def main_program(shuffi, isCrossValidation):
list_all_test = shuffi[0]
list_all_station = shuffi[1]
global D
#time.sleep(100)
print len(list_all_station)
trend_curve = generateGlobalMatrix(date_from, date_to, list_all_station)
D = generateDistanceMatrix(list_all_station)
area_id = 0
genModelForArea(area_id, list_all_station, list_data, G, D)
reGenerateSemiMatrix(list_all_station)
print 'GLOBAL DATA'
print global_data
derivation = np.zeros([row,col])
result = np.zeros([row,col])
for i in xrange(0,row):
#print i
for j in xrange(0, col):
#print i, j
point = (minLat + i*shift,minLon + j*shift)
f.write(str(i)+' '+ str(j)+ ':'+ str(point) +'\n')
#if isInVietnam(point):
result[i][j], derivation[i][j] = krigeOne(point, neighbor, list_all_station, list_data, G, D)
trendValue = getTrendValue(trend_curve, point[0], point[1])
#print trendValue
result[i][j] = result[i][j] + trendValue
#exportGeotiff('filename', raster, row, col, shift, minLon, minLat)
if isCrossValidation == False:
exportGeotiff('t08_GEOTIFF_simple_kriging_one_model', result, row, col, shift, minLon, minLat )
saveAsPng(result)
saveAsPng(derivation)
if isCrossValidation:
return calculateSE(result, list_all_test, minLat, minLon, shift, date_to)
def main_program(shuffi, isCrossValidation):
f_se = open("Moiseture_f_se.txt", "a")
f_se.write("Moiseture_se statistic\n")
f_da = open("Moiseture_f_da.txt", "a")
f_da.write("Moiseture_da statistic\n")
list_all_test = shuffi[0]
list_all_station = shuffi[1]
global D
# time.sleep(100)
print len(list_all_station)
trend_curve = generateGlobalMatrix(date_from, date_to, list_all_station)
D = generateDistanceMatrix(list_all_station)
area_id = 0
genModelForArea(area_id, list_all_station, list_data, G, D)
reGenerateSemiMatrix(list_all_station)
print "GLOBAL DATA"
print global_data
derivation = np.zeros([row, col])
result = np.zeros([row, col])
result_kri_backup = np.zeros([row, col])
for i in xrange(0, row):
# print i
for j in xrange(0, col):
# print i, j
point = (minLat + i * shift, minLon + j * shift)
f.write(str(i) + " " + str(j) + ":" + str(point) + "\n")
result[i][j], derivation[i][j] = krigeOne(point, neighbor, list_all_station, list_data, G, D)
trendValue = getTrendValue(trend_curve, point[0], point[1])
# print trendValue
result[i][j] = result[i][j] + trendValue
result_kri_backup[i][j] = result[i][j]
# DATA ASSIMILATION: assimilate
# if sate_lite_value[i][j] != 0:
# result[i][j] = assimilate(result[i][j], derivation[i][j], sate_lite_value[i][j], sate_lite_deri[i][j], f_da)
# Assimilate with Utility.assimilate
if isCrossValidation == False:
exportGeotiff("Moiseture__GEOTIFF_universal_kriging", result, row, col, shift, minLon, minLat)
exportGeotiff("Moiseture__GEOTIFF_universal_kriging_derivation", derivation, row, col, shift, minLon, minLat)
saveAsPng(result)
saveAsPng(derivation)
f_da.close()
f_se.close()
if isCrossValidation:
return calculateSE(
result, result_kri_backup, list_all_test, minLat, minLon, shift, date_to, f, sate_lite_value, f_se
)
def krigeOne(point, neighbor, list_all_station, list_data, G, D):
#print 'KRIGE ONE'
list_station = list_all_station
#Cal distances from current point to all others
item0 = (0, point[0], point[1])
meteo_size = len(list_station)
list_d = []
#cal all distance from point to others stations
for i in xrange(0,meteo_size):
#Cal distance one-one
item1 = list_station[i]
d = haversine(item0[2], item0[1] ,item1[2], item1[1])
newitem = (item1[0], d, item1[3], item1[1], item1[2])#id, distance, temp, lat, lon
list_d.append(newitem)
#Sort by distance, get n first nearest neighbors
sorted_list = sorted(list_d, key=itemgetter(1))
list_neighbor = sorted_list[:neighbor]
#print list_neighbor
'''
#Get areaid for each station
list_neighbor_with_areaid = []
for item in list_neighbor:
item1 = list_station[item[0]]
new_item = (item[0], item[1], getAreaIdFromPoint([item1[1], item1[2]]))
list_neighbor_with_areaid.append(new_item)
'''
K1 = np.zeros([neighbor, neighbor])
K = np.zeros([neighbor, neighbor])
k = np.zeros([neighbor, 1])
T = []# list of temp of neighbor station at interpolate time
string = ''
i = 0
for item in list_neighbor:
f.write('getting T\n')
#Nhiet do 1 thang cua tram item
#print len(list_data)
t = list_data[item[0]-1]
trendValue = getTrendValue(trend_curve, item[3], item[4])
#print trend_curve, trendValue
f.write(str(t) + '\n')
t_size = len(t)
T.append(t[t_size - 1] - trendValue)
j = 0
string = string + str(item[0])
#print item
#Area key = 0, because of we have only one model for all
a0, c0 = global_data[0][0], global_data[0][1]
h = item[1]#distance
k[i][0] = semiToCov( h, a0, c0)
for sub_item in list_neighbor:
string = string+' ' + str(sub_item[0])
#print i, j
K1[i][j] = G_COV[item[0]-1][sub_item[0]-1]
K [j][i] = K1[i][j]#Inverse of K1
j = j + 1
string = string + '\n'
i = i + 1
f.write('T: ' + str(T) + '\n')
f.write('neighbors list: \n' + string + '\n')
#print string
#print k
#print K
f.write('K: \n' + str(K) +' \n')
f.write('k: \n' + str(k) +' \n')
weight = np.dot(K,k)
f.write('ori-weight: \n' + str(weight) +' \n')
if sum(weight) == 0:
f.write('NOT RELATED\n')
weight[:] = 1 #assign the same weight for all elements
weight = weight/sum(weight)
krige = np.dot(T, weight)
f.write('nor-weight: \n' + str(weight) +' \n')
f.write('T: \n' + str(T)+' \n')
f.write('krige: ' + str(krige)+' \n')
#print krige
#print krige[0],' ',
#print k
weight = inverseMatrix(weight)
derivation = np.dot(weight, k)
f.write('derivation: ' + str(derivation[0][0])+' \n')
#time.sleep(2)
return krige[0], derivation[0][0]
'''