Open Issues Need Help
View All on GitHubAI Summary: This issue identifies a potential temporal bias in the S2 GeoMAD analysis for mangrove and seagrass stock estimation. The bias arises because the statistical selection process for the annual snapshot can be skewed by factors like cloud cover, leading to an inaccurate representation of stocks in certain years. The user is seeking clarification and potential solutions to mitigate this temporal skewing.
AI Summary: Investigate and resolve the issue of increased noise and inaccuracies in GeoMAD results for earlier years (2017-2020) compared to later years. This involves testing the GeoMAD implementation's performance on large datasets, evaluating different chunk/thread configurations, validating MAD outputs, and potentially adapting the algorithm to handle data characteristics specific to earlier years. After resolving the Landsat issues, extend the solution to Sentinel-2 data.