There's an issue with the converted netCDFs:

plia_id = Path(f"{ROOT_DATA}openEO_local/s1_parameters/S1_CSAR_IWGRDH/PLIA-TAG/V01R03/EQUI7_EU020M/E054N006T3/PLIA-TAG-MEAN_20200101T000000_20201231T235959__D080_E054N006T3_EU020M_V01R03_S1IWGRDH.nc")
sig0_id = Path(f"{ROOT_DATA}openEO_local/s1_parameters/S1_CSAR_IWGRDH/SIG0/V1M1R1/EQUI7_EU020M/E054N006T3/SIG0_20180228T043908__VV_D080_E054N006T3_EU020M_V1M1R1_S1AIWGRDH_TUWIEN.nc")

Running

sig0_dc_xr = sig0_dc.execute()
sig0_dc_xr

Same problem for the plia file.

You will notice that there is a band dimension which shouldn't be there, since it's a Dataset and the data is under variables.

I guess that the fix would be to add a .drop("band") or add "band" inside squeeze:

In openEO we defined and implemented two processes to work with harmonic models fitting and predicting, which are:

fit_curve:

• definition https://processes.openeo.org/#fit_curve
• implementation: https://github.com/Open-EO/openeo-processes-dask/blob/c09c805e337f357af48e393cc84a0719639b6b05/openeo_processes_dask/process_implementations/ml/curve_fitting.py#L17

predict_curve:

• definition: https://processes.openeo.org/#predict_curve
• implementation: https://github.com/Open-EO/openeo-processes-dask/blob/c09c805e337f357af48e393cc84a0719639b6b05/openeo_processes_dask/process_implementations/ml/curve_fitting.py#L102

Currently in the code there is a prediction step for a single date (when it's already known there's a flood). I guess it would be interesting to predict values for a longer time series and then work with the result, implementing the flood detection/alarm part.

@ms801221 @fl0roth @MartinSchobben @SwamyDev
I was wondering what's the plan in terms of usage of this code.

From what I understand:

• Currently it's using pre-defined netCDFs with Sentinel-1 data (backscatter and local incidence angle) and a netCDF containing pre-computed parameters of the fitted function to the current area.
• It demonstrates how to use openEO to build and run the flood detection workflow.
• The plan for moving forward is to use the Sentinel-1 Sigma0 data and openEO back-end provided by EODC (instead of the local data and processing).

Some open points which are not documented yet are:

• How to compute the fitted parameters? Is it planned to provide the code for it?
• What are the specifications of the input data? (Resolution, other important parameters used for the pre-processing?) Please specify them if possible.

Since it now does both local and remote processing