-
Notifications
You must be signed in to change notification settings - Fork 0
/
process_palm_results.py
319 lines (249 loc) · 12.5 KB
/
process_palm_results.py
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
'''
Script to process the PALM results
- first: potential temperature is extracted and converted to air temperature, then buildings are clipped out of the data
- second: urban rural temperatures are calculated for both domains
- third: palm results at the location of the LMSS stations are extracted for a comparison
- fourth: palm energy balance results at the location of the LMSS stations are extracted
- fifth: palm results at the location of the Netatmo stations are extracted for statistical evaluation
Results from step 1 and 2 are saved as netcdf, coordinates are defined in those files to allow for display in QGIS at the correct georeferende location
Results from step 3 and 4 are saved as csv files for analysis in R
'''
## load required libraries
import sys
import xarray as xr
import numpy as np
import geopandas
import rioxarray as rio
from shapely.geometry import mapping
from pyproj import Proj
from xarray.core.common import T
import pandas as pd
## load data
# palm results
palm_parent = xr.open_dataset("/media/lara/2TB SSD/Data_MA/Model_output/city/isotropic/v9/MA_v9_av_xy.nc")
palm_child = xr.open_dataset("/media/lara/2TB SSD/Data_MA/Model_output/city/isotropic/v9/MA_v9_av_xy_N02.nc")
child_3d = xr.open_dataset("/media/lara/2TB SSD/Data_MA/Model_output/city/isotropic/v9/MA_v9_av_3d_N02.nc")
# shape files
buildings_p = geopandas.read_file("/media/lara/2TB SSD/Data_MA/Model_input/10m_city/buildings_10m_city.shp", crs = "epsg:25832")
buildings_c = geopandas.read_file("/media/lara/2TB SSD/Data_MA/Model_input/city/buildings_city_new2.shp", crs = "epsg:25832")
lcz = geopandas.read_file("~/Dokumente/Data_processing/data_in/LCZ_14.shp", crs = "epsg:25832")
# netatmo data
netatmo = pd.read_csv("~/Dokumente/Data_processing/data_in/netatmo_utm.csv")
##### part 1: convert potential temperature to air temperature for the parent and child domain ###############################################
# extract theta at 2m
theta_2m_par = palm_parent.variables["theta_2m*_xy"]
theta_2m_child = palm_child.variables["theta_2m*_xy"]
## convert to absolute temperature
# necessary variables for conversion
# gas constant of dry air
r = 287.05
# specific heat capacity of air
cp = 1004.67
# reference pressure
p0 = 1000
# current pressure
p = 1006
# calculate ta
divider = (p0/p)**(r/cp)
ta_2m_p = theta_2m_par/divider
ta_2m_c = theta_2m_child/divider
# convert to °C
ta_2m_p = np.round(ta_2m_p - 273.15, 1)
ta_2m_p = np.squeeze(ta_2m_p)
ta_2m_c = np.round(ta_2m_c - 273.15, 1)
ta_2m_c = np.squeeze(ta_2m_c)
## extract latitude and longitude to define crs for display in QGIS
# utm coordinates
utm_y_p = palm_parent.variables["N_UTM"]
utm_x_p = palm_parent.variables["E_UTM"]
utm_y_c = palm_child.variables["N_UTM"]
utm_x_c = palm_child.variables["E_UTM"]
# define utm projection as proj object
utm_proj = Proj("+proj=utm +zone=32 +ellps=GRS80 +towgs84=0,0,0,0,0,0,0 +units=m +no_defs")
# coordinates as arrays
utm_x_arr_p = np.array(utm_x_p)
utm_y_arr_p = np.array(utm_y_p)
utm_x_arr_c = np.array(utm_x_c)
utm_y_arr_c = np.array(utm_y_c)
# meshgrid of coordinates
x_mesh_p, y_mesh_p = np.meshgrid(utm_x_arr_p, utm_y_arr_p)
x_mesh_c, y_mesh_c = np.meshgrid(utm_x_arr_c, utm_y_arr_c)
# convert utm coordinates to lat/lon
lon_p, lat_p = utm_proj(x_mesh_p, y_mesh_p, inverse=True)
lon_c, lat_c = utm_proj(x_mesh_c, y_mesh_c, inverse=True)
## extract time
time = palm_parent.coords["time"]
## convert to xarray DataArray, x and y coordinates must contain UTM coordinates
ta_2m_parent =xr.DataArray(data=ta_2m_p, dims=["time","y","x"], \
coords=dict(x=(["x"], utm_x_p), y=(["y"], utm_y_p), time = time, lat = (["y","x"], lat_p), lon = (["y","x"], lon_p)))
ta_2m_child =xr.DataArray(data=ta_2m_c, dims=["time","y","x"], \
coords=dict(x=(["x"], utm_x_c), y=(["y"], utm_y_c), time = time, lat = (["y","x"], lat_c), lon = (["y","x"], lon_c)))
## save as tiff
ta_2m_parent = ta_2m_parent.rio.set_spatial_dims(x_dim="x",y_dim="y")
ta_2m_parent.rio.write_crs("epsg:25832", inplace=True)
ta_2m_parent.rio.to_raster("data_paper/ta_2m_parent.tiff")
ta_2m_child = ta_2m_child.rio.set_spatial_dims(x_dim="x",y_dim="y")
ta_2m_child.rio.write_crs("epsg:25832", inplace=True)
ta_2m_child.rio.to_raster("data_paper/ta_2m_child.tiff")
## clip out values at buildings locations
# use an invert = True in clip to set values to NA at building locations
ta_2m_parent_clip = ta_2m_parent.rio.clip(geometries = buildings_p.geometry.apply(mapping), crs=25832, invert = True)
ta_2m_child_clip = ta_2m_child.rio.clip(geometries = buildings_c.geometry.apply(mapping), crs=25832, invert = True)
## save as tiff
ta_2m_parent_clip = ta_2m_parent_clip.rio.set_spatial_dims(x_dim="x",y_dim="y")
ta_2m_parent_clip.rio.write_crs("epsg:25832", inplace=True)
ta_2m_parent_clip.rio.to_raster("data_paper/ta_2m_parent_clip.tiff")
ta_2m_child_clip = ta_2m_child_clip.rio.set_spatial_dims(x_dim="x",y_dim="y")
ta_2m_child_clip.rio.write_crs("epsg:25832", inplace=True)
ta_2m_child_clip.rio.to_raster("data_paper/ta_2m_child_clip.tiff")
#### part 2: calculate the urban rural temperature differences ##############################################################################
## clip out results pixel within LCZ 14, shape with LCZ preprocessed in QGIS
ta_rural_parent = ta_2m_parent_clip.rio.clip(geometries = lcz.geometry.apply(mapping), crs = 25832)
## calculate mean "rural" ta for city and la
ta_rural_mean = ta_rural_parent.reduce(np.nanmean, dim = ("x","y"))
## calculate uhi
uhi_p = np.round(ta_2m_p - ta_rural_mean, 1)
uhi_c = np.round(ta_2m_c - ta_rural_mean, 1)
## combine as DataArray
uhi_parent =xr.DataArray(data=uhi_p, dims=["time","y","x"], \
coords=dict(x=(["x"], utm_x_p), y=(["y"], utm_y_p), time = time, lat = (["y","x"], lat_p), lon = (["y","x"], lon_p)))
uhi_child =xr.DataArray(data=uhi_c, dims=["time","y","x"], \
coords=dict(x=(["x"], utm_x_c), y=(["y"], utm_y_c), time = time, lat = (["y","x"], lat_c), lon = (["y","x"], lon_c)))
## save as tiff
uhi_parent = uhi_parent.rio.set_spatial_dims(x_dim="x",y_dim="y")
uhi_parent.rio.write_crs("epsg:25832", inplace=True)
uhi_parent.rio.to_raster("data_paper/uhi_parent.tiff")
uhi_child = uhi_child.rio.set_spatial_dims(x_dim="x",y_dim="y")
uhi_child.rio.write_crs("epsg:25832", inplace=True)
uhi_child.rio.to_raster("data_paper/uhi_child.tiff")
##### part 3: combine the palm results with the lmss data ######################################################################################
## get the index of the required coordinates
# garden
i = np.argmin([np.abs(utm_x_c - 376130.12)])
j = np.argmin([np.abs(utm_y_c - 5706005.70)])
# roof
k = np.argmin([np.abs(utm_x_c - 375959.80)])
l = np.argmin([np.abs(utm_y_c - 5705596.18)])
# extract wind and rh variables from dataset
wind = palm_child.variables["wspeed_10m*_xy"]
rh = child_3d.variables["rh"]
# extract variable for desired location via indexing
ta_lmss = ta_2m_c[:,j-1:j+2,i-1:i+2]
wind_lmss = wind[:,:,l-1:l+2,k-1:k+2]
rh_lmss = rh[:,:,j-1:j+2,i-1:i+2]
# calculate mean
ta_lmss_mean = ta_lmss.reduce(np.nanmean, dim=("x","y"))
wind_lmss_mean = wind_lmss.reduce(np.nanmean, dim=("x","y"))
rh_lmss_mean = rh_lmss.reduce(np.nanmean, dim=("x","y"))
# change into array and round values
ta_lmss_mean = np.array(ta_lmss_mean).round(1)
wind_lmss_mean = np.array(wind_lmss_mean).round(1)
rh_lmss_mean = np.array(rh_lmss_mean).round(1)
# remove unneccessary dimensions
ta_lmss_squz = np.squeeze(ta_lmss_mean)
wind_lmss_sqz = np.squeeze(wind_lmss_mean)
# get index of first non nan value in rh array and slice array
rh_index = np.where(np.isnan(rh_lmss_mean[0,:]))[0][-1] +1
rh_lmss_mean = rh_lmss_mean[:,rh_index]
## get time array
# add timedelta to origin time to get each time step as datetime
date = np.datetime64('2020-08-08T00:00:00') + np.array(time)
## combine variable and datetime
# join in pd dataframe and print to check
ta_rh_time = pd.DataFrame({"date":date,"palm_ta":ta_lmss_squz,"palm_rh":rh_lmss_mean})
wind_time = pd.DataFrame({"date":date, "palm_ws":wind_lmss_sqz})
# save as csv
ta_rh_time.to_csv("~/Dokumente/Data_processing/data_paper/palm_lmss.csv")
wind_time.to_csv("~/Dokumente/Data_processing/data_paper/palm_lmss_ws.csv")
##### part 4: extract palm energy balance at lmss location #########################################################################################
# get energy balance variables
sens_hf = palm_child.variables["shf*_xy"]
lat_hf = palm_child.variables["qsws*_xy"]
ground_hf = palm_child.variables["ghf*_xy"]
# extract energy balance at lmss location
sens_hf_lmss = sens_hf[:,:,j-1:j+2,i-1:i+2]
lat_hf_lmss = lat_hf[:,:,j-1:j+2,i-1:i+2]
ground_hf_lmss = ground_hf[:,:,j-1:j+2,i-1:i+2]
# calculate means
sens_hf_mean = sens_hf_lmss.reduce(np.nanmean, dim=("x","y"))
lat_hf_mean = lat_hf_lmss.reduce(np.nanmean, dim=("x","y"))
ground_hf_mean = ground_hf_lmss.reduce(np.nanmean, dim=("x","y"))
# round and change into np array
sens_hf_mean = np.array(sens_hf_mean).round(1)
lat_hf_mean = np.array(lat_hf_mean).round(1)
ground_hf_mean = np.array(ground_hf_mean).round(1)
# remove empty dimensions
sens_hf_squz = np.squeeze(sens_hf_mean)
lat_hf_squz = np.squeeze(lat_hf_mean)
ground_hf_squz = np.squeeze(ground_hf_mean)
# combine variables in dataframe
energy = pd.DataFrame({"date":date,"sens_hf":sens_hf_squz,"lat_hf":lat_hf_squz,"ground_hf":ground_hf_squz})
# save
energy.to_csv("~/Dokumente/Data_processing/data_out/palm_lmss_energy.csv")
##### part 5: combine palm results with netatmo data ################################################################################################
## reduce netatmo dataframe to individual stations and only utm coordinates
stations = netatmo.drop_duplicates("intern_id")
stations = stations[["intern_id","utm_x","utm_y"]]
# filter for study areas by coordinates
# parent
stations = stations.loc[(stations["utm_x"] >= 372200.0) & (stations["utm_x"] <=381000.0)]
stations = stations.loc[(stations["utm_y"] >= 5702550.0) & (stations["utm_y"] <=5711050.0)]
# child
stations2 = stations.loc[(stations["utm_x"] >= 375480.0) & (stations["utm_x"] <=377280.0)]
stations2 = stations2.loc[(stations2["utm_y"] >= 5705110.0) & (stations2["utm_y"] <=5706610.0)]
## extract palm cells for each station
# change coordinates to numpy array
x = np.array(utm_x_p)
y = np.array(utm_y_p)
x2 = np.array(utm_x_c)
y2 = np.array(utm_y_c)
# parent
# initialise empty numpy arrays, specify data type for time array
palm_data = np.zeros(0)
time_arr = np.zeros(0, dtype=np.datetime64)
ids = np.zeros(0)
# use for loop to get data at netatmo locations and average data from radius
for i in range(0,len(stations.index)):
# get index off cell with station
index_x = np.abs(x - stations.iloc[i,1]).argmin()
index_y = np.abs(y - stations.iloc[i,2]).argmin()
# slice out cell with station and surrounding stations, calculate mean
# parent: 9 cells
cells = ta_2m_parent_clip[:,index_y-1:index_y+2,index_x-1:index_x+2]
cell_mean = cells.reduce(np.nanmean, dim=("x","y"))
# convert to numpy array and round values
cell_mean = np.array(cell_mean).round(1)
# create numpy array with station id
station_id = np.full(len(cell_mean), stations.iloc[i,0])
# append to numpy arrays
palm_data = np.r_[palm_data, cell_mean]
ids = np.r_[ids, station_id]
time_arr = np.r_[time_arr, date]
# child
# initialise empty numpy arrays, specify data type for time array
palm_data2 = np.zeros(0)
time_arr2 = np.zeros(0, dtype=np.datetime64)
ids2 = np.zeros(0)
# use for loop to get data at netatmo locations and average data from radius
for i in range(0,len(stations2.index)):
# get index off cell with station
index_x = np.abs(x2 - stations2.iloc[i,1]).argmin()
index_y = np.abs(y2 - stations2.iloc[i,2]).argmin()
# slice out cell with station and surrounding stations, calculate mean
# child: 25 cells
cells = ta_2m_child_clip[:,index_y-2:index_y+3,index_x-2:index_x+3]
cell_mean = cells.reduce(np.nanmean, dim=("x","y"))
# convert to numpy array and round values
cell_mean = np.array(cell_mean).round(1)
# create numpy array with station id
station_id = np.full(len(cell_mean), stations2.iloc[i,0])
# append to numpy arrays
palm_data2 = np.r_[palm_data2, cell_mean]
ids2 = np.r_[ids2, station_id]
time_arr2 = np.r_[time_arr2, date]
# merge all arrays in dataframe
palm_df = pd.DataFrame({"station_id":ids, "time":time_arr, "palm_ta":palm_data}, columns=["station_id","time","palm_ta"])
palm_df2 = pd.DataFrame({"station_id":ids2, "time":time_arr2, "palm_ta":palm_data2}, columns=["station_id","time","palm_ta"])
# save as csv
palm_df.to_csv("~/Dokumente/Data_processing/data_paper/palm_netatmo_parent.csv")
palm_df2.to_csv("~/Dokumente/Data_processing/data_paper/palm_netatmo_child.csv")