Allow pixels to be treated as areas on demand

[rajadain]

Overview

Previously, we would always treat raster pixels as points, so if a pixel's centroid was covered in an area of interest it would be included, otherwise not. For certain rasters, where the pixel sizes are very large, for small areas of interest, no pixels are included since the areas do not cover their centroids.

By allowing the calling JSON to specify whether to treat pixels as points or areas, we allow treating certain rasters (as determined by the caller) as having pixels as areas, which will make small areas of interests run again.

Connects https://github.com/WikiWatershed/model-my-watershed/issues/2253

Demo

$ http --print HhBb :8090/run < SquareKmPpt.json

POST /run HTTP/1.1
Accept: application/json, */*
Accept-Encoding: gzip, deflate
Connection: keep-alive
Content-Length: 498
Content-Type: application/json
Host: localhost:8090
User-Agent: HTTPie/0.9.9

{
    "input": {
        "operationType": "RasterGroupedAverage",
        "pixelIsArea": true,
        "polygon": [
            "{\"type\":\"MultiPolygon\",\"coordinates\":[[[[-75.17123123758348,39.94946918388007],[-75.1595030153462,39.94946918388007],[-75.1595030153462,39.95845957765271],[-75.17123123758348,39.95845957765271],[-75.17123123758348,39.94946918388007]]]]}"
        ],
        "polygonCRS": "LatLng",
        "rasterCRS": "ConusAlbers",
        "rasters": [],
        "targetRaster": "climatology-ppt-09-epsg5070",
        "zoom": 0
    }
}

HTTP/1.1 200 OK
Content-Length: 40
Content-Type: application/json
Date: Wed, 13 Sep 2017 21:55:48 GMT
Server: akka-http/10.0.9

{
    "result": {
        "List(0)": 93.80207061767578
    }
}

Testing Instructions

• Download SquareKmPpt.json.txt
• On develop, POST it to the geoprocessing service. Ensure you see a 0 in the output.
• Check out this branch. POSTing the same file should now show a real value.
• Build and deploy the JAR to your worker VM. Restart the geoprocessing service.
• Apply the following patch to your base.py:

diff --git a/src/mmw/mmw/settings/base.py b/src/mmw/mmw/settings/base.py
index 8349e16..409a032 100644
--- a/src/mmw/mmw/settings/base.py
+++ b/src/mmw/mmw/settings/base.py
@@ -536,6 +536,7 @@ GEOP = {
                 'polygonCRS': 'LatLng',
                 'rasters': [],
                 'targetRaster': 'climatology-ppt-{:02d}-epsg5070',
+                'pixelIsArea': True,
                 'rasterCRS': 'ConusAlbers',
                 'operationType': 'RasterGroupedAverage',
                 'zoom': 0
@@ -547,6 +548,7 @@ GEOP = {
                 'polygonCRS': 'LatLng',
                 'rasters': [],
                 'targetRaster': 'climatology-tmean-{:02d}-epsg5070',
+                'pixelIsArea': True,
                 'rasterCRS': 'ConusAlbers',
                 'operationType': 'RasterGroupedAverage',
                 'zoom': 0

• Check the MMW UI. Ensure that all the regular values are identical to that on staging.
• Ensure climate values are populated for small areas of interest.

[kellyi]

Here's the current output from develop when posting the above file:

{
    "result": {
        "List(0)": null
    }
}

... and here's the output from this branch:

{
    "result": {
        "List(0)": 93.80207061767578
    }
}

[kellyi]

Tried out the prior operations & the values remain the same as on develop.

[kellyi]

Seeing values for 1 sq km in MMW:

[kellyi]

This looks good to me!

I tested it with the jar and in MMW/BigCZ and both worked as described above. The values for the existing operations remained the same post-pixelIsArea, and the value for the new pixel op is now coming back as not null.