ANIN: SPEI

[EmileSonneveld]

Implement SPEI from here: https://git-ext.gmv.com/anin-external/drought-indices.git Same principle as SPI

[EmileSonneveld]

Will probably need hourly data to calculate proper MIN/MAX. Depends on https://github.com/Open-EO/openeo-geopyspark-driver/issues/548

[EmileSonneveld]

Daily min/max based on hourly data is available here: /data/users/Public/johan.schreurs/ANIN/reanalysis-era5-land_southafrica_float32/ Thanks to @JohanKJSchreurs

[EmileSonneveld]

When loading agera5 land collections we need to apply an offset and a scale. In GMV code that is done by reading the netcdf metadata. We will probably need to implement it under this ticket: https://github.com/Open-EO/openeo-geopyspark-driver/issues/338

The current workaround is to apply the offset and scale in the UDF, by manually pasting the values in the code.

[EmileSonneveld]

Status update:

• ERA5 was re-imported to fix some invalid images, make CRS explicitly embedded, and to preserve scaling and offset in metadata.
• There is a scaling and offset issue difference between the GMV data and the openEO imported data, causing a significant difference in output. It is only a scaling issue, because the linear correlation between 2 input sets is 0.9999...
• Applying scaling in different ways did not fix it. I can try to use GMV's input data to see if I at least get same results then.
• MIN/MEAN/MAX temperature based on hourly data can be validated if the check above shows differences

[EmileSonneveld]

Using ERA5 based on GMV inputs did fix the differences. However, there is a sub-pixel offset. Maybe this ticket combined with load_stac would fix it: https://github.com/Open-EO/openeo-geopyspark-driver/issues/648 But the corelation is already top

[EmileSonneveld]

If NRT is required, this can be implemented in a new ticket