def write_gauge_stage_all_cells(reccsv, dirtif, wdttif, gaugelfp, stagelfp):
print(" writing gauge and stage files...")
# Reading rec file
rec = pd.read_csv(reccsv)
# Create a width dataframe
dat = gdalutils.get_data(wdttif)
geo = gdalutils.get_geo(wdttif)
wdt = gdalutils.array_to_pandas(dat, geo, 0, 'gt')
wdt.columns = ['x', 'y', 'width']
# Create directions dataframe
dat = gdalutils.get_data(dirtif)
geo = gdalutils.get_geo(dirtif)
drc = gdalutils.array_to_pandas(dat, geo, 0, 'gt')
drc.columns = ['x', 'y', 'direction']
# Find widths and directions for every lon, lat in river network
gdalutils.assign_val(df2=rec,
df2_x='lon',
df2_y='lat',
df1=wdt,
df1_x='x',
df1_y='y',
label='width',
copy=False)
gdalutils.assign_val(df2=rec,
df2_x='lon',
df2_y='lat',
df1=drc,
df1_x='x',
df1_y='y',
label='direction',
copy=False)
# Change numbers (1,2,3,4,5,6,7) to letters (N,S,E,W)
rec['direction_let'] = rec['direction'].apply(getdirletter)
# Writing .gauge file
with open(gaugelfp, 'w') as f:
f.write(str(rec.shape[0]) + '\n')
rec[['lon', 'lat', 'direction_let', 'width']].to_csv(gaugelfp,
index=False,
sep=' ',
header=False,
float_format='%.7f',
mode='a')
# Writing .stage file
with open(stagelfp, 'w') as f:
f.write(str(rec.shape[0]) + '\n')
rec[['lon', 'lat']].to_csv(stagelfp,
index=False,
sep=' ',
header=False,
float_format='%.7f',
mode='a')
def multiply_rasters(rast1, rast2, out):
dat1 = gdalutils.get_data(rast1)
dat2 = gdalutils.get_data(rast2)
geo1 = gdalutils.get_geo(rast1)
geo2 = gdalutils.get_geo(rast2)
dat_masked1 = np.ma.masked_where(dat1 == geo1[11], dat1)
dat_masked2 = np.ma.masked_where(dat2 == geo2[11], dat2)
res = dat_masked1 * dat_masked2
res.set_fill_value(-9999)
gdalutils.write_raster(res.filled(), out, geo1, "Float32", -9999)
def write_tiles_layout(mylist, outfile):
codes = []
xmins = []
ymins = []
for tile in mylist:
geo = gu.get_geo(tile)
xmin = np.round(geo[0])
ymin = np.round(geo[1])
xmins.append(xmin)
ymins.append(ymin)
codes.append(return_code(xmin, ymin))
xmins = np.unique(np.round(xmins)).astype(int)
ymins = np.unique(np.round(ymins)).astype(int)
resx = abs(xmins[0] - xmins[1])
resy = abs(ymins[1] - ymins[0])
with open(outfile, 'w') as fout:
for y in reversed(range(min(ymins), max(ymins) + resy, resy)):
for x in range(min(xmins), max(xmins) + resx, resx):
mycode = return_code(x, y)
if mycode in codes:
idx = np.where(np.array(codes) == mycode)[0][0]
fout.write(mylist[idx] + ',')
else:
fout.write(' ' * 15 + ',')
fout.write('\n')
def get_ascii_geo(filename, proj4=None):
"""
Reads LISFLOOD-FP outputs either in gz or wd file extension GEO info
Assumes WGS_84 projection by default
"""
ext = os.path.basename(filename).split('.')[-1]
if ext == 'gz':
file = _uncompress_gz(filename)
geo = gdalutils.get_geo(file, proj4=proj4)
os.remove(file)
else:
geo = gdalutils.get_geo(filename, proj4=proj4)
return geo
def run_simulation(reccsv, dirtif, widthtif, bedtif, runoffcsv, date1, date2, lisfloodfp, outfolder):
# Determine end of the simulation, how many days
t = (pd.to_datetime(date2, format='%Y-%m-%d') -
pd.to_datetime(date1, format='%Y-%m-%d')).days + 1
# Create 1D DEM, synthetic
demtif = outfolder + 'dem1d.tif'
wdt = gu.get_data(widthtif)
geo = gu.get_geo(widthtif)
dem = np.where(wdt > 0, 10000, 0)
gu.write_raster(dem, demtif, geo, 'Int16', 0)
# Convert input files to ASCII
widthasc = outfolder + 'width.asc'
call(['gdal_translate',
'-of', 'AAIGRID',
widthtif, widthasc])
demasc = outfolder + 'dem.asc'
call(['gdal_translate',
'-of', 'AAIGRID',
demtif, demasc])
bedasc = outfolder + 'bed.asc'
call(['gdal_translate',
'-of', 'AAIGRID',
bedtif, bedasc])
# Write LISFLOOD-FP files
bcilfp = outfolder + 'lfp.bci'
write_bci(bcilfp, runoffcsv)
bdylfp = outfolder + 'lfp.bdy'
write_bdy(bdylfp, runoffcsv, t)
evaplfp = outfolder + 'lfp.evap'
write_evap(evaplfp, t)
gaugelfp = outfolder + 'lfp.gauge'
stagelfp = outfolder + 'lfp.stage'
write_gauge_stage_all_cells(reccsv, dirtif, widthtif, gaugelfp, stagelfp)
parlfp = outfolder + 'lfp.par'
write_par(parlfp=parlfp,
bcilfp=bcilfp,
bdylfp=bdylfp,
evaplfp=evaplfp,
gaugelfp=gaugelfp,
stagelfp=stagelfp,
dembnktif=demasc,
wdttif=widthasc,
bedtif=bedasc,
t=t)
# Run simulation
call([lisfloodfp, '-v', 'lfp.par'], cwd=outfolder)
def create_dir_d4(dirtaud4, dirtaud8, dirtau_maskd4):
dat1 = gdalutils.get_data(dirtaud8)
dat2 = gdalutils.get_data(dirtau_maskd4)
geo = gdalutils.get_geo(dirtaud8)
A = np.where(dat2 > 0)
dat1[A] = dat2[A]
gdalutils.write_raster(dat1, dirtaud4, geo, "Int16", -32768)
def d82d4(filedir, filediro, fileneto):
"""
Returns direction and river netwrok maps in D4
"""
nodata = -32768.
dirdata = gdalutils.get_data(filedir)
dirgeo = gdalutils.get_geo(filedir)
data, net = cy_d82d4(np.int16(dirdata), np.int16(nodata))
gdalutils.write_raster(np.int16(data), filediro, dirgeo, "Int16", nodata)
gdalutils.write_raster(np.int16(net), fileneto, dirgeo, "Int16", nodata)
def rastermask(file, mask, fmt, outp):
"""
Mask input array following a defined mask (1,0)
"""
nodata = -32768
filedata = gdalutils.get_data(file)
maskdata = gdalutils.get_data(mask)
filegeo = gdalutils.get_geo(file)
data = cy_rastermask(np.float64(filedata), np.int16(maskdata))
gdalutils.write_raster(np.float64(data), outp, filegeo, fmt, nodata)
def rasterthreshold(file, thres, fmt, outp):
"""
Output a raster based on a threshold (larger-equal-than)
"""
nodata = -1
filedata = gdalutils.get_data(file)
filegeo = gdalutils.get_geo(file)
data = cy_rasterthreshold(np.float64(filedata), np.float64(thres),
np.float64(nodata))
gdalutils.write_raster(np.float64(data), outp, filegeo, fmt, nodata)
def step_04():
A = gu.get_data('delta_surf_interp.tif')
B = gu.get_data(fill_demf)
C = gu.get_data(void_demf)
geo = gu.get_geo(void_demf)
mysum = A+B
final = np.where(C==nodata,mysum,C)
final[(final>=8000) | (final<=-8000)] = nodata
gu.write_raster(final,'dem.tif',geo,'Float64',nodata)
def calculate_area(filename, output):
geo = gdalutils.get_geo(filename)
nx = np.int32(geo[4])
ny = np.int32(geo[5])
resx = np.float32(geo[6])
resy = np.float32(geo[7])
x = np.float32(geo[8])
y = np.float32(geo[9])
dat = calc_area(nx, ny, resx, resy, x, y)
gdalutils.write_raster(np.array(dat), output, geo, "Float32", -9999)
def step_01():
geo = gu.get_geo(void_demf)
void_dem = gu.get_data(void_demf)
fill_dem = gu.get_data(fill_demf)
delta_surf = void_dem - fill_dem
delta_surf[(delta_surf>=8000) | (delta_surf<=-8000)] = nodata
delta_surf[delta_surf==0] = nodata
gu.write_raster(delta_surf,'delta_surf_wt_voids.tif',geo,'Float64',nodata)
def burn_banks_dem(dembnktif, demtif, fixbnktif):
print(" burning banks in dem...")
nodata = -9999
fout = dembnktif
base = gdalutils.get_data(demtif)
basegeo = gdalutils.get_geo(demtif)
new = gdalutils.get_data(fixbnktif)
out = np.where(new > 0, new, base)
gdalutils.write_raster(out, fout, basegeo, "Float32", nodata)
def get_tiles_by_extent(xmin, ymin, xmax, ymax, tilesdir):
# Reading tiles filenames, assuming they are tif files
mylist = sorted(glob(tilesdir + '/*'))
mylist_area = []
for tile in mylist:
geo = gu.get_geo(tile)
if (geo[0] >= xmin) & (geo[1] >= ymin) & (geo[2] <= xmax) & (geo[3] <=
ymax):
mylist_area.append(tile)
return mylist_area
def burn_banks_dem_1D(dembnktif, demtif, fixbnktif):
print(" burning banks in dem 1D...")
nodata = -9999
fout = dembnktif
base = gdalutils.get_data(demtif)
basegeo = gdalutils.get_geo(demtif)
new = (np.ma.masked_values(gdalutils.get_data(fixbnktif), nodata) +
10000).filled(nodata)
out = np.where(new > 0, new, base)
gdalutils.write_raster(out, fout, basegeo, "Float32", nodata)
def write_outlets(outshp, dirtif_mask):
proj = '+proj=longlat +ellps=WGS84 +datum=WGS84 +no_defs'
dat = gdalutils.get_data(dirtif_mask)
geo = gdalutils.get_geo(dirtif_mask)
rows, cols = np.where(dat > 0)
x = []
y = []
for row, col in zip(rows, cols):
A = find_neighbours(dat, row, col)
if np.any(A < 0):
x.append(geo[8][col])
y.append(geo[9][row])
# Initiate shapefile
w = shapefile.Writer(shapefile.POINT)
w.field('x')
w.field('y')
w.field('id')
# Write coordinate points in shapefile
for i in range(len(x)):
w.point(x[i], y[i])
w.record(x[i], y[i], i)
w.save(outshp)
fname = os.path.dirname(outshp)+'/' + \
os.path.basename(outshp).split('.')[0] + '.prj'
prj = open(fname, "w")
srs = osr.SpatialReference()
srs.ImportFromProj4(proj)
prj.write(srs.ExportToWkt())
prj.close()
typ = "Byte"
fmt = "GTiff"
nodata = 0
name1 = os.path.dirname(outshp)+'/' + \
os.path.basename(outshp).split('.')[0] + '.shp'
name2 = os.path.dirname(outshp)+'/' + \
os.path.basename(outshp).split('.')[0] + '.tif'
subprocess.call([
"gdal_rasterize", "-a_nodata",
str(nodata), "-ot", typ, "-of", fmt, "-tr",
str(geo[6]),
str(geo[7]), "-burn", "1", "-a_srs", proj, "-te",
str(geo[0]),
str(geo[1]),
str(geo[2]),
str(geo[3]), name1, name2
])
def getdir(rec, dirtif):
dat = gdalutils.get_data(dirtif)
geo = gdalutils.get_geo(dirtif)
dirdf = gdalutils.array_to_pandas(dat, geo, 0, 'gt')
recdf = gdalutils.assign_val(df2=rec.reset_index(
), df2_x='lon', df2_y='lat', df1=dirdf, df1_x='x', df1_y='y', label='z', copy=True)
# Direction of outlet is given by the maximum repetitions of directions in the last 10 points
_dir = recdf.sort_values(by='distance').iloc[0:10].groupby('z')[
'z'].count().idxmax()
return _dir
def directions_esri(inputrast, outputrast):
"""
Function to change convetion from a DIR file
This script changes these numbers from TauDEM convention, 1,2,3...
to ESRI convention 128,64,32,..
"""
nodata = -32768
data = gdalutils.get_data(inputrast)
datageo = gdalutils.get_geo(inputrast)
data_esri = cy_directions_esri(np.int16(data), np.int16(nodata))
gdalutils.write_raster(np.int16(data_esri), outputrast, datageo, "Int16",
nodata)
def directions_tau(inputrast, outputrast):
"""
Function to use in Shell to change convetion from a DIR file
HydroSHEDS uses ESRI convention 128,64,32,.. this script
changes these numbers to TauDEM convention, 1,2,3...
"""
nodata = -32768
data = gdalutils.get_data(inputrast)
datageo = gdalutils.get_geo(inputrast)
datatau = cy_directions_tau(np.int16(data), np.int16(nodata))
gdalutils.write_raster(np.float64(datatau), outputrast, datageo, "Int16",
nodata)
def rasterresample(method, demf, netf, output, outlier, hrnodata, thresh,
nproc):
print(" running rasterresample.py...")
fname1 = demf
fname2 = output
# coordinates for bank elevations are based in river network mask
net = gdalutils.get_data(netf)
geo = gdalutils.get_geo(netf)
# consider all pixels in net30 including river network pixels
iy, ix = np.where(net > -1)
x = geo[8][ix]
y = geo[9][iy]
# Split x and y in nproc parts
split_x = np.array_split(x, nproc)
split_y = np.array_split(y, nproc)
# Define a queue
queue = mp.Queue()
# Setup a list of processes that we want to run
processes = []
processes = [
mp.Process(target=calc_resampling_mp,
args=(i, queue, fname1, hrnodata, split_x[i], split_y[i],
thresh, outlier, method)) for i in range(len(split_x))
]
# Run processes
for p in processes:
p.start()
# Get process results from the queue
results = [queue.get() for p in processes]
# Retrieve results in a particular order
results.sort()
results = [r[1] for r in results]
# Stack results horizontally
elev = np.hstack(results).reshape(net.shape)
# Replace NaN by hrnodata
elev[np.isnan(elev)] = hrnodata
# elev = calc_resampling(fname1,hrnodata,x,y,ix,iy,thresh,outlier,method)
gdalutils.write_raster(elev, fname2, geo, "Float32", hrnodata)
def depth_raster(w, netf, fdepth, thresh):
"""
From a raster of depths this subroutine finds nearest depth to every river pixel in grid
"""
# Reading river network file
dat_net = gdalutils.get_data(netf)
geo_net = gdalutils.get_geo(netf)
iy, ix = np.where(dat_net > 0)
xx = geo_net[8][ix]
yy = geo_net[9][iy]
# Reading depth source file
dat = gdalutils.get_data(fdepth)
geo = gdalutils.get_geo(fdepth)
iy, ix = np.where(dat > -9999)
xdat = geo[8][ix]
ydat = geo[9][iy]
depth = []
for x, y in zip(xx, yy):
try:
dis, ind = misc_utils.near_euc(xdat, ydat, (x, y))
if dis <= thresh:
val = dat[iy[ind], ix[ind]]
depth.append(val)
else:
depth.append(np.nan)
except ValueError:
depth.append(np.nan)
for x, y, mydepth in zip(xx, yy, depth):
w.point(x, y)
w.record(x, y, mydepth)
return w
def find_nearest_mean_mask(ncf,
ncproj,
lon,
lat,
proj,
thresh_dis,
thresh_mean=5):
"""
Apply a threshold to the mean discharge
Based on the thresholded map, find nearest value to lon, lat in a given perimiter and var threhsold
For JRC data default values are 5 m3s-1 and 2.5 Km
"""
# JRC data set projection is EPSG:3035
# It's required to convert to WGS84 to perform distance calculation
crs_wgs84 = Proj(init=proj)
crs_nc = Proj(init=ncproj)
# Reading mean mask
dat = gu.get_data(ncf)
geo = gu.get_geo(ncf)
# Create df, a pandas dataframe with values larger than "thresh_mean=5"
df = gu.array_to_pandas(dat, geo, thresh_mean, 'ge')
# Creating two new columns with projected values
coords = transform(crs_nc, crs_wgs84, df['x'].values, df['y'].values)
df['lon'] = coords[0]
df['lat'] = coords[1]
# Calcualte distance to lat and lon point to every point in the dataframe
vec = gu.haversine.haversine_array(np.array(df['lat'], dtype='float32'),
np.array(df['lon'], dtype='float32'),
np.float32(lat), np.float32(lon))
idx = np.argmin(vec)
dis = gu.haversine.haversine(df.loc[idx, 'lat'], df.loc[idx, 'lon'], lat,
lon)
if dis <= thresh_dis:
near_x = df.loc[idx, 'x']
near_y = df.loc[idx, 'y']
mymean = df.loc[idx, 'z']
df = None
return near_x, near_y, mymean, dis
else:
df = None
return None, None, None, None
def getdepths(proj, netf, method, output, **kwargs):
print(" runnning getdepths.py...")
fname = output
w = shapefile.Writer(shapefile.POINT)
w.field('x')
w.field('y')
w.field('depth')
if method == "depth_raster":
depth_raster(w, netf, **kwargs)
elif method == "depth_geometry":
depth_geometry(w, **kwargs)
elif method == "depth_manning":
depth_manning(w, **kwargs)
else:
sys.exit("ERROR method not recognised")
# write final value in a shapefile
w.save("%s.shp" % fname)
# write .prj file
prj = open("%s.prj" % fname, "w")
srs = osr.SpatialReference()
srs.ImportFromProj4(proj)
prj.write(srs.ExportToWkt())
prj.close()
nodata = -9999
fmt = "GTiff"
name1 = output + ".shp"
name2 = output + ".tif"
mygeo = gdalutils.get_geo(netf)
subprocess.call([
"gdal_rasterize", "-a_nodata",
str(nodata), "-of", fmt, "-tr",
str(mygeo[6]),
str(mygeo[7]), "-a", "depth", "-a_srs", proj, "-te",
str(mygeo[0]),
str(mygeo[1]),
str(mygeo[2]),
str(mygeo[3]), name1, name2
])
def step_03():
geo = gu.get_geo(void_demf)
nx = geo[4]
ny = geo[5]
xmin = geo[0]
xmax = geo[2]
ymin = geo[1]
ymax = geo[3]
subprocess.call(['gdal_grid','--config','GDAL_NUM_THREADS','ALL_CPUS',
'-a','invdist',
'-of','GTiff',
'-ot','Float64',
'-txe', str(xmin), str(xmax),
'-tye', str(ymin), str(ymax),
'-outsize', str(nx), str(ny),
'-l','delta_surf_wt_voids',
'delta_surf_wt_voids.vrt','delta_surf_interp.tif'])
def extract_from_zip(zipresults, date, date1, date2, var, proj4, outfile):
"""
Extract variable per date then convert to GTIFF
Example
-------
import lfptools.utils as lfp
proj4 = '+proj=longlat +ellps=WGS84 +datum=WGS84 +no_defs'
lfp.extract_from_zip('./176.zip','2002-08-20','1990-01-01','2014-12-31','wd',proj4,'./tmp.tif')
"""
tmpdir = os.path.dirname(outfile) + '/tmp/'
try:
os.mkdir(tmpdir)
except FileExistsError:
pass
# Open zip
myzip = zipfile.ZipFile(zipresults)
# Get list of files in zip
mylist = sorted(myzip.namelist())
myvar = [i for i in mylist if i.endswith('.' + var)]
# Simulation times
dates = pd.date_range(date1, date2)
# Retrieve filenames based on dates
ix = np.where(dates == date)[0][0]
# Extract ASCII file
myzip.extract(myvar[ix], tmpdir)
# Get info from ASCII
fname = tmpdir + myvar[ix]
dat = gdalutils.get_data(fname)
geo = gdalutils.get_geo(fname)
geo[10] = _return_projection(proj4)
gdalutils.write_raster(dat, outfile, geo, 'Float64', geo[-1])
# Remove temp folder
shutil.rmtree(tmpdir)
def getbedelevs(bnkf, dptf, netf, output, proj):
print(" running getbedelevs.py...")
bnk = gpd.read_file(bnkf)
dpt = gpd.read_file(dptf)
print('loaded data')
print('calculating bed from banks and depth')
bnk['bedelev'] = bnk['elevadj'].astype(np.float32) - dpt['depth'].astype(
np.float32)
print(bnk.keys())
# print(bnk['bedelev'])
bed = bnk[['x', 'y', 'geometry', 'bedelev']]
# bnk.columns = ['x', 'y', 'bedelev']
print('Writing out data')
# Just write bed dataframe to file, rather than creating a new dataframe with the same data
# mybed = gpd.GeoDataFrame(bnk, crs={'init': 'epsg:4326'}, geometry=[
# Point(xy) for xy in zip(bed.x.astype(float), bed.y.astype(float))])
bed.to_file(output + '.shp')
nodata = -9999
fmt = "GTiff"
# name1 = output
# name2 = os.path.dirname(output) + '/' + \
# os.path.basename(output).split('.')[0] + '.tif'
name1 = output + '.shp'
name2 = output + '.tif'
mygeo = gdalutils.get_geo(netf)
subprocess.call([
"gdal_rasterize", "-a_nodata",
str(nodata), "-of", fmt, "-ot", "Float32", "-co", "COMPRESS=DEFLATE",
"-tr",
str(mygeo[6]),
str(mygeo[7]), "-a", "bedelev", "-a_srs", proj, "-te",
str(mygeo[0]),
str(mygeo[1]),
str(mygeo[2]),
str(mygeo[3]), name1, name2
])
dptf=dpt_shp + '.shp',
netf=netf,
output=bed_shp,
proj='+proj=longlat +ellps=WGS84 +datum=WGS84 +no_defs')
#############################################################################################
# Crop LISFLOOD-FP domain
# First clip floats:
output_tifs = [
bed_shp + '.tif', wdt_shp + '.tif', dem, bnkfix_shp + '.tif',
slp_shp + '.tif', bfq_shp + '.tif', dpt_shp + '.tif'
]
for tif in output_tifs:
# Snap extent to match input tif grid cells
geo = gdalutils.get_geo(tif)
xmin0, ymin0, xmax0, ymax0 = regbound
# Geo has format [xmin, ymin, xmax, ymax, xn, yn, xres, yres, ....]
xmin = geo[0] + np.floor((xmin0 - geo[0]) / geo[6]) * geo[6]
ymin = geo[1] + np.floor((ymin0 - geo[1]) / geo[7]) * geo[7]
xmax = geo[2] + np.floor((xmax0 - geo[2]) / geo[6]) * geo[6]
ymax = geo[3] + np.floor((ymax0 - geo[3]) / geo[7]) * geo[7]
print('Clipextent:', xmin, xmax, ymin, ymax)
tifclip = outdir1 + '/' + os.path.basename(
tif[:-4]) + '_' + regname + '.tif'
if not os.path.exists(tifclip) or overwrite:
call([
"gdalwarp", '-ot', 'Float32', "-te",
str(xmin),
str(ymin),
str(xmax),
def fixelevs(source, output, netf, recf, proj, method):
print(" running fixelevs.py...")
# Reading XXX_net.tif file
geo = gdalutils.get_geo(netf)
# Reading XXX_rec.csv file
rec = pd.read_csv(recf)
# Reading XXX_bnk.shp file
bnk_gdf = gpd.read_file(source)
# Initiate output shapefile
w = shapefile.Writer(shapefile.POINT)
w.field('x')
w.field('y')
w.field('elevadj')
# Retrieving bank elevations from XXX_bnk.shp file
# Values are stored in rec['bnk']
rec['bnk'] = bnk_gdf['elev'].astype(float)
# Adjusting bank values, resulting values
# are stored in rec['bnk_adj']
# coordinates are grouped by REACH number
rec['bnk_adj'] = 0
recgrp = rec.groupby('reach')
for reach, df in recgrp:
ids = df.index
dem = df['bnk']
# calc bank elevation
if method == 'yamazaki':
adjusted_dem = bank4flood(dem)
elif method == 'lowless':
adjusted_dem = lowless(dem)
else:
sys.exit('Method not recognised')
rec['bnk_adj'][ids] = adjusted_dem
# Writing .shp resulting file
for i in rec.index:
w.point(rec['lon'][i], rec['lat'][i])
w.record(rec['lon'][i], rec['lat'][i], rec['bnk_adj'][i])
w.save("%s.shp" % output)
# write .prj file
prj = open("%s.prj" % output, "w")
srs = osr.SpatialReference()
srs.ImportFromProj4(proj)
prj.write(srs.ExportToWkt())
prj.close()
nodata = -9999
fmt = "GTiff"
name1 = output + ".shp"
name2 = output + ".tif"
subprocess.call([
"gdal_rasterize", "-a_nodata",
str(nodata), "-of", fmt, "-co", "COMPRESS=DEFLATE", "-tr",
str(geo[6]),
str(geo[7]), "-a", "elevadj", "-a_srs", proj, "-te",
str(geo[0]),
str(geo[1]),
str(geo[2]),
str(geo[3]), name1, name2
])
dem_downsample = os.path.join(
indir, 'dem_downsample_' + str(nwindow * 3) + 's.tif')
acc_downsample = os.path.join(
indir, 'acc_downsample_' + str(nwindow * 3) + 's.tif')
ord_downsample = os.path.join(
indir, 'ord_downsample_' + str(nwindow * 3) + 's.tif')
f_dir = os.path.join(indir,
'dir_d8_downsample_' + str(nwindow * 3) + 's.tif')
f_net = os.path.join(indir, 'net_downsample_' + str(nwindow * 3) + 's.tif')
f_outlets = os.path.join(
indir, 'outlets_downsample_' + str(nwindow * 3) + 's.tif')
#########################################################################################
# Get geometry information from acctif, assume all arrays have the same geometry
#
geo = gdalutils.get_geo(acctif)
print(geo)
# modify number of cells: divide by nwindow and round up (block_reduce pads the arrays)
geo[4] = int(np.ceil(geo[4] / nwindow))
geo[5] = int(np.ceil(geo[5] / nwindow))
# modify resolution: multiply by nwindow
geo[6] = geo[6] * nwindow
geo[7] = geo[7] * nwindow
#########################################################################################
# Downsample dem array
if not os.path.exists(dem_downsample):
data = gdalutils.get_data(demtif)
print('inshape', data.shape)
downsample_dem = block_reduce(data,
block_size=(nwindow, nwindow),
def calc_banks(banktif, bedtif, fname_disch, fname_stage, reccsv, return_per, layer, outfolder):
# Loading stage and discharge files
# Try statement added since some discharge and stage files are empty, exit program
try:
stage = lfp.read_stage(fname_stage)
df_locs = lfp.read_stage_locs(fname_stage)
df_locs.index = range(len(stage.columns))
discharge = lfp.read_discharge(fname_disch)
stage.columns = range(len(discharge.columns))
discharge.columns = range(len(stage.columns))
except ValueError:
sys.exit('ERROR: Probably stage or discharge file is empty')
# Loading Return Perid database (eg. FLOPROS)
gdf_defenses = gpd.read_file(return_per)
# Getting protection level from Return Period dataset at every cell
# River points have been buffered to allow disaggrement between geolocations
# By buffering some points get more than one value, maximum flood protection is selected
mygeom = [Point(x, y) for x, y in zip(df_locs['x'], df_locs['y'])]
gdf_locs = gpd.GeoDataFrame(crs={'init': 'epsg:4326'}, geometry=mygeom)
gdf_locs_buf = gpd.GeoDataFrame(
crs={'init': 'epsg:4326'}, geometry=gdf_locs.buffer(0.1))
gdf_locs_ret = gpd.sjoin(gdf_locs_buf, gdf_defenses, op='intersects')
gdf_locs_ret['index'] = gdf_locs_ret.index
gdf_locs_ret = gdf_locs_ret.sort_values(
layer, ascending=False).drop_duplicates('index').sort_values('index')
# Estimating error in discharge fitting
dis_err = []
for i in range(discharge.shape[1]):
try:
dis_err.append(get_discharge_error(discharge[i]))
except (KeyError,np.core._internal.AxisError):
dis_err.append(0)
# Estimating a defenses-related discharge
dis_df = []
for i in range(discharge.shape[1]):
ret_pe = gdf_locs_ret['MerL_Riv'][i]
try:
dis_df.append(get_discharge_returnperiod(discharge[i], ret_pe))
except (KeyError,np.core._internal.AxisError):
dis_df.append(np.nan)
# Estimating error in stage fitting
stg_err = []
for i in range(discharge.shape[1]):
try:
stg_err.append(get_stage_error(discharge[i], stage[i]))
except (RuntimeError, TypeError):
stg_err.append(0)
# Estimating a defenses-related stage
stg_df = []
for i in range(discharge.shape[1]):
try:
stg_df.append(get_stage_discharge(
discharge[i], stage[i], dis_df[i]))
except (RuntimeError, TypeError):
stg_df.append(np.nan)
# Preparing a summary with variables retrived
df_locs['dis_df'] = dis_df
df_locs['stg_df'] = stg_df
df_locs['dis_err'] = dis_err
df_locs['stg_err'] = stg_err
# Read REC file
rec = pd.read_csv(reccsv)
# Convert dataframe to geodataframe, join with rec
gdf_sum = gpd.GeoDataFrame(df_locs, crs={'init': 'epsg:4326'}, geometry=[
Point(x, y) for x, y in zip(df_locs['x'], df_locs['y'])])
gdf_rec = gpd.GeoDataFrame(rec, crs={'init': 'epsg:4326'}, geometry=[
Point(x, y) for x, y in zip(rec['lon'], rec['lat'])])
gdf_rec_buf = gpd.GeoDataFrame(
rec, crs={'init': 'epsg:4326'}, geometry=gdf_rec.buffer(0.001))
gdf_sum_rec = gpd.sjoin(gdf_sum, gdf_rec_buf,
how='inner', op='intersects')
gdf_sum_rec.sort_values('index_right', inplace=True)
# Save errors in a GeoJSON file
try:
gdf_sum_rec.to_file(outfolder + 'bnk_err.geojson', driver='GeoJSON')
except:
os.remove(outfolder + 'bnk_err.geojson')
gdf_sum_rec.to_file(outfolder + 'bnk_err.geojson', driver='GeoJSON')
# Score should greater than 0.85 for both Discharge and Stage to be accepted, otherwise NaN
gdf_err = gdf_sum_rec['stg_df'].where(
(gdf_sum_rec['dis_err'] > 0.85) & (gdf_sum_rec['stg_err'] > 0.85))
# Fill with NaN stg_df not filling that condition
gdf_sum_rec['stg_df'] = gdf_err
# NaNs are filled repating last/first number per link
gdf_sum_rec_fillna = gdf_sum_rec.groupby('link').fillna(
method='bfill').fillna(method='ffill')
gdf_sum_rec_fillna['link'] = gdf_sum_rec['link']
# Read data and geo for bedtif
bed = gu.get_data(bedtif)
geo = gu.get_geo(bedtif)
# Convert dataframes to arrays
df_locs_stgdf = gdf_sum_rec_fillna[['x', 'y', 'stg_df']]
df_locs_stgdf.columns = ['x', 'y', 'z']
arr_stgdf = gu.pandas_to_array(df_locs_stgdf, geo, 0)
# Sum bankfull stage and defenses-related stage to bed
arr_bnkdf = (bed + arr_stgdf)
# Write burned banks in ASC and TIF files
gu.write_raster(arr_bnkdf, banktif, geo, 'Float64', 0)
