Closed drnextgis closed 3 months ago
@drnextgis The cloud cover should be the percent of the valid data area covered by clouds. This way it will provide a normalized metric for the clouds, otherwise it could heavily depend on how well an image fits into a north-up rectangle and wouldn't be a great metric for cloud cover.
@matthewhanson, I appreciate your prompt response! Would it be alright if I leave this ticket unresolved until we provide further clarification in the README file? The current information there isn't very clear:
Estimate of cloud cover as a percentage (0-100) of the entire scene.
Should the
eo:cloud_cover
percentage be calculated for the entire raster or exclusively for the valid data regions (excluding "nodata" areas)? In the given example, the black regions represent areas not captured by a satellite. It's crucial to determine whether these areas should be factored in when calculatingeo:cloud_cover
, as this could yield markedly different results. If nodata areas are excluded and cloud coverage is computed solely over valid regions, the percentage will approach 100%. Conversely, if nodata areas are included, the percentage will be substantially lower.