A cutting-edge artificial intelligence/machine learning (AI/ML) model for predicting monsoon rainfall has been developed through a collaborative effort between the Department of Science and Technology Centre of Excellence in Climate Modeling at the Indian Institute of Technology in Delhi (IIT-Delhi), the Indraprastha Institute of Information Technology (IIIT-Delhi), and universities in the United States and Japan.
According to the research team, the AI/ML model they developed has predicted an All India Summer Monsoon Rainfall (AISMR) of approximately 790mm for the upcoming monsoon season. This prediction suggests that the country can expect a normal monsoon this year. This information can be valuable for various sectors that rely on monsoon rainfall, such as agriculture and water resource management.
The prediction is made using a training process that incorporated historical AISMR data, the Niño3.4 index data, and categorical Indian Ocean Dipole (IOD) data spanning the period from 1901 to 2001. The model has exhibited superior performance compared to the current physical models employed for monsoon predictions in the country. It has achieved an impressive forecast success rate of 61.9% during the test period from 2002 to 2022. The success rate is determined by assessing whether the model can predict the AISMR within a range of +/-5% of the actual values observed each year.
The model can make forecasts several months in advance, contingent upon the availability of the Niño3.4 index and IOD forecasts. These inputs can be continuously monitored and updated to reflect their evolving conditions. Thus, the data-driven models are flexible to inputs and can better capture the nonlinear relations among the monsoon drivers, all while being less computationally intensive.
Compared to the resource-intensive process associated with traditional physical models, a small team of individuals running the ML models on a personal computer can deliver a more precise monsoon rainfall forecast.
The significance of this study extends to the entire country, as the availability of accurate monsoon forecasts well in advance plays a crucial role in making critical decisions across multiple socioeconomic sectors. Accurate monsoon predictions enable effective planning in agriculture, the efficient management of energy resources, optimal utilisation of water resources, proactive measures in disaster management, and the ability to address health-related concerns. The data-driven techniques developed in this study will be further extended to provide state-wise monsoon rainfall predictions, enhancing their usefulness for regional applications.
Rainfall erosivity, which represents the potential of rain to cause soil degradation, is a significant contributor to the erosion-induced by water in India. Approximately 68.4% of the total eroded soil in the country is affected by erosion induced by water.
Globally, soil erosion caused by rainfall is recognised as a significant environmental challenge. However, traditional assessments of rainfall erosivity in India are often limited to specific catchments or regions, which hinders a comprehensive evaluation of this phenomenon for a country as geographically diverse as India, with varying climate properties.
Last year, a study carried out at IIT-Delhi resulted in the first-ever pan-India assessment of rainfall erosivity. The researchers employed multiple national and global gridded precipitation datasets, including the Indian Monsoon Data Assimilation and Analysis (IMDAA) at an hourly temporal scale, the India Meteorological Department (IMD) on a daily scale, and the Global Climate Hazards Group Infrared Precipitation with Station data (CHIRPS) on a daily scale. By using these datasets, the team created a high-resolution map that identifies areas in India prone to erosion caused by rainfall.
The study is a step toward building a national-scale soil erosion model for India. The map will enable watershed managers to identify rainfall erosivity potential at diverse locations and plan, prioritise, and implement essential watershed development activities to minimise soil erosion.