def multi_plot_regression(date_start, date_end, frequency='daily'):
datetime_start = datetime.strptime(date_start, "%Y%m%d")
datetime_end = datetime.strptime(date_end, "%Y%m%d")
stats_data = defaultdict(list)
while datetime_start <= datetime_end:
date_str = datetime_start.strftime("%Y%m%d")
datas = plot_verification_picture(date_str,
date_end=date_end,
frequency=frequency)
if datas:
for key, value in datas.items():
stats_data[key].append(value)
if frequency == 'daily':
datetime_start = datetime_start + relativedelta(days=1)
elif frequency == 'monthly':
datetime_start = datetime_start + relativedelta(months=1)
elif frequency == 'seasonly':
datetime_start = datetime_start + relativedelta(months=3)
elif frequency == 'yearly':
datetime_start = datetime_start + relativedelta(years=1)
else:
break
stats_data = pd.DataFrame(stats_data)
out_dir = os.path.join(AOD_PICTURE_DIR, 'STATS', MATCH)
make_sure_path_exists(out_dir)
out_file = os.path.join(
out_dir, '{}_{}_{}_{}.csv'.format(AREA, frequency, date_start,
date_end))
stats_data.to_csv(out_file, index=False)
print('>>> {}'.format(out_file))
def match_file():
# FY3D_MERSI_ORBT_L2_AOD_MLT_NUL_20190102_0500_1000M_MS.HDF
fy3d_file_dict = defaultdict(list)
for root, dirs, files in os.walk(AOD_FY3D_1KM_DIR):
for name in files:
if name[-3:].lower() != 'hdf':
continue
ymd, hm = name.split('_')[7:9]
date_str = ymd
fy3d_file_dict[date_str].append(os.path.join(root, name))
# SATE_FY4A_AGRI_MULT_4000M_DISK_LDA_GLL_HYT_20190101000000_VPRJ5_L2.NC
fy4a_file_dict = defaultdict(list)
for root, dirs, files in os.walk(AOD_FY4A_4KM_DIR):
for name in files:
if name[-2:].lower() != 'nc':
continue
date_str = name.split('_')[9][0:8]
fy4a_file_dict[date_str].append(os.path.join(root, name))
for fy3d_date_str in fy3d_file_dict.keys():
if fy3d_date_str in fy4a_file_dict:
fy3d_files = fy3d_file_dict[fy3d_date_str]
fy4a_files = fy4a_file_dict[fy3d_date_str]
for fy3d_file in fy3d_files:
for fy4a_file in fy4a_files:
ymdhm_fy3d = os.path.basename(fy3d_file).split('_')[7:9]
datetime_fy3d = datetime.strptime(''.join(ymdhm_fy3d),
'%Y%m%d%H%M')
ymdhms_fy4a = os.path.basename(fy4a_file).split(
'_')[9][0:12]
datetime_fy4a = datetime.strptime(ymdhms_fy4a,
'%Y%m%d%H%M')
datetime_start = datetime_fy3d - relativedelta(minutes=5)
datetime_end = datetime_fy3d + relativedelta(minutes=5)
if not (datetime_start <= datetime_fy4a <= datetime_end):
continue
match_dir = os.path.join(AOD_MATCH_DIR,
"FY3D_1KM_FY4A_4KM")
make_sure_path_exists(match_dir)
out_file = get_out_file(fy3d_file, fy4a_file, match_dir)
if os.path.isfile(out_file):
print('already exist {}'.format(out_file))
continue
geo_fy3d_name = 'FY3D_MERSI_GBAL_L1_{}_{}_GEO1K_MS.HDF'.format(
ymdhm_fy3d[0], ymdhm_fy3d[1])
fy3d_geo_file = os.path.join(GEO_FY3D_1KM_DIR,
geo_fy3d_name)
match_fy3d_1km_fy4a_4km(fy3d_file,
fy3d_geo_file,
fy4a_file,
out_file,
longitude_range=LONGITUDE_RANGE,
latitude_range=LATITUDE_RANGE)
def match_file():
print("<<< === {}".format(AOD_FY3D_1KM_DIR))
print("<<< === {}".format(AOD_MODIS_3KM_DIR))
fy3d_file_dict = defaultdict(list)
for root, dirs, files in os.walk(AOD_FY3D_1KM_DIR):
for name in files:
if name[-3:].lower() != 'hdf':
continue
date_str = name.split('_')[7]
fy3d_file_dict[date_str].append(os.path.join(root, name))
modis_file_dict = defaultdict(list)
for root, dirs, files in os.walk(AOD_MODIS_3KM_DIR):
for name in files:
if name[-3:].lower() != 'hdf':
continue
date_j_str = name.split('.')[1][1:]
date_str = datetime.strptime(date_j_str, "%Y%j").strftime("%Y%m%d")
modis_file_dict[date_str].append(os.path.join(root, name))
for fy3d_date_str in fy3d_file_dict.keys():
if fy3d_date_str in modis_file_dict:
fy3d_files = fy3d_file_dict[fy3d_date_str]
modis_files = modis_file_dict[fy3d_date_str]
for fy3d_file in fy3d_files:
for modis_file in modis_files:
ymdhm_fy3d = os.path.basename(fy3d_file).split('_')[7:9]
datetime_fy3d = datetime.strptime(''.join(ymdhm_fy3d),
'%Y%m%d%H%M')
ymdhm_modis = os.path.basename(modis_file).split('.')[1:3]
date_j_str = ymdhm_modis[0][1:] + ymdhm_modis[1]
datetime_modis = datetime.strptime(date_j_str, "%Y%j%H%M")
datetime_start = datetime_fy3d - relativedelta(minutes=60)
datetime_end = datetime_fy3d + relativedelta(minutes=60)
if not (datetime_start <= datetime_modis <= datetime_end):
continue
match_dir = os.path.join(AOD_MATCH_DIR,
"FY3D_1KM_MODIS_3KM")
make_sure_path_exists(match_dir)
out_file = get_out_file(fy3d_file, modis_file, match_dir)
if os.path.isfile(out_file):
print('already exist {}'.format(out_file))
continue
ymd_hm = os.path.basename(fy3d_file).split('_')[7:9]
geo_fy3d_name = 'FY3D_MERSI_GBAL_L1_{}_{}_GEO1K_MS.HDF'.format(
ymd_hm[0], ymd_hm[1])
fy3d_geo_file = os.path.join(GEO_FY3D_1KM_DIR,
geo_fy3d_name)
match_fy3d_1km_modis_3km(fy3d_file,
fy3d_geo_file,
modis_file,
out_file,
longitude_range=LONGITUDE_RANGE,
latitude_range=LATITUDE_RANGE)
def match_file():
fy3d_file_dict = defaultdict(list)
for root, dirs, files in os.walk(AOD_FY3D_5KM_DIR):
for name in files:
if name[-3:].lower() != 'hdf':
continue
date_str = name.split('_')[7]
fy3d_file_dict[date_str].append(os.path.join(root, name))
modis_file_dict = defaultdict(list)
for root, dirs, files in os.walk(AOD_MODIS_10KM_DIR):
for name in files:
if name[-3:].lower() != 'hdf':
continue
date_j_str = name.split('.')[1][1:]
date_str = datetime.strptime(date_j_str, "%Y%j").strftime("%Y%m%d")
modis_file_dict[date_str].append(os.path.join(root, name))
for fy3d_date_str in fy3d_file_dict.keys():
if fy3d_date_str in modis_file_dict:
fy3d_files = fy3d_file_dict[fy3d_date_str]
modis_files = modis_file_dict[fy3d_date_str]
for fy3d_file in fy3d_files:
for modis_file in modis_files:
match_dir = os.path.join(AOD_MATCH_DIR,
"FY3D_5KM_MODIS_10KM")
make_sure_path_exists(match_dir)
out_file = get_out_file(fy3d_file, modis_file, match_dir)
if os.path.isfile(out_file):
print('already exist {}'.format(out_file))
continue
match_fy3d_5km_modis_10km(fy3d_file,
modis_file,
out_file,
longitude_range=LONGITUDE_RANGE,
latitude_range=LATITUDE_RANGE)
def combine_aqua_daily(datetime_start=None,
datetime_end=None,
data_dir=None,
out_dir=None):
# FY3D_MERSI_ORBT_L2_AOD_MLT_NUL_20190916_0910_1000M_MS.HDF
# FY3D_MERSI_GBAL_L2_AOD_MLT_GLL_20190705_POAD_5000M_MS.HDF
print('data_dir :{}'.format(data_dir))
file_dict = defaultdict(list)
for root, dirs, files in os.walk(data_dir):
for name in files:
if name[-3:].lower() != 'hdf':
continue
date_str_file = name[10:17]
date_ = datetime.strptime(date_str_file, "%Y%j")
if datetime_start is not None and datetime_end is not None:
if not (datetime_start <= date_ <= datetime_end):
continue
date_str = date_.strftime("%Y%m%d")
file_dict[date_str].append(os.path.join(root, name))
if not file_dict:
print('没有找到数据')
return
# 创建投影查找表
print('创建投影查找表')
res_degree = 0.01 # 分辨率，1km
projstr = "+proj=longlat +ellps=WGS84 +datum=WGS84 +no_defs"
proj = ProjCore(projstr,
res_degree,
unit="deg",
pt_tl=(69.995, 54.995),
pt_br=(139.995, 9.995)) # 角点也要放在格点中心位置
for d_, files in file_dict.items():
aod_sum = np.zeros((proj.row, proj.col), dtype=np.float)
aod_count = np.zeros_like(aod_sum, dtype=np.float)
filename_out = 'AQUA_MODIS_GBAL_L2_AOD_MLT_GLL_{}_POAD_1000M_MS.HDF'.format(
d_)
file_out = os.path.join(out_dir, filename_out)
if os.path.isfile(file_out):
print('already exist {}'.format(file_out))
continue
print('<<< {}'.format(d_))
for file_ in files:
print('<<< {}'.format(file_))
file_loader = AodModis(file_)
aod = file_loader.get_aod()
lons, lats = file_loader.get_lon_lat()
print(np.nanmin(aod), np.nanmax(aod), np.nanmean(aod))
if aod is None:
print('aod 为 None: {}'.format(file_))
continue
# 投影
print('投影')
ii, jj = proj.lonslats2ij(lons, lats)
valid = np.logical_and.reduce((
ii >= 0,
ii < proj.row,
jj >= 0,
jj < proj.col,
aod > 0,
aod < 1.5,
))
if valid.sum() == 0:
print('valid.size == 0, continue')
continue
print('valid.sum() == {}'.format(valid.sum()))
ii = ii[valid]
jj = jj[valid]
aod = aod[valid]
aod_sum[ii, jj] += aod
aod_count[ii, jj] += 1
print(np.nanmin(aod), np.nanmax(aod), np.nanmean(aod))
print(np.nanmin(aod_sum), np.nanmax(aod_sum), np.nanmean(aod_sum))
if aod_sum is not None and aod_count is not None:
aod_mean = get_mean(aod_sum, aod_count)
else:
continue
# 新的网格的经纬度
lons_grid, lats_grid = proj.grid_lonslats()
make_sure_path_exists(out_dir)
print((aod_mean != -999).sum())
data_write = {
'Optical_Depth_Land_And_Ocean': aod_mean,
'Longitude': lons_grid,
'Latitude': lats_grid
}
write_hdf5_and_compress(data_write, file_out)
def combine(frequency='Monthly',
datetime_start=None,
datetime_end=None,
data_dir=None,
out_dir=None,
res_type='1KM',
data_loader=None,
satellite_sensor=None):
print('frequency === {}'.format(frequency))
print('datetime_start === {}'.format(datetime_start))
print('datetime_end === {}'.format(datetime_end))
print('data_dir === {}'.format(data_dir))
print('out_dir === {}'.format(out_dir))
print('res_type === {}'.format(res_type))
print('satellite_sensor === {}'.format(satellite_sensor))
if data_loader is None:
data_loader = AodCombine
file_dict = defaultdict(list)
for root, dirs, files in os.walk(data_dir):
for name in files:
if name[-3:].lower() != 'hdf':
continue
in_file = os.path.join(root, name)
date_ = data_loader(in_file).dt
if datetime_start is not None and datetime_end is not None:
if not (datetime_start <= date_ <= datetime_end):
continue
if frequency == 'Monthly':
date_str = get_month(date_)
elif frequency == 'Seasonly':
date_str = get_season(date_)
elif frequency == 'Yearly':
date_str = get_year(date_)
else:
raise ValueError(frequency)
file_dict[date_str].append(in_file)
res_dict = {
'1KM': '1000M',
'5KM': '5000M',
}
for d_, files in file_dict.items():
aod_sum = None
aod_count = None
lons = None
lats = None
res = res_dict[res_type]
filename_out = '{}_GBAL_L2_AOD_MLT_GLL_{}_POAD_{}_MS.HDF'.format(
satellite_sensor, d_, res)
file_out = os.path.join(out_dir, filename_out)
if os.path.isfile(file_out):
print('already exist {}'.format(file_out))
continue
for file_ in files:
print('<<< {}'.format(file_))
loader = data_loader(file_)
aod = loader.get_aod() # 无效值赋值为-999
if aod is None:
print('aod 为 None: {}'.format(file_))
continue
valid = np.logical_and(aod > 0, aod < 1.5)
aod[~valid] = -999
# print(np.nanmin(aod), np.nanmax(aod), np.nanmean(aod))
aod_sum, aod_count = get_sum_count(aod_sum, aod_count, aod)
if lons is None and lats is None:
lons, lats = loader.get_lon_lat()
if aod_sum is not None and aod_count is not None:
aod_mean = get_mean(aod_sum, aod_count)
else:
continue
make_sure_path_exists(out_dir)
print((aod_mean != -999).sum())
data_write = {
'Optical_Depth_Land_And_Ocean': aod_mean,
'Longitude': lons,
'Latitude': lats
}
write_hdf5_and_compress(data_write, file_out)
def main(data_type, in_path, out_path, geo_path):
assert data_type in {
'FY3D_MERSI_1KM', 'FY3D_MERSI_5KM', 'AQUA_MODIS_3KM', 'AQUA_MODIS_10KM'
}, "{} 输入错误".format(data_type)
if os.path.isfile(in_path):
filelist = [in_path]
elif os.path.isdir(in_path):
filelist = list()
for root, dirs, files in os.walk(in_path):
for name in files:
if name[-3:].lower() != 'hdf':
continue
filelist.append(os.path.join(root, name))
else:
raise ValueError('{} 不是文件夹也不是有效文件'.format(in_path))
assert len(filelist) > 0, "输入文件的数量小于1"
make_sure_path_exists(out_path)
if data_type in {'FY3D_MERSI_1KM'}:
data_loader = AodFy3d1km
elif data_type in {'FY3D_MERSI_5KM'}:
data_loader = AodFy3d5km
elif data_type in {'AQUA_MODIS_3KM', 'AQUA_MODIS_10KM'}:
data_loader = AodModis
else:
raise ValueError(data_type)
for in_file in filelist:
if geo_path is not None:
if os.path.isfile(geo_path):
loader = data_loader(in_file, geo_file=geo_path)
else:
loader = data_loader(in_file, geo_path=geo_path)
else:
loader = data_loader(in_file)
dt = loader.dt
data = loader.get_aod() # 无效值为-999
lons, lats = loader.get_lon_lat()
data, lons, lats = proj_china(data, lons, lats, 0, 1.5)
if data_type in {"FY3D_MERSI_5KM"}:
fill_points_2d(data, -999)
fill_points_2d(data, -999)
for area_type in ["China", "YRD"]:
title = get_title(data_type, dt, area_type)
filename = loader.filename + '.{}.png'.format(area_type)
out_file = os.path.join(out_path, filename)
vmin = 0
vmax = 1.5
ticks = np.arange(0, 1.51, 0.3)
if area_type == 'China':
nanhai = True
else:
nanhai = False
mksize = 5
areas = get_areas(area_type)
mask = get_province_mask(areas)
valid = np.logical_and.reduce((data > vmin, data < vmax, mask))
data_mask = data[valid]
lons_mask = lons[valid]
lats_mask = lats[valid]
count = len(data_mask)
print('count == {}'.format(count))
longitude_range, latitude_range = get_area_range(area_type)
box = [
latitude_range[1], latitude_range[0], longitude_range[0],
longitude_range[1]
]
plot_map_picture(data_mask,
lons_mask,
lats_mask,
title=title,
vmin=vmin,
vmax=vmax,
areas=areas,
box=box,
ticks=ticks,
file_out=out_file,
mksize=mksize,
nanhai=nanhai)
