Quality of images for smaller polygon

Apps / Services
1

Hello,

Currently we are using the process api to retrieve images from our polygons.

For larger polygons (> 500ha) the resolution is good, we are using 1000px x 1000px. (width and height)

But when the polygon is small the resolution is very bad.

Is there any workaround to acquire best images for small polygons ?

Used payload

{
   "input":{
      "bounds":{
      	    "properties": {
                "crs": "http://www.opengis.net/def/crs/EPSG/0/4326"
            },
         "geometry":{
            "type":"Polygon",
            "coordinates":[
               [
                  [
                     -51.0705149173737,
                     -23.2945561408997
                  ],
                  [
                     -51.0683262348175,
                     -23.294792175293
                  ],
                  [
                     -51.0701501369476,
                     -23.288140296936
                  ],
                  [
                     -51.0706865787506,
                     -23.2888913154602
                  ],
                  [
                     -51.0710299015045,
                     -23.289749622345
                  ],
                  [
                     -51.0715019702911,
                     -23.2906079292297
                  ],
                  [
                     -51.0718023777008,
                     -23.2909083366394
                  ],
                  [
                     -51.0705149173737,
                     -23.2945561408997
                  ]
               ]
            ]
         }
         
      },
      "data":[
         {
            "type":"S2L1C",
            "dataFilter":{
               "timeRange":{
                  "from":"2020-01-18T00:00:00Z",
                  "to":"2020-01-18T23:59:59Z"
               }
            }
         }
      ]
   },
   "output":{
        "width": 1000,
        "height": 1000,
        "responses": [
            {
                "identifier": "default",
                "format": {
                    "type": "image/png",
                    "quality": 100
                }
            }
        ]
   },
   "evalscript":"function setup(){return{input:[\"B08\",\"B04\",\"B02\",\"dataMask\"],output:{bands:4}}}function evaluatePixel(t){let a=(t.B08-t.B04)/(t.B08+t.B04);return 1==t.dataMask?colorBlend(a,[-.2,0,.1,.2,.3,.4,.5,.6,.7,.8,.9,1],[[0,0,0,1],[165/255,0,38/255,1],[215/255,48/255,39/255,1],[244/255,109/255,67/255,1],[253/255,174/255,97/255,1],[254/255,224/255,139/255,1],[1,1,191/255,1],[217/255,239/255,139/255,1],[166/255,217/255,106/255,1],[.4,189/255,99/255,1],[26/255,152/255,80/255,1],[0,104/255,55/255,1]]):[99,99,99,0]}"
}

Image of response

response
response1000×1000 9.52 KB

Regards

Guilherme

2

What you are probably noticing is a limitation of 10 meters resolution. E.g. each pixel you see is 10x10 meters, in line with the sensor capabilities.

You can use upsampling (I suggest bicubic) - note that this does not actually improve resolution, it simply interpolates between pixel values to produce a nicer-looking result.

On another note - make sure that width and height used are proportional to actual width/height of the polygon. E.g. if a polygon has a 3-1 ratio (width/height), you should put width = 1000, height = 333.