Section / TensorFlow Rainfall Forecast

All-data rainfall analytics and fast TensorFlow LSTM pilot forecast

Combine stored NetCDF/CSV rainfall files, manual uploads, full-period filtering, QA/QC, graph-first visualization, and optional fast LSTM-style forecasting.

Rainfall AI Lab

From NetCDF archive to fast flood-risk forecast

Load the full available IMD-style rainfall record first, then optionally run a capped TensorFlow LSTM pilot forecast without slowing normal graph review.

Default Mode All available data Monthly aggregation, TensorFlow off until selected

01SourceStored NetCDF/CSV or manual upload

02PrepareSort, combine, dedupe, aggregate

03VisualizeGraph loads before model forecast

04ForecastFast LSTM/proxy pilot output

Data Preparation

Repository and period controls

Data Period Start Date End Date Aggregation

Manual data input CSV / NetCDF / NC4 Expected columns: date/time and rainfall/precipitation. NetCDF support depends on server xarray availability.

Permanent repositorySave uploads into data/netcdf TensorFlow LSTM pilotOff by default for fast graph viewing

Fast Model Preset

Demo-safe LSTM parameters

Epochs Lookback Forecast Steps LSTM Units Max Samples

Click inside India/Madhya Pradesh to plot the nearest 0.25 degree IMD rainfall grid cell.

Extreme Rainfall Events

S.No.	District	Year	Date	Rainfall (mm)
Select a point or load analysis to populate district-wise annual extreme events.

Analytics Output

Rainfall signal, risk band, and forecast

Load the module to view all-data rainfall graph, QA/QC, and optional forecast.

Recordsn/aselected rows

Total Rainfalln/aperiod sum

Wet Daysn/a> 2.5 mm

Max Dailyn/aextreme signal

Risk Bandn/ascreening score

TensorFlown/aruntime status

Historical Rainfall All available period

Pilot Forecast Forecast disabled until LSTM is selected