Climate Forecast Dashboard

The Climate Forecast Dashboard builds on the functionality of the Climate Report Dashboard, adding predictive capabilities through the use of Machine Learning (ML) models. These tools allow users to explore climate trends for the next 7 days, providing valuable insights for farmers, local authorities, researchers and disaster response teams. The forecast was designed to serve both as a planning tool and an early warning system for heat waves, floods, or other extreme weather events at a local scale.
The forecasted parameters include precipitation, average temperature, relative humidity, solar radiation, but in the future more parameters will be included.
For each parameter, the blue trend line represents observed historical data, the red trend line shows the forecasted values, and the shaded portion represents a 95% confidence interval.

Implemented Models:

Research was done on the best performing models for climate data forecasting, several models were trained, and the two best performing models were used.

• Machine Learning : XG-Boost was the best performing amongst the machine learning models.
• Deep learning: : LSTM – best-performing amongst deep learning models.

These models were trained using processed historical weather data and validated with recent observations to ensure reliability and accuracy in forecasting tasks up to 7 days. Full documentation of the models can be accessed on request with information about, Metadata and data processing method, Description of the models used, model validation procedures and guidelines for interpreting forecast results.