best way to calculating time series mean&median value #337

[orttak]

I try to get mean or median values of Sentinel COG time series over same tile. It took 3.5-4.5 min. I think my code block take too much time, when I compared gdal or qgis results. Is it right way? Any recommendation?

band_url1='https://sentinel-cogs.s3.us-west-2.amazonaws.com/sentinel-s2-l2a-cogs/37/S/DA/2018/11/S2A_37SDA_20181109_0_L2A/B02.tif'
band_url2='https://sentinel-cogs.s3.us-west-2.amazonaws.com/sentinel-s2-l2a-cogs/37/S/DA/2018/10/S2A_37SDA_20181030_0_L2A/B02.tif'

r38=rioxarray.open_rasterio(band_url1, masked=False, chunks=(1, "auto", -1))
r48=rioxarray.open_rasterio(band_url2, masked=False, chunks=(1, "auto", -1))
stack_bands2 = xr.concat([r38,r48], dim='band', join='outer' )
stack_bands2=stack_bands2.astype(np.float64)
stack_bands2.rio.write_nodata(np.nan, inplace=True)

#time info belongs to below code block
stack_bands2 = stack_bands2.mean(dim='band',skipna=True).compute()
#CPU times: user 6.28 s, sys: 2.52 s, total: 8.8 s
#Wall time: 3min 24s    

[keewis]

just checking: since the wall time is much bigger than the CPU time, did you try downloading those files before opening? I did have some trouble with downloads through rasterio some time ago.