Last active
April 24, 2020 11:23
-
-
Save tastatham/cfed5c4fcd3a5ed74d78948d0cc78162 to your computer and use it in GitHub Desktop.
dask_geomodeling_zonalstats.py
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
def compute_zonal_stats(raster, vector, fid, col, statistics,tile,mode): | |
""" | |
Calculate zonal statistics using dask | |
Parameters | |
---------- | |
raster : A GeoDataFrame to create multiple copies based on the list of density thresholds | |
vector : ESRI Shapefile vector grid (1km) | |
fid : Primary key (default is fid for exporting ESRI Shapefiles) | |
col : Column name for zonal stats | |
statistics : Zonal statistics aggregation | |
tile : How large to split the raster database to fit in memory | |
mode : Centroid / extent | |
Returns | |
------- | |
A GeoDataframe containing the computed zonal statistics | |
""" | |
# construct the View | |
rs = RasterFileSource(raster) | |
gs = GeometryFileSource(vector, id_field=fid) | |
# Define zonal statistics to calculate | |
agg = AggregateRaster(gs, rs, column_name=col, statistic=statistics) | |
# Build the data request | |
x1, x2, y1, y2 = rs.geometry.GetEnvelope() | |
projection = rs.projection | |
request = dict(geometry=box(x1, y1, x2, y2), projection=projection) | |
# Tile geometry | |
tiled = parallelize.GeometryTiler(agg, size=tile, projection=projection) | |
# Print statement | |
print('Geometry Tiler done for %s' %(raster)) | |
# Define graph to store how to compute using dask | |
graph, name = tiled.get_compute_graph(**request, mode=mode) | |
# Print statement | |
print('Compute graph done for %s' %(raster)) | |
# Compute using dask | |
tuple_output = get(graph,name) # Dask compute always returns tuples | |
# Convert to GeoDataFrame | |
gdf = list(tuple_output.items())[0][1] | |
return(gdf) |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment