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 and climate models. These tools allow users to explore climate trends for the next 10 days, providing valuable insights for farmers, local authorities, researchers and disaster response teams. The dashboard serves 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, Air temperature, Relative humidity, Sunshine duration, Solar radiation.

For each parameter, the black trend line represents observed historical data, the dash trend line shows the Moving average, while the red trend line shows the forecasted values. Model-predicted accuracy is displayed for each parameter based on the selected model.

Implemented Models:

• Machine Learning and deep learning: Tree-based models, Support Vector Regression (SVR), Multivariate Adaptive Regression Splines (MARS) – best-performing model selected. Fast forward Neural network and
• Statistical: ARIMA and SARIMA – best-performing model selected.

These models were trained using processed historical weather data and validated with recent observations to ensure reliability and accuracy in forecasting tasks such as rainfall prediction and temperature forecasting up to 10 days. Full documentation of the models can be accessed at (inserts link), with information about, Metadata and data processing method, Description of the ML and statistical models used, Model validation procedures and Guidelines for interpreting forecast results.