/statistics endpoint usage ?

[nuayge]

Hi everyone, I have TiTiler-PGStac running with the noaa-emergency-response collection from the tests folder.

I'm trying to get some statistics for multiple geometries.

If I send two points, I receive back a FeatureCollection with two points, but they have the same values

noaa_points = {
    "type": "FeatureCollection",
    "features": [
        {
            "type": "Feature",
            "properties": {},
            "geometry": {"coordinates": [-85.47477772241675, 36.152795257931984], "type": "Point"},
        },
        {
            "type": "Feature",
            "properties": {},
            "geometry": {"coordinates": [-85.46603751623104, 36.136352659372946], "type": "Point"},
        },
    ],
}
res = httpx.post(
    url="http://127.0.0.1:8081/collections/noaa-emergency-response/statistics?assets=cog&pixel_selection=first",
    json=noaa_points,
)

res.json()

{'type': 'FeatureCollection',
'features': [{'type': 'Feature',
'geometry': {'type': 'Point',
'coordinates': [-85.47477772241675, 36.152795257931984]},
'properties': {'statistics': {'cog_b1': {'min': 81.0,
'max': 81.0,
'mean': 81.0,
'count': 0.009999999776482582,
'sum': 0.8100000023841858,
'std': 0.0,
'median': 81.0,
'majority': 81.0,
'minority': 81.0,
'unique': 1.0,
'histogram': [[0, 0, 0, 0, 0, 1, 0, 0, 0, 0],
[80.5, 80.6, 80.7, 80.8, 80.9, 81.0, 81.1, 81.2, 81.3, 81.4, 81.5]],
'valid_percent': 100.0,
'masked_pixels': 0.0,
'valid_pixels': 1.0,
'percentile_2': 81.0,
'percentile_98': 81.0},
'cog_b2': {'min': 72.0,
'max': 72.0,
'mean': 72.0,
'count': 0.009999999776482582,
'sum': 0.7199999690055847,
'std': 0.0,
'median': 72.0,
'majority': 72.0,
'minority': 72.0,
'unique': 1.0,
'histogram': [[0, 0, 0, 0, 0, 1, 0, 0, 0, 0],
[71.5, 71.6, 71.7, 71.8, 71.9, 72.0, 72.1, 72.2, 72.3, 72.4, 72.5]],
'valid_percent': 100.0,
'masked_pixels': 0.0,
'valid_pixels': 1.0,
'percentile_2': 72.0,
'percentile_98': 72.0},
'cog_b3': {'min': 73.0,
'max': 73.0,
'mean': 73.0,
'count': 0.009999999776482582,
'sum': 0.7299999594688416,
'std': 0.0,
'median': 73.0,
'majority': 73.0,
'minority': 73.0,
'unique': 1.0,
'histogram': [[0, 0, 0, 0, 0, 1, 0, 0, 0, 0],
[72.5, 72.6, 72.7, 72.8, 72.9, 73.0, 73.1, 73.2, 73.3, 73.4, 73.5]],
'valid_percent': 100.0,
'masked_pixels': 0.0,
'valid_pixels': 1.0,
'percentile_2': 73.0,
'percentile_98': 73.0}}}},
{'type': 'Feature',
'geometry': {'type': 'Point',
'coordinates': [-85.46603751623104, 36.136352659372946]},
'properties': {'statistics': {'cog_b1': {'min': 81.0,
'max': 81.0,
'mean': 81.0,
'count': 0.009999999776482582,
'sum': 0.8100000023841858,
'std': 0.0,
'median': 81.0,
'majority': 81.0,
'minority': 81.0,
'unique': 1.0,
'histogram': [[0, 0, 0, 0, 0, 1, 0, 0, 0, 0],
[80.5, 80.6, 80.7, 80.8, 80.9, 81.0, 81.1, 81.2, 81.3, 81.4, 81.5]],
'valid_percent': 100.0,
'masked_pixels': 0.0,
'valid_pixels': 1.0,
'percentile_2': 81.0,
'percentile_98': 81.0},
'cog_b2': {'min': 72.0,
'max': 72.0,
'mean': 72.0,
'count': 0.009999999776482582,
'sum': 0.7199999690055847,
'std': 0.0,
'median': 72.0,
'majority': 72.0,
'minority': 72.0,
'unique': 1.0,
'histogram': [[0, 0, 0, 0, 0, 1, 0, 0, 0, 0],
[71.5, 71.6, 71.7, 71.8, 71.9, 72.0, 72.1, 72.2, 72.3, 72.4, 72.5]],
'valid_percent': 100.0,
'masked_pixels': 0.0,
'valid_pixels': 1.0,
'percentile_2': 72.0,
'percentile_98': 72.0},
'cog_b3': {'min': 73.0,
'max': 73.0,
'mean': 73.0,
'count': 0.009999999776482582,
'sum': 0.7299999594688416,
'std': 0.0,
'median': 73.0,
'majority': 73.0,
'minority': 73.0,
'unique': 1.0,
'histogram': [[0, 0, 0, 0, 0, 1, 0, 0, 0, 0],
[72.5, 72.6, 72.7, 72.8, 72.9, 73.0, 73.1, 73.2, 73.3, 73.4, 73.5]],
'valid_percent': 100.0,
'masked_pixels': 0.0,
'valid_pixels': 1.0,
'percentile_2': 73.0,
'percentile_98': 73.0}}}}]}

If I send two Polygons,

noaa_features = {
    "type": "FeatureCollection",
    "features": [
        {
            "type": "Feature",
            "properties": {},
            "geometry": {
                "coordinates": [
                    [
                        [-85.47628747804225, 36.14168866536674],
                        [-85.47628747804225, 36.13835851797647],
                        [-85.47216385822023, 36.13835851797647],
                        [-85.47216385822023, 36.14168866536674],
                        [-85.47628747804225, 36.14168866536674],
                    ]
                ],
                "type": "Polygon",
            },
        },
        {
            "type": "Feature",
            "properties": {},
            "geometry": {
                "coordinates": [
                    [
                        [-85.46872575693266, 36.141548605973554],
                        [-85.46872575693266, 36.133918423537935],
                        [-85.45807530007667, 36.133918423537935],
                        [-85.45807530007667, 36.141548605973554],
                        [-85.46872575693266, 36.141548605973554],
                    ]
                ],
                "type": "Polygon",
            },
        },
    ],
}

res = httpx.post(
    url="http://127.0.0.1:8081/collections/noaa-emergency-response/statistics?assets=cog&pixel_selection=first",
    json=noaa_features,
)

the server crashes with

pidex = self.mosaic.mask & ~array.mask
            ~~~~~~~~~~~~~~~~~^~~~~~~~~~~~~
ValueError: operands could not be broadcast together with shapes (3,828,1024) (3,734,1024)

I understand the two points are "merged" and then re-splitted into the original features due to them having the exact same values.

This issue: #122 hints at something similar, but I maybe did not understand how I should use the statistics endpoint: I would expect each geometry to get its own statistics, am I getting this wrong ?

[vincentsarago]

@nuayge can you share the version of titiler-pgstac you're using?

[nuayge]

I'm on 1.9.0