AI can boost predicting if it can handle obstacles

• India uses AI for weather forecasting and early warnings due to intense heat waves and rainfall.

• Traditional weather forecasting uses numerical models, which simulate atmospheric behavior using fluid dynamics and thermodynamics.

• AI models start with data, learning relationships between inputs and outputs without prior knowledge of underlying earth system processes.

• AI models can explore hidden links between earth-system variables to uncover cause-effect relationships.

• The Indian government announced ‘Mission Mausam’ in September 2024 with an allocation of ₹2,000 crore over two years to enhance weather and climate observations.

• The Ministry of Earth Sciences has set up a dedicated AI and machine-learning (ML) centre to develop and test techniques to improve short-range rain forecasts, develop high-resolution urban meteorological datasets, and nowcast rainfall and snow using data from Doppler radars.

• Researchers are also using AI to predict weather, with an international team developing an ML model to predict monsoon rainfall.

• Challenges include the nonlinear and chaotic nature of weather systems, the need for large, high-quality datasets, and the complexity of AI outputs.

• Efforts are underway to develop hybrid approaches by combining AI/ML with physics-based modelling for weather forecasting.

• Many weather forecasters in India use information generated from other agencies or a combination of data produced by multiple models, overlaying local knowledge.

Challenges in AI/ML Predicting Weather

• Human resources, particularly at the interface between AI and weather prediction, are a significant hurdle.

• Climate science, a discipline involving scientists from various disciplines, is often viewed as akin to a black box without the necessary AI/ML expertise.

• The lack of AI/ML expertise in climate science limits the scope of deep research and progress.

• India’s diverse topography and climate zones demand regionally tailored models, increasing development complexity.

• The need for more data and computing power is a never-satiable demand, and collaborations between climate scientists and AI/ML scientists are essential.

• A critical shortage of professionals with expertise in both meteorology and machine learning hinders the development and deployment of advanced models.

• The availability of long-term data of high quality is a significant challenge.

• Scientists worldwide are trying to overcome challenges in using ML for climate science.

• Hybrid modelling, combining physics-based climate models with ML tools, is emerging as an emerging enterprise.

• AI/ML models can be particularly useful to predict extreme weather events like heat waves and torrential rainfall.

• However, accurate predicting and modeling extreme weather events is crucial due to their localized and rapid development.

• The complexity and interpretability of ML models, the difficulty of generalising across different contexts, and the quantification of uncertainty are concerns.