def write_raster_to_disk(raster,
out_filename,
in_filename=r"data/Strat4/DTM_metres_clip.tif"):
import rasterio
with rasterio.open(
in_filename
) as src: #src file is needed to output with the same metadata and attributes
profile = src.profile
_, wtd_old, _, _, _ = preprocess_data.read_preprocess_rasters(
in_filename, in_filename, in_filename, in_filename, in_filename)
profile.update(nodata=None) # overrun nodata value given by input raster
profile.update(
width=wtd_old.shape[1]
) # Shape of rater is changed for hydro simulation inside read_preprocess_rasters. Here we take that shape to output consistently.
profile.update(height=wtd_old.shape[0])
# profile.update(dtype='float32', compress='lzw') # What compression to use?
profile.update(
dtype='float32'
) # instead of 64. To save space, we don't need so much precision. float16 is not supported by GDAL, check: https://github.com/mapbox/rasterio/blob/master/rasterio/dtypes.py
with rasterio.open(out_filename, 'w', **profile) as dst:
dst.write(raster.astype(dtype='float32'), 1)
return 0
def write_raster_multiband(nbands,
rasters_array,
STUDY_AREA,
out_filename,
ref_filename=r"data/Strat4/DTM_metres_clip.tif"):
# Rasters_array should be a numpy array with the form [bands, rows, col]
# This means that if there are 4 bands in the raster, the shape of the array
# should be (4, nrows, ncols)
# Metadata is copied from the raster at ref_filename
import rasterio
import preprocess_data
with rasterio.open(
ref_filename
) as src: #src file is needed to output with the same metadata and attributes
profile = src.profile
_, _, ref_raster, _, _, _, _ = preprocess_data.read_preprocess_rasters(
STUDY_AREA, ref_filename, ref_filename, ref_filename, ref_filename,
ref_filename, ref_filename, ref_filename)
if nbands != rasters_array.shape[0]:
raise ValueError("Number of bands does not match raster array shape")
profile.update(nodata=None) # overrun nodata value given by input raster
profile.update(
width=ref_raster.shape[1]
) # Shape of rater is changed for hydro simulation inside read_preprocess_rasters. Here we take that shape to output consistently.
profile.update(height=ref_raster.shape[0])
profile.update(count=nbands)
profile.update(
dtype='float32'
) # instead of 64. To save space, we don't need so much precision. float16 is not supported by GDAL, check: https://github.com/mapbox/rasterio/blob/master/rasterio/dtypes.py
with rasterio.open(out_filename, 'w', **profile) as dst:
dst.write(rasters_array.astype(dtype='float32'))
return 0
#land_use_rst_fn = preprocessed_datafolder + r"/Landcover2017_clip.tif" # Not used
peat_depth_rst_fn = preprocessed_datafolder + r"/Peattypedepth_clip.tif" # peat depth, peat type in the same raster
abs_path_data = os.path.abspath(
'./data'
) # Absolute path to data folder needed for Excel file with parameters
params_fn = abs_path_data + r"/params.xlsx"
if 'CNM' and 'cr' and 'c_to_r_list' not in globals():
CNM, cr, c_to_r_list = preprocess_data.gen_can_matrix_and_raster_from_raster(
can_rst_fn=can_rst_fn, dem_rst_fn=dem_rst_fn)
#else:
# print("Canal adjacency matrix and raster loaded from memory.")
_, dem, peat_type_arr, peat_depth_arr = preprocess_data.read_preprocess_rasters(
can_rst_fn, dem_rst_fn, peat_depth_rst_fn, peat_depth_rst_fn)
PARAMS_df = preprocess_data.read_params(params_fn)
BLOCK_HEIGHT = PARAMS_df.block_height[0]
CANAL_WATER_LEVEL = PARAMS_df.canal_water_level[0]
DIRI_BC = PARAMS_df.diri_bc[0]
HINI = PARAMS_df.hini[0]
P = PARAMS_df.P[0]
ET = PARAMS_df.ET[0]
TIMESTEP = PARAMS_df.timeStep[0]
KADJUST = PARAMS_df.Kadjust[0]
print(">>>>> WARNING, OVERWRITING PEAT DEPTH")
peat_depth_arr[peat_depth_arr < 2.] = 2.
# catchment mask
cnm_sim = cnm + cnm.T
CNM = scipy.sparse.csr_matrix(cnm) # In order to use the same sparse type as the big ass true adjacency matrix
else:
filenames_df = pd.read_excel('file_pointers.xlsx', header=2, dtype=str)
dem_rst_fn = Path(filenames_df[filenames_df.Content == 'DEM'].Path.values[0])
can_rst_fn = Path(filenames_df[filenames_df.Content == 'canal_raster'].Path.values[0])
# TODO: use peat depth as depth of canals??
peat_depth_rst_fn = Path(filenames_df[filenames_df.Content == 'peat_depth_raster'].Path.values[0])
# Choose smaller study area
STUDY_AREA = (0,-1), (0,-1) # E.g., a study area of (0,-1), (0,-1) is the whole domain
can_arr, wtd_old , dem, _, peat_depth_arr, _, _ = preprocess_data.read_preprocess_rasters(STUDY_AREA, dem_rst_fn, can_rst_fn, dem_rst_fn, peat_depth_rst_fn, peat_depth_rst_fn, dem_rst_fn, dem_rst_fn)
labelled_canals = preprocess_data.label_canal_pixels(can_arr, dem)
CNM, c_to_r_list = preprocess_data.gen_can_matrix_and_label_map(labelled_canals, dem)
dem_nodes = [dem[loc] for loc in c_to_r_list]
dem_nodes[0] = 3.0 # something strange happens with this node
dem_nodes = np.array(dem_nodes)
n_edges = np.sum(CNM)
n_nodes = CNM.shape[0]
DIST_BETWEEN_NODES = 100. # m
#%%
# Hydrology-relevant data: BC, IC, rainfall, etc.
[get_day_rainfall(api_url)]
) * 25.4 # From inches to mm. array type is to allow for list of precip. Usually, single value is used.
except:
raise ConnectionError(
"Could not get precipitation data from the weather station.")
"""
LOOP FOR HISTORICAL PRECIPITATION DATA
If only daily precipitation, then precip is a list of length 1.
"""
for d, P in enumerate(precip):
if HISTORIC_PRECIPITATION and d == 0:
wtd_old = 0. # The initial condition is fully saturated.
wtd_old_fn = dem_rst_fn # this works for raster reading and writing functions
_, _, dem, peat_type_arr, peat_depth_arr = preprocess_data.read_preprocess_rasters(
wtd_old_fn, can_rst_fn, dem_rst_fn, peat_depth_rst_fn,
peat_depth_rst_fn)
else:
wtd_old_fn = previous_wtd_fname(WTD_folder, hp=HISTORIC_PRECIPITATION)
_, wtd_old, dem, peat_type_arr, peat_depth_arr = preprocess_data.read_preprocess_rasters(
wtd_old_fn, can_rst_fn, dem_rst_fn, peat_depth_rst_fn,
peat_depth_rst_fn)
"""
RUN HYDROLOGICAL MODEL
"""
DAYS = 1 # number of days with daily timestep for hydrological module
N_BLOCKS = 0
# Generate adjacency matrix, and dictionary. Need to do this every time?
CNM, cr, c_to_r_list = preprocess_data.gen_can_matrix_and_raster_from_raster(
"""
filenames_df = pd.read_excel('file_pointers.xlsx', header=2, dtype=str)
dem_rst_fn = Path(filenames_df[filenames_df.Content == 'DEM'].Path.values[0])
can_rst_fn = Path(filenames_df[filenames_df.Content == 'canal_raster'].Path.values[0])
peat_depth_rst_fn = Path(filenames_df[filenames_df.Content == 'peat_depth_raster'].Path.values[0])
blocks_fn = Path(filenames_df[filenames_df.Content == 'canal_blocks_raster'].Path.values[0])
sensor_loc_fn = Path(filenames_df[filenames_df.Content == 'sensor_locations'].Path.values[0])
params_fn = Path(filenames_df[filenames_df.Content == 'parameters'].Path.values[0])
WTD_folder = Path(filenames_df[filenames_df.Content == 'WTD_input_and_output_folder'].Path.values[0])
weather_fn = Path(filenames_df[filenames_df.Content == 'historic_precipitation'].Path.values[0])
# Choose smaller study area
STUDY_AREA = (0,-1), (0,-1)
wtd_old_fn = dem_rst_fn # not reading from previous wtd raster
can_arr, wtd_old , dem, peat_type_arr, peat_depth_arr, blocks_arr, sensor_loc_arr = preprocess_data.read_preprocess_rasters(STUDY_AREA, wtd_old_fn, can_rst_fn, dem_rst_fn, peat_depth_rst_fn, peat_depth_rst_fn, blocks_fn, sensor_loc_fn)
sensor_loc_indices = utilities.get_sensor_loc_array_indices(sensor_loc_arr)
if 'CNM' and 'labelled_canals' and 'c_to_r_list' not in globals():
labelled_canals = preprocess_data.label_canal_pixels(can_arr, dem)
CNM, c_to_r_list = preprocess_data.gen_can_matrix_and_label_map(labelled_canals, dem)
built_block_positions = utilities.get_already_built_block_positions(blocks_arr, labelled_canals)
PARAMS_df = preprocess_data.read_params(params_fn)
BLOCK_HEIGHT = PARAMS_df.block_height[0]; CANAL_WATER_LEVEL = PARAMS_df.canal_water_level[0]
DIRI_BC = PARAMS_df.diri_bc[0]; HINI = PARAMS_df.hini[0]; P = PARAMS_df.P[0]
ET = PARAMS_df.ET[0]; TIMESTEP = PARAMS_df.timeStep[0]; KADJUST = PARAMS_df.Kadjust[0]
print(">>>>> WARNING, OVERWRITING PEAT DEPTH")