zonal statistics on Vector and NetCDF data

[rylanlee]

Since zonal statistics on Vector and \Raster data is possible as Integrating with GeoPandas and Numpy. Is zonal statistics on Vector and NetCDF data possible? Moreover, is calculation in each zone (each region in vector data) possible based ?

For example, I have 1980-2010 globla temperature, How can I get the 1980-2010 mean temperature series in each vectors? Can anyone help me figure it out?

[perrygeo]

rasterstats is designed around summarizing a single raster band by multiple vector shapes. It sounds like what you want to do is summarize multiple raster bands (temperature for each year) by multiple shapes?

There is an open issue about multiband support: see #73

For now, you can iterate over each band and combine the data structure with a little post processing.
The following example demonstrates the approach I would take:

from rasterstats import zonal_stats
import json
import rasterio

multiband_raster = "rgba.tif"
vect = "shapes.geojson"

# Use rasterio to determine the band_indexes
with rasterio.open(multiband_raster) as src:
    band_indexes = src.indexes

all_band_summaries = []

for b in band_indexes:
    band_summary = zonal_stats(
        vect, multiband_raster, prefix=f"band{b}_", band=b, stats="min")  # note the prefix
    all_band_summaries.append(band_summary)

# all_band_summaries is now a 2D list of lists
# first dimension: bands
# second dimension: shapes
# We can flip the dimensions using zip
shape_summaries = list(zip(*all_band_summaries))
# then aggregate into a single list (dimension: shapes) containing bands
final = [{k: v for d in s for k, v in d.items()}
         for s in shape_summaries]

print(json.dumps(final, indent=2))

Which will print a list of dictionaries, one for each vector shape, containing the minimum value for each band:

[
  {
    "band1_min": 214.0,
    "band2_min": 220.0,
    "band3_min": 255.0,
    "band4_min": 255.0
  },
  {
    "band1_min": 133.0,
    "band2_min": 155.0,
    "band3_min": 171.0,
    "band4_min": 255.0
  },
  {
    "band1_min": 65.0,
    "band2_min": 69.0,
    "band3_min": 80.0,
    "band4_min": 255.0
  }
]

Something like this could be adapted to NetCDF files and annual temperature grids as bands, and of course all the vector formats as described in the docs.

Please give this a shot and report back if you have any questions.

[perrygeo]

Closing issue due to inactivity. The above code is the recommended approach.