues["building_pars"]

#     assign green fraction on roofs
      building_pars[3,:,:] = np.where( ( buildings_2d != fillvalues["buildings_2d"] ) & ( green_roofs != fillvalues["building_pars"] ) & ( green_roofs != 0.0 ),1,fillvalues["building_pars"] )

#     assign leaf area index for vegetation on roofs
      building_pars[4,:,:] = np.where( ( building_pars[3,:,:] != fillvalues["building_pars"] ) & ( green_roofs >= 0.5 ),settings_lai_roof_intensive,fillvalues["building_pars"])
      building_pars[4,:,:] = np.where( ( building_pars[3,:,:] != fillvalues["building_pars"] ) & ( green_roofs < 0.5 ),settings_lai_roof_extensive,building_pars[4,:,:])
   
   
      nc_write_dimension(filename[i], 'nbuilding_pars', nbuilding_pars, datatypes["nbuilding_pars"]) 
      nc_write_to_file_3d(filename[i], 'building_pars', building_pars, datatypes["building_pars"],'nbuilding_pars','y','x',fillvalues["building_pars"]) 
      nc_write_attribute(filename[i], 'building_pars', 'long_name', 'building_pars')
      nc_write_attribute(filename[i], 'building_pars', 'units', '')
      nc_write_attribute(filename[i], 'building_pars', 'res_orig', domain_px[i])  
      nc_write_attribute(filename[i], 'building_pars', 'coordinates', 'E_UTM N_UTM lon lat') 
      nc_write_attribute(filename[i], 'building_pars', 'grid_mapping', 'E_UTM N_UTM lon lat')   
         
      del building_pars, buildings_2d, x, y


# Read tree data and create LAD and BAD arrays using the canopy generator
for i in range(0,ndomains):  
   lai = nc_read_from_file_2d(input_file_lai[ii[i]], 'Band1', domain_x0[i], domain_x1[i], domain_y0[i], domain_y1[i])   
     
   vegetation_type = nc_read_from_file_2d_all(filename[i], 'vegetation_type')
     
   lai = np.where(vegetation_type == fillvalues["vegetation_type"],fillvalues["vegetation_pars"],lai)

      
   x = nc_read_from_file_2d_all(filename[i], 'x')
   y = nc_read_from_file_2d_all(filename[i], 'y')  
   nvegetation_pars = np.arange(0,12)
   vegetation_pars = np.ones((len(nvegetation_pars),len(y),len(x)))
   vegetation_pars[:,:,:] = fillvalues["vegetation_pars"]  
      
   vegetation_pars[1,:,:] = lai
      

#  Write out first version of LAI. Will later be overwritten.
   nc_write_dimension(filename[i], 'nvegetation_pars', nvegetation_pars, datatypes["nvegetation_pars"]) 
   nc_write_to_file_3d(filename[i], 'vegetation_pars', vegetation_pars, datatypes["vegetation_pars"],'nvegetation_pars','y','x',fillvalues["vegetation_pars"]) 
   nc_write_attribute(filename[i], 'vegetation_pars', 'long_name', 'vegetation_pars')
   nc_write_attribute(filename[i], 'vegetation_pars', 'units', '')
   nc_write_attribute(filename[i], 'vegetation_pars', 'res_orig', domain_px[i])  
   nc_write_attribute(filename[i], 'vegetation_pars', 'coordinates', 'E_UTM N_UTM lon lat') 
   nc_write_attribute(filename[i], 'vegetation_pars', 'grid_mapping', 'E_UTM N_UTM lon lat')   

   del lai, vegetation_pars, vegetation_type

# Read tree data and create LAD and BAD arrays using the canopy generator
for i in range(0,ndomains):  
   if domain_street_trees[i]: 

      vegetation_pars =  nc_read_from_file_2d_all(filename[i], 'vegetation_pars')

      lai = nc_read_from_file_2d(input_file_lai[ii[i]], 'Band1', domain_x0[i], domain_x1[i], domain_y0[i], domain_y1[i])   

      x = nc_read_from_file_2d_all(filename[i], 'x')
      y = nc_read_from_file_2d_all(filename[i], 'y')  

#     Save lai data as default for low and high vegetation
      lai_low = lai
      lai_high = lai
   
#     Read all tree parameters from file
      tree_height = nc_read_from_file_2d(input_file_tree_height[ii[i]], 'Band1', domain_x0[i], domain_x1[i], domain_y0[i], domain_y1[i])   

      if (input_file_tree_crown_diameter[ii[i]] is not None):
         tree_crown_diameter = nc_read_from_file_2d(input_file_tree_crown_diameter[ii[i]], 'Band1', domain_x0[i], domain_x1[i], domain_y0[i], domain_y1[i])   
         tree_crown_diameter = np.where( (tree_crown_diameter == 0.0) | (tree_crown_diameter == -1.0) ,fillvalues["tree_data"],tree_crown_diameter)
      else:
         tree_crown_diameter = np.ones((len(y),len(x)))
         tree_crown_diameter[:,:] = fillvalues["tree_data"]


      tree_trunk_diameter = nc_read_from_file_2d(input_file_tree_trunk_diameter[ii[i]], 'Band1', domain_x0[i], domain_x1[i], domain_y0[i], domain_y1[i])         
      tree_type = nc_read_from_file_2d(input_file_tree_type[ii[i]], 'Band1', domain_x0[i], domain_x1[i], domain_y0[i], domain_y1[i]) 
      patch_height = nc_read_from_file_2d(input_file_patch_height[ii[i]], 'Band1', domain_x0[i], domain_x1[i], domain_y0[i], domain_y1[i]) 
      
#     Remove missing values from the data. Reasonable values will be set by the tree generator    
      tree_height = np.where( (tree_height == 0.0) | (tree_height == -1.0) ,fillvalues["tree_data"],tree_height)
      tree_trunk_diameter = np.where( (tree_trunk_diameter == 0.0) | (tree_trunk_diameter == -1.0) ,fillvalues["tree_data"],tree_trunk_diameter)
      tree_type = np.where( (tree_type == 0.0) | (tree_type == -1.0) ,fillvalues["tree_data"],tree_type)
      
#     For vegetation pixel with missing height information (-1), set default patch height   
      patch_height = np.where( patch_height == -1.0 ,settings_patch_height_default,patch_height)     

#     Convert trunk diameter from cm to m
      tree_trunk_diameter = np.where(tree_trunk_diameter != fillvalues["tree_data"], tree_trunk_diameter * 0.01,tree_trunk_diameter)


#     Temporarily change missing value for tree_type
      tree_type = np.where( (tree_type == fillvalues["tree_type"]),fillvalues["tree_data"],tree_type)    

#     Compare patch height array with vegetation type and correct accordingly
      vegetation_type = nc_read_from_file_2d_all(filename[i], 'vegetation_type') 


#     For zero-height patches, set vegetation_type to short grass and remove these pixels from the patch height array
      vegetation_type = np.where( (patch_height == 0.0) & ( (vegetation_type == 4) | (vegetation_type == 5) | (vegetation_type == 6) |(vegetation_type == 7) | (vegetation_type == 17) | (vegetation_type == 18) ),3,vegetation_type) 
      patch_height = np.where( (patch_height == 0.0) & ( (vegetation_type == 4) | (vegetation_type == 5) | (vegetation_type == 6) |(vegetation_type == 7) | (vegetation_type == 17) | (vegetation_type == 18) ),fillvalues["tree_data"],patch_height)       

      max_tree_height = max(tree_height.flatten())
      max_patch_height = max(patch_height.flatten())
      
#     Call canopy generator for single trees only if there is any tree height available in the domain.
#     This does not guarantee that there are street trees that can be processed. This is checked in the
#     canopy generator.
      if ( (max_tree_height != fillvalues["tree_data"]) | (max_patch_height != fillvalues["tree_data"]) ):

         lad, bad, tree_ids, tree_types, zlad = generate_single_tree_lad(x,y,domain_dz[i],max_tree_height,max_patch_height,tree_type,tree_height,tree_crown_diameter,tree_trunk_diameter,lai,settings_season,fillvalues["tree_data"])
  
#        Remove LAD volumes that are inside buildings
         if not domain_overhanging_trees[i]:
            buildings_2d = nc_read_from_file_2d_all(filename[i], 'buildings_2d') 
            for k in range(0,len(zlad)-1):
  
               lad[k,:,:] = np.where(buildings_2d == fillvalues["buildings_2d"],lad[k,:,:],fillvalues["tree_data"])
               bad[k,:,:] = np.where(buildings_2d == fillvalues["buildings_2d"],bad[k,:,:],fillvalues["tree_data"])
               tree_ids[k,:,:] = np.where(buildings_2d == fillvalues["buildings_2d"],tree_ids[k,:,:],fillvalues["tree_id"])      

            del buildings_2d
  
 
         nc_write_dimension(filename[i], 'zlad', zlad, datatypes["tree_data"]) 

         nc_write_to_file_3d(filename[i], 'lad', lad, datatypes["tree_data"],'zlad','y','x',fillvalues["tree_data"]) 
         nc_write_attribute(filename[i], 'lad', 'long_name', 'leaf area density')
         nc_write_attribute(filename[i], 'lad', 'units', '')
         nc_write_attribute(filename[i], 'lad', 'res_orig', domain_px[i])  
         nc_write_attribute(filename[i], 'lad', 'coordinates', 'E_UTM N_UTM lon lat') 
         nc_write_attribute(filename[i], 'lad', 'grid_mapping', 'E_UTM N_UTM lon lat') 
 
         nc_write_to_file_3d(filename[i], 'bad', bad, datatypes["tree_data"],'zlad','y','x',fillvalues["tree_data"]) 
         nc_write_attribute(filename[i], 'bad', 'long_name', 'basal area density')
         nc_write_attribute(filename[i], 'bad', 'units', '')
         nc_write_attribute(filename[i], 'bad', 'res_orig', domain_px[i])  
         nc_write_attribute(filename[i], 'bad', 'coordinates', 'E_UTM N_UTM lon lat') 
         nc_write_attribute(filename[i], 'bad', 'grid_mapping', 'E_UTM N_UTM lon lat')
            
         nc_write_to_file_3d(filename[i], 'tree_id', tree_ids, datatypes["tree_id"],'zlad','y','x',fillvalues["tree_id"]) 
         nc_write_attribute(filename[i], 'tree_id', 'long_name', 'tree id')
         nc_write_attribute(filename[i], 'tree_id', 'units', '')
         nc_write_attribute(filename[i], 'tree_id', 'res_orig', domain_px[i])  
         nc_write_attribute(filename[i], 'tree_id', 'coordinates', 'E_UTM N_UTM lon lat') 
         nc_write_attribute(filename[i], 'tree_id', 'grid_mapping', 'E_UTM N_UTM lon lat')           

         nc_write_to_file_3d(filename[i], 'tree_type', tree_types, datatypes["tree_types"],'zlad','y','x',fillvalues["tree_types"]) 
         nc_write_attribute(filename[i], 'tree_type', 'long_name', 'tree type')
         nc_write_attribute(filename[i], 'tree_type', 'units', '')
         nc_write_attribute(filename[i], 'tree_type', 'res_orig', domain_px[i])  
         nc_write_attribute(filename[i], 'tree_type', 'coordinates', 'E_UTM N_UTM lon lat') 
         nc_write_attribute(filename[i], 'tree_type', 'grid_mapping', 'E_UTM N_UTM lon lat')   
            
         del lai, lad, bad, tree_ids, tree_types, zlad

      else:
         print('No street trees generated in domain ' + str(i))
         
      del vegetation_pars, tree_height, tree_crown_diameter, tree_trunk_diameter, tree_type, patch_height, x, y


# Create vegetation patches for locations with high vegetation type
for i in range(0,ndomains):  
   if domain_canopy_patches[i]: 

      patch_height = nc_read_from_file_2d(input_file_patch_height[ii[i]], 'Band1', domain_x0[i], domain_x1[i], domain_y0[i], domain_y1[i]) 

#     For vegetation pixel with missing height information (-1), set default patch height      
      patch_height = np.where( patch_height == -1.0 ,settings_patch_height_default,patch_height)
      
      max_patch_height = max(patch_height.flatten())
      if ( max_patch_height != fillvalues["tree_data"] ):
#        Call canopy generator for single trees only if there is any tree height available in the domain.
#        This does not guarantee that there are street trees that can be processed. This is checked in the
#        canopy generator.

#        Load vegetation_type and lad array (at level z = 0) for re-processing
         vegetation_type = nc_read_from_file_2d_all(filename[i], 'vegetation_type') 
         lad = nc_read_from_file_3d_all(filename[i], 'lad') 
         tree_id = nc_read_from_file_3d_all(filename[i], 'tree_id') 
         tree_types = nc_read_from_file_3d_all(filename[i], 'tree_type') 
         zlad = nc_read_from_file_1d_all(filename[i], 'zlad') 
         vegetation_pars = nc_read_from_file_3d_all(filename[i], 'vegetation_pars') 
         lai = nc_read_from_file_2d(input_file_lai[ii[i]], 'Band1', domain_x0[i], domain_x1[i], domain_y0[i], domain_y1[i])   
    

#        Determine all pixels that do not already have an LAD but which are high vegetation to a dummy value of 1.0 and remove all other pixels
         lai_high = np.where( (lad[0,:,:] == fillvalues["tree_data"]) & ( ( (vegetation_type == 4) | (vegetation_type == 5) | (vegetation_type == 6) |(vegetation_type == 7) | (vegetation_type == 17) | (vegetation_type == 18) ) & ( (patch_height == fillvalues["tree_data"]) | (patch_height >= domain_dz[i])) ),1.0,fillvalues["tree_data"])


#        Treat all pixels where short grass is defined, but where a patch_height >= dz is found, as high vegetation (often the case in backyards)
         lai_high = np.where( (lai_high == fillvalues["tree_data"]) & (patch_height >= domain_dz[i]) & (vegetation_type == 3), 1.0, lai_high)
                          
#        If overhanging trees are allowed, assign pixels with patch_height > dz that are not included in vegetation_type to high
         if domain_overhanging_trees[i]:
            lai_high = np.where( (lai_high == fillvalues["tree_data"]) & (patch_height >= domain_dz[i]), 1.0, lai_high)

#        Now, assign either the default LAI for high vegetation or keep 1.0 from the lai_high array.
         lai_high = np.where( (lai_high != fillvalues["tree_data"]) & (lai == fillvalues["tree_data"]), settings_lai_high_default, lai_high)

#        If LAI values are available in the LAI array, write them on the lai_high array
         lai_high = np.where( (lai_high != fillvalues["tree_data"]) & (lai != fillvalues["tree_data"]), lai, lai_high)


#        Define a patch height wherever it is missing, but where a high vegetation LAI was set
         patch_height = np.where ( (lai_high != fillvalues["tree_data"]) & (patch_height == fillvalues["tree_data"]), settings_patch_height_default, patch_height)


#        Remove pixels where street trees were already se
         patch_height = np.where ( (lad[0,:,:] != fillvalues["tree_data"]), fillvalues["tree_data"], patch_height)

#        Remove patch heights that have no lai_high value
         patch_height = np.where ( (lai_high == fillvalues["tree_data"]), fillvalues["tree_data"], patch_height)




#        For missing LAI values, set either the high vegetation default or the low vegetation default    
         lai_high = np.where( (patch_height > 2.0) & (patch_height != fillvalues["tree_data"]) & (lai_high == fillvalues["tree_data"]),settings_lai_high_default,lai_high)
         lai_high = np.where( (patch_height <= 2.0) & (patch_height != fillvalues["tree_data"]) & (lai_high == fillvalues["tree_data"]),settings_lai_low_default,lai_high)
         
         if ( max(patch_height.flatten()) >= (2.0 * domain_dz[i]) ):
            print("    start calculating LAD (this might take some time)")       

         
            lad_patch, patch_id, patch_types, patch_nz, status = process_patch(domain_dz[i],patch_height,vegetation_type,max(zlad),lai_high,settings_lai_alpha,settings_lai_beta)

#           Set negative ids for patches
            patch_id = np.where( patch_id != fillvalues["tree_id"], -patch_id, patch_id)

#           2D loop in order to avoid memory problems with large arrays
            for iii in range(0,domain_nx[i]):
               for jj in range(0,domain_ny[i]):            

                  tree_id[0:patch_nz+1,jj,iii] = np.where( (lad[0:patch_nz+1,jj,iii] == fillvalues["tree_data"]),patch_id[0:patch_nz+1,jj,iii], tree_id[0:patch_nz+1,jj,iii] )
                  tree_types[0:patch_nz+1,jj,iii] = np.where( (lad[0:patch_nz+1,jj,iii] == fillvalues["tree_data"]),patch_types[0:patch_nz+1,jj,iii], tree_types[0:patch_nz+1,jj,iii] )                  
                  lad[0:patch_nz+1,jj,iii]     = np.where( (lad[0:patch_nz+1,jj,iii] == fillvalues["tree_data"]),lad_patch[0:patch_nz+1,jj,iii], lad[0:patch_nz+1,jj,iii] )

#        Remove high vegetation wherever it is replaced by a leaf area density. This should effectively remove all high vegetation pixels
         vegetation_type = np.where((lad[0,:,:] != fillvalues["tree_data"]) & (vegetation_type != fillvalues["vegetation_type"]),settings_veg_type_below_trees,vegetation_type)
              
#        If desired, remove all high vegetation. TODO: check if this is still necessary
         if not domain_high_vegetation[i]:
            vegetation_type = np.where((vegetation_type != fillvalues["vegetation_type"]) & ( (vegetation_type == 4) | (vegetation_type == 5) | (vegetation_type == 6) |(vegetation_type == 7) | (vegetation_type == 17) | (vegetation_type == 18) ),3,vegetation_type)   

  
#        Set default low LAI for pixels with an LAD (short grass below trees)
         lai_low = np.where( (lad[0,:,:] == fillvalues["tree_data"]), lai, settings_lai_low_default)
    
#        Fill low vegetation pixels without LAI set or with LAI = 0 with default value
         lai_low = np.where( ( (lai_low == fillvalues["tree_data"]) | (lai_low == 0.0) ) & (vegetation_type != fillvalues["vegetation_type"] ), settings_lai_low_default, lai_low)

#        Remove lai for pixels that have no vegetation_type
         lai_low = np.where( ( (vegetation_type != fillvalues["vegetation_type"]) & (vegetation_type != 1) ), lai_low, fillvalues["tree_data"])
        
#        Overwrite lai in vegetation_parameters  
         vegetation_pars[1,:,:] = np.copy(lai_low)
         nc_overwrite_to_file_3d(filename[i], 'vegetation_pars', vegetation_pars) 

#        Overwrite lad and id arrays
         nc_overwrite_to_file_3d(filename[i], 'lad', lad) 
         nc_overwrite_to_file_3d(filename[i], 'tree_id', tree_id) 
         nc_overwrite_to_file_3d(filename[i], 'tree_type', tree_types) 

         nc_overwrite_to_file_2d(filename[i], 'vegetation_type', vegetation_type)  


         del vegetation_type, lad, lai, patch_height, vegetation_pars, zlad
      else:
         print('No tree patches found in domain ' + str(i))

# Final adjustment of vegetation parameters: remove LAI where a bare soil was set
for i in range(0,ndomains): 
   
      vegetation_type = nc_read_from_file_2d_all(filename[i], 'vegetation_type') 
      vegetation_pars = nc_read_from_file_3d_all(filename[i], 'vegetation_pars') 
      lai = vegetation_pars[1,:,:]   
      
      
#     Remove lai for pixels that have no vegetation_type
      lai = np.where( ( (vegetation_type != fillvalues["vegetation_type"]) & (vegetation_type != 1) ), lai, fillvalues["tree_data"])
        
#     Overwrite lai in vegetation_parameters  
      vegetation_pars[1,:,:] = np.copy(lai)
      nc_overwrite_to_file_3d(filename[i], 'vegetation_pars', vegetation_pars) 
      
      del vegetation_type, lai, vegetation_pars


# Final consistency check
for i in range(0,ndomains): 
   vegetation_type = nc_read_from_file_2d_all(filename[i], 'vegetation_type')  
   pavement_type = nc_read_from_file_2d_all(filename[i], 'pavement_type')  
   building_type = nc_read_from_file_2d_all(filename[i], 'building_type')  
   water_type = nc_read_from_file_2d_all(filename[i], 'water_type')     
   soil_type = nc_read_from_file_2d_all(filename[i], 'soil_type')  

#  Check for consistency and fill empty fields with default vegetation type
   consistency_array, test = check_consistency_4(vegetation_type,building_type,pavement_type,water_type,fillvalues["vegetation_type"],fillvalues["building_type"],fillvalues["pavement_type"],fillvalues["water_type"])

#  Check for consistency and fill empty fields with default vegetation type
   consistency_array, test = check_consistency_3(vegetation_type,pavement_type,soil_type,fillvalues["vegetation_type"],fillvalues["pavement_type"],fillvalues["soil_type"])

   surface_fraction = nc_read_from_file_3d_all(filename[i], 'surface_fraction')       
   surface_fraction[0,:,:] = np.where(vegetation_type != fillvalues["vegetation_type"], 1.0, 0.0)
   surface_fraction[1,:,:] = np.where(pavement_type != fillvalues["pavement_type"], 1.0, 0.0)
   surface_fraction[2,:,:] = np.where(water_type != fillvalues["water_type"], 1.0, 0.0)
   nc_overwrite_to_file_3d(filename[i], 'surface_fraction', surface_fraction) 
       
   del vegetation_type, pavement_type, building_type, water_type, soil_type