soil-moisture-prediction

1. Download HDF5 files from SMAP

Use the python code to download the SMAP L4 data in HDF5 format for the required dates. For more information visit SMAP L4

2. Pre-process files

The Deep learning model uses images as it's inputs. So convert the HDF5 file to JPEG image using the Pre-processing code. This is done for two features Soil Moisture and Precipitation.

3. Train the model

Use the tranining code to train a ConvLSTM model

4. Predictions

Use the prediction code to predict for the required date. Note that for validation purposes all the dates , including the date to be predicted, data must be loaded.

5. Post process the predicted CSV file

Finally post-process the global predicted CSV file to include the estimated bias for selected Ground Points.

Questions?

Contact: Archana Kannan

Email: [email protected]