def sample_pixel_area_raster(pxlist, rasterpath, dtype, maj):
pxvals = []
for i in range(len(pxlist)):
tmp = extr_TS.extr_raster_pixel_values(rasterpath, 1, pxlist[i][1], pxlist[i][0], dtype)
pxvals.append(tmp)
pxvals = np.array(pxvals, dtype=np.float32)
if maj == True: out_maj = np.argmax(np.bincount(np.array(pxvals,dtype=np.int)))
pxvals[pxvals == -9999] = np.nan
pxvals[pxvals == 32767] = np.nan
out_pxval = np.nanmean(pxvals)
if out_pxval == np.nan: out_pxval = -9999
if maj == True:
return((out_pxval, out_maj))
else:
return(out_pxval)
def sample_pixel_area_raster(pxlist, rasterpath, dtype, maj):
pxvals = []
for i in range(len(pxlist)):
tmp = extr_TS.extr_raster_pixel_values(rasterpath, 1, pxlist[i][1],
pxlist[i][0], dtype)
pxvals.append(tmp)
pxvals = np.array(pxvals, dtype=np.float32)
if maj == True:
out_maj = np.argmax(np.bincount(np.array(pxvals, dtype=np.int)))
pxvals[pxvals == -9999] = np.nan
pxvals[pxvals == 32767] = np.nan
out_pxval = np.nanmean(pxvals)
if out_pxval == np.nan: out_pxval = -9999
if maj == True:
return ((out_pxval, out_maj))
else:
return (out_pxval)
def sample_pixel_area_sig0(lon, lat, pxsize, landcover, asapath, outpath):
#this function calculates an average sig0 over a certain area by sampling it with 5 random points
pxlist = []
#pxlist.append([lon,lat])
random.seed()
goodlc = np.array([10,12,13,15,16,17,18,19,20,21,26,27,28,29])
meanSIG0 = []
SIG0list = []
SIG0Nlist = []
LIAlist = []
DAYlist = []
while len(pxlist) <= 10:
if len(pxlist) == 0:
rlon = lon
rlat = lat
else:
rlon = random.uniform(lon-(pxsize/3),lon+(pxsize/3))
rlat = random.uniform(lat-(pxsize/3),lat+(pxsize/3))
lc = extr_TS.extr_raster_pixel_values(landcover, 1, rlat, rlon, 1)
if np.any(goodlc == lc):
tmp = extr_TS.extr_SIG0_LIA_ts(asapath, 'ASAWS', 'SGRT21A01', 'resampled', 75, rlon, rlat)
tmpSIG0 = tmp[1]
tmpLIA=tmp[2]
#normalise SIG0 to a constant incidence angle of 30deg
mask = np.array((tmpLIA > 2000) & (tmpLIA < 6000) & (tmpSIG0 != -9999)).squeeze()
slope = stats.linregress(tmpLIA[mask], tmpSIG0[mask])[0]
# print(slope)
tmpSIG0norm = tmpSIG0 - (slope*(tmpLIA-3000))
tmpSIG0norm[np.logical_not(mask)] = -9999
SIG0list.append(tmp[1])
SIG0Nlist.append(tmpSIG0norm)
LIAlist.append(tmp[2])
DAYlist.append(tmp[0])
meanSIG0.append(10*math.log10(np.power(10,(np.array(tmpSIG0norm[mask])/10)).mean()))
pxlist.append([rlon,rlat])
#find minimum and maximum day
minDay = min([x for xs in DAYlist for x in xs])
maxDay = max([x for xs in DAYlist for x in xs])
ndays = int(maxDay) - int(minDay)
ndays = ndays + 1
# create arrays for SIG0 and LIA
DAYseq = range(int(minDay), int(maxDay)+1)
SIG0arr = np.zeros((ndays, len(pxlist)+1), dtype=np.float)
SIG0arr[:,1:] = -9999
SIG0Narr = np.zeros((ndays, len(pxlist)+1), dtype=np.float)
SIG0Narr[:,1:] = -9999
LIAarr = np.zeros((ndays, len(pxlist)+1), dtype=np.float)
LIAarr[:,1:] = -9999
for i in range(ndays):
SIG0arr[i,0] = DAYseq[i]
SIG0Narr[i,0] = DAYseq[i]
LIAarr[i,0] = DAYseq[i]
# fill SIG0 and LIA into array
for i in range(len(pxlist)):
for j in range(len(SIG0list[i])):
SIG0arr[SIG0arr[:,0] == DAYlist[i][j], i+1] = SIG0list[i][j]
SIG0Narr[SIG0Narr[:,0] == DAYlist[i][j], i+1] = SIG0Nlist[i][j]
LIAarr[LIAarr[:,0] == DAYlist[i][j], i+1] = LIAlist[i][j]
# create array with averages
outSIG0arr = np.zeros((ndays,2))
outSIG0Narr = np.zeros((ndays,2))
outLIAarr = np.zeros((ndays,2))
for i in range(ndays):
outSIG0arr[i,0] = DAYseq[i]
outSIG0Narr[i,0] = DAYseq[i]
outLIAarr[i,0] = DAYseq[i]
SIG0arr[SIG0arr == -9999] = np.nan
SIG0Narr[SIG0Narr == -9999] = np.nan
LIAarr[LIAarr == -9999] = np.nan
outSIG0arr[:,1] = 10*np.log10(np.nanmean(np.power(10,SIG0arr[:,1:]/10), axis=1))
outSIG0Narr[:,1] = 10*np.log10(np.nanmean(np.power(10,SIG0Narr[:,1:]/10), axis=1))
outLIAarr[:,1] = np.nanmean(LIAarr[:,1:], axis=1)
outSIG0arr[np.isfinite(outSIG0arr) == False] = -9999
outSIG0Narr[np.isfinite(outSIG0Narr) == False] = -9999
outLIAarr[np.isfinite(outLIAarr) == False] = -9999
return (outSIG0arr.astype(np.int), outSIG0Narr.astype(np.int), outLIAarr.astype(np.int), pxlist, np.nanmean(meanSIG0))
def sample_pixel_area_sig0(lon, lat, pxsize, landcover, asapath, outpath):
#this function calculates an average sig0 over a certain area by sampling it with 5 random points
pxlist = []
#pxlist.append([lon,lat])
random.seed()
goodlc = np.array([10, 12, 13, 15, 16, 17, 18, 19, 20, 21, 26, 27, 28, 29])
meanSIG0 = []
SIG0list = []
SIG0Nlist = []
LIAlist = []
DAYlist = []
while len(pxlist) <= 10:
if len(pxlist) == 0:
rlon = lon
rlat = lat
else:
rlon = random.uniform(lon - (pxsize / 3), lon + (pxsize / 3))
rlat = random.uniform(lat - (pxsize / 3), lat + (pxsize / 3))
lc = extr_TS.extr_raster_pixel_values(landcover, 1, rlat, rlon, 1)
if np.any(goodlc == lc):
tmp = extr_TS.extr_SIG0_LIA_ts(asapath, 'ASAWS', 'SGRT21A01',
'resampled', 75, rlon, rlat)
tmpSIG0 = tmp[1]
tmpLIA = tmp[2]
#normalise SIG0 to a constant incidence angle of 30deg
mask = np.array((tmpLIA > 2000) & (tmpLIA < 6000)
& (tmpSIG0 != -9999)).squeeze()
slope = stats.linregress(tmpLIA[mask], tmpSIG0[mask])[0]
# print(slope)
tmpSIG0norm = tmpSIG0 - (slope * (tmpLIA - 3000))
tmpSIG0norm[np.logical_not(mask)] = -9999
SIG0list.append(tmp[1])
SIG0Nlist.append(tmpSIG0norm)
LIAlist.append(tmp[2])
DAYlist.append(tmp[0])
meanSIG0.append(10 * math.log10(
np.power(10, (np.array(tmpSIG0norm[mask]) / 10)).mean()))
pxlist.append([rlon, rlat])
#find minimum and maximum day
minDay = min([x for xs in DAYlist for x in xs])
maxDay = max([x for xs in DAYlist for x in xs])
ndays = int(maxDay) - int(minDay)
ndays = ndays + 1
# create arrays for SIG0 and LIA
DAYseq = range(int(minDay), int(maxDay) + 1)
SIG0arr = np.zeros((ndays, len(pxlist) + 1), dtype=np.float)
SIG0arr[:, 1:] = -9999
SIG0Narr = np.zeros((ndays, len(pxlist) + 1), dtype=np.float)
SIG0Narr[:, 1:] = -9999
LIAarr = np.zeros((ndays, len(pxlist) + 1), dtype=np.float)
LIAarr[:, 1:] = -9999
for i in range(ndays):
SIG0arr[i, 0] = DAYseq[i]
SIG0Narr[i, 0] = DAYseq[i]
LIAarr[i, 0] = DAYseq[i]
# fill SIG0 and LIA into array
for i in range(len(pxlist)):
for j in range(len(SIG0list[i])):
SIG0arr[SIG0arr[:, 0] == DAYlist[i][j], i + 1] = SIG0list[i][j]
SIG0Narr[SIG0Narr[:, 0] == DAYlist[i][j], i + 1] = SIG0Nlist[i][j]
LIAarr[LIAarr[:, 0] == DAYlist[i][j], i + 1] = LIAlist[i][j]
# create array with averages
outSIG0arr = np.zeros((ndays, 2))
outSIG0Narr = np.zeros((ndays, 2))
outLIAarr = np.zeros((ndays, 2))
for i in range(ndays):
outSIG0arr[i, 0] = DAYseq[i]
outSIG0Narr[i, 0] = DAYseq[i]
outLIAarr[i, 0] = DAYseq[i]
SIG0arr[SIG0arr == -9999] = np.nan
SIG0Narr[SIG0Narr == -9999] = np.nan
LIAarr[LIAarr == -9999] = np.nan
outSIG0arr[:, 1] = 10 * np.log10(
np.nanmean(np.power(10, SIG0arr[:, 1:] / 10), axis=1))
outSIG0Narr[:, 1] = 10 * np.log10(
np.nanmean(np.power(10, SIG0Narr[:, 1:] / 10), axis=1))
outLIAarr[:, 1] = np.nanmean(LIAarr[:, 1:], axis=1)
outSIG0arr[np.isfinite(outSIG0arr) == False] = -9999
outSIG0Narr[np.isfinite(outSIG0Narr) == False] = -9999
outLIAarr[np.isfinite(outLIAarr) == False] = -9999
return (outSIG0arr.astype(np.int), outSIG0Narr.astype(np.int),
outLIAarr.astype(np.int), pxlist, np.nanmean(meanSIG0))