Zonal stats with large time series

[hannahker]

Hi,

Following past tutorials on zonal stats, I'm trying to get the max CHIRPS precipitation values by admin regions within my area of interest (admin 2 boundaries of Malawi). I'm trying to process an ImageCollection with 20 years worth of daily data. Perhaps unsurprisingly, the following code returns User memory limit exceeded..

Is there a better way to try and calculate zonal stats based on an ImageCollection of this size?

import ee
import geemap
import geopandas as gpd

ee.Initialize()

CHIRPS_START = '2000-01-01'
CHIRPS_END = '2020-12-31'

mwi_adm2 = ee.FeatureCollection('users/hker/mwi_adm2')

chirps_col = ee.ImageCollection('UCSB-CHG/CHIRPS/DAILY').filter(ee.Filter.date(CHIRPS_START, CHIRPS_END))
chirps_col = chirps_col.select('precipitation').filterBounds(mwi_adm2).map(lambda image: image.clip(mwi_adm2))

# Create a single multiband image where each band is the chirps data from a different date
chirps_image = chirps_col.toBands() 
out_dir = os.path.join(os.path.expanduser('~'), 'Downloads')
out_chirps_stats = os.path.join(out_dir, 'chirps_stats.csv')  
geemap.zonal_statistics(chirps_image, mwi_adm2, out_chirps_stats, statistics_type='MAXIMUM')

[giswqs]

Try using a smaller ImageColleciton, e.g., using a for loop to do it annually.

[hannahker]

Ah that makes sense, thank you!

Do you know if there is a way for geemap.zonal_statistics() to output a dataframe instead of writing to a csv, so that I can clean and combine the files before outputting them?

[giswqs]

FeatureCollection exported from GEE has to be either CSV or shapefile. If you want a dataframe, you can read the CSV back as a dataframe within the loop. Something like this:

for i in range(0, 10):
  out_csv = str(i) + ".csv"
  geemap.zonal_statistics(image, zone, out_csv)
  df = pd.read_csv(out_csv)
  # whatever you want to do with df

[hannahker]

Thank you!