COG bbox queries return same statistics as the entire file

[ammarsalman94]

So I have interest in requesting screen bounds and I am using Leaflet.js and retrieving correct coordinates from the view, then requesting the statistics using

/cog/statistics?url=/data/landmass_cog.tif&bbox=10,0,30,20&coord_crs=epsg:4326

Those statistics never update irrespective of what the bounding box is.

Things I have tried:

1. Set max_size: I set it to different values, but it seems to just resample the entire map instead of the area in the bbox.
2. Set bbox-crs and coord_crs: neither had any effect
3. Validate the TiTiler is receiving the full query from nginx behind the reverse proxy
4. Test with and without COG overviews, tested with many levels of overviews, and no levels.
5. Different values for bbox: 0,0,60,60... 0,0,1,1... 0,1,0,1... All return the same thing

The reason I want this is because my COG file contains 1 band of elevation data for all landmasses in the globe. I want to use the min/max to change the rescale parameters when the view changes, automatically.

TiTiler docker image tags tested: 2.0.0, 2.0.1

Here is the gdalinfo of my COG file:

gdalinfo landmass_cog.tif 

Driver: GTiff/GeoTIFF
Files: landmass_cog.tif
Size is 162000, 81000
Coordinate System is:
GEOGCRS["WGS 84",
    ENSEMBLE["World Geodetic System 1984 ensemble",
        MEMBER["World Geodetic System 1984 (Transit)"],
        MEMBER["World Geodetic System 1984 (G730)"],
        MEMBER["World Geodetic System 1984 (G873)"],
        MEMBER["World Geodetic System 1984 (G1150)"],
        MEMBER["World Geodetic System 1984 (G1674)"],
        MEMBER["World Geodetic System 1984 (G1762)"],
        MEMBER["World Geodetic System 1984 (G2139)"],
        ELLIPSOID["WGS 84",6378137,298.257223563,
            LENGTHUNIT["metre",1]],
        ENSEMBLEACCURACY[2.0]],
    PRIMEM["Greenwich",0,
        ANGLEUNIT["degree",0.0174532925199433]],
    CS[ellipsoidal,2],
        AXIS["geodetic latitude (Lat)",north,
            ORDER[1],
            ANGLEUNIT["degree",0.0174532925199433]],
        AXIS["geodetic longitude (Lon)",east,
            ORDER[2],
            ANGLEUNIT["degree",0.0174532925199433]],
    USAGE[
        SCOPE["Horizontal component of 3D system."],
        AREA["World."],
        BBOX[-90,-180,90,180]],
    ID["EPSG",4326]]
Data axis to CRS axis mapping: 2,1
Origin = (-180.000138888888898,90.000000000000000)
Pixel Size = (0.002222222222222,-0.002222222222222)
Metadata:
  AREA_OR_POINT=Area
Image Structure Metadata:
  COMPRESSION=DEFLATE
  INTERLEAVE=BAND
  LAYOUT=COG
  PREDICTOR=2
Corner Coordinates:
Upper Left  (-180.0001389,  90.0000000) (180d 0' 0.50"W, 90d 0' 0.00"N)
Lower Left  (-180.0001389, -90.0000000) (180d 0' 0.50"W, 90d 0' 0.00"S)
Upper Right ( 179.9998611,  90.0000000) (179d59'59.50"E, 90d 0' 0.00"N)
Lower Right ( 179.9998611, -90.0000000) (179d59'59.50"E, 90d 0' 0.00"S)
Center      (  -0.0001389,  -0.0000000) (  0d 0' 0.50"W,  0d 0' 0.00"S)
Band 1 Block=512x512 Type=Int16, ColorInterp=Gray
  Min=-11603.000 Max=22768.000 
  NoData Value=-32768
  Overviews: 81000x40500, 40500x20250, 20250x10125, 10125x5062, 5062x2531, 2531x1265, 1265x632, 632x316, 316x158
  Metadata:
    STATISTICS_MAXIMUM=22768
    STATISTICS_MINIMUM=-11603

Here is the ONLY config I get from the requests:

{"b1":{"min":-884.0,"max":5924.0,"mean":1092.0330227588536,"count":180633.0,"sum":197257201.0,"std":1143.7629668494217,"median":559.0,"majority":3.0,"minority":-884.0,"unique":5316.0,"histogram":[[54,83269,35065,17230,13806,20515,8240,1206,1075,173],[-884.0,-203.20000000000005,477.5999999999999,1158.3999999999999,1839.1999999999998,2520.0,3200.7999999999997,3881.5999999999995,4562.4,5243.2,5924.0]],"valid_percent":34.45,"masked_pixels":343655.0,"valid_pixels":180633.0,"description":"b1","percentile_2":4.0,"percentile_98":3707.0}}

I know that both Min=-11603.000 Max=22768.000 values are wrong, but I am kinda stuck and have no idea what to do next.

[vincentsarago]

The statistics endpoint does not accept bbox https://developmentseed.org/titiler/endpoints/cog/#statistics

You can use the POST - /statistics endpoint with a GeoJSON feature to achieve this

[ammarsalman94]

Thanks, it worked!

I tested it with httpx and provided the geojson and it pulled the data. Although it is a little slow on higher resolutions so I likely need to calculate max_size based on the polygon's coverage area to make sure contrast becomes better at more zoom levels.