CDC Challenge Weather Data Code

narr_simple.py

import rasterstats as rs
import geopandas as gpd
import os
import pandas as pd
import glob

shape_frame = gpd.read_file('./tl_2017_us_county/tl_2017_us_county.shp')
shape = shape_frame.to_crs('+proj=longlat +datum=WGS84 +no_defs')
shape['GEOID'] = shape['GEOID'].astype(str)

output_columns = ['mean']
var = 'apcp'

for year in range(2000, 2019):
    year_dir = './' + var + '/data_NEW_NEW/' + str(year) + '/'
    if not os.path.exists(year_dir):
        print('Making', year_dir)
        os.makedirs(year_dir)

    year_files = glob.glob('./' + var + '/masked_daily_geotiff/' + str(year) + '/*.geotiff')
    print('PROCESSING', len(year_files), 'FILES')
    year_files = sorted(year_files)

    for file in year_files:
        name = file[file.rfind('/')+1:file.rfind('.')]
        stats = rs.zonal_stats(shape, file, stats=output_columns, all_touched=True)
        frame = pd.DataFrame.from_dict(stats).set_index(shape_frame['GEOID'])
        frame.to_pickle(year_dir + name + '.pkl')
        print(year_dir + name + '.pkl')