IMD’s Multi-Model Ensemble System Improves Monsoon Forecast Accuracy
The India Meteorological Department (IMD) has significantly improved the accuracy of its long-range weather forecasts by adopting the Multi-Model Ensemble (MME) forecasting system since 2021. Developed under the Ministry of Earth Sciences (MoES), the MME approach combines predictions from multiple weather and climate models to produce more reliable forecasts while reducing the chances of errors from any single model. According to the government, all operational monsoon forecasts issued between 2021 and 2025 remained within the prescribed error limits, with an average absolute forecast error of just 2.2% of the Long Period Average (LPA). Over the past decade, the forecast error exceeded 10% of the LPA only once—in 2019, when it reached 14%. The improved forecasting system has enhanced the reliability of seasonal monsoon predictions, providing better support for agriculture, water resource management, and disaster preparedness across the country.
1. What is “Forecast Error”?
A forecast error is simply the difference between what was predicted and what actually happened.
Formula
Forecast Error = Actual Rainfall − Forecast Rainfall
When IMD says the error is 2.2% of the Long Period Average (LPA), it means the forecast was very close to the actual rainfall.
Example
Suppose the Long Period Average (LPA) rainfall is 100 cm.
| Forecast | Actual Rainfall | Error |
| 98 cm | 100 cm | 2 cm (2%) |
| 103 cm | 100 cm | 3 cm (3%) |
| 90 cm | 100 cm | 10 cm (10%) |
So,
- 2% error = Excellent forecast
- 5% error = Good forecast
- 10%+ error = Less accurate forecast