Earth, Atmosphere & Environment Sciences
Title : | Machine learning-based computational models for the analysis of monsoon activities over the Indian region at temporal and spatial scale |
Area of research : | Earth, Atmosphere & Environment Sciences |
Principal Investigator : | Dr. Parashjyoti Borah, Indian Institute Of Information Technology, Guwahati, Assam |
Timeline Start Year : | 2022 |
Timeline End Year : | 2024 |
Contact info : | parashjyoti@hotmail.com |
Equipments : | Computer Monitor
Printer
Keyboard and Mouse
Workstation |
Details
Executive Summary : | Rain is an essential phenomenon for the lifeline of the Indian civilization, specially the agricultural industry. A majority of the Indian population primarily depends on agricultural practices for livelihood. Subsequently, the agriculture sector adds up a large portion of the Indian economy. For water resources Indian farmers are primarily dependent on rainfall during the monsoon season. An erratic monsoon season can hugely impact the agricultural productivity and growth and thereby influencing the Indian economy. A deficit monsoon might develop droughts affecting crops and groundwater depletion, whereas an excessive monsoon may lead to natural calamities such as floods, landslides, etc. India is spread across a huge geographic area; therefore, it is quite obvious that there is a huge variation in monsoonal rainfall in different parts of the country. When, variation of monsoon is considered, apart from the year to year variation, there exists large intra-seasonal variation. An excess rainfall during a certain period and a deficient during another period within the same season is also a common feature of monsoon. However, a precise and early forecasting of monsoon activity at both temporal and spatial scale significantly helps the policy makers, so that well-planned and timely actions could be undertaken. The operational weather forecasting responsibilities are being carried out by the India Meteorological Department (IMD), Ministry of Earth Sciences, Govt. of India. IMD predicts the Indian monsoon by analysing different climatic variables and predictors using some statistical or/and multi-model ensembles. These models are continually updated for improved accuracy in monsoon forecasting. However, it is evident these days that machine learning (including deep learning) and artificial intelligence performs great in analysing complex data when fed with the appropriate input, and even outperforms traditional techniques in most cases. Machine learning is an evolving research area that deals with computational algorithms to analyse complex data. Machine learning algorithms are widely applied in different application areas and are proven to be very efficient. There are some studies available in the literature that applies machine learning tools in monsoon forecasting. However, it is felt that machine learning field is not adequately explored in analysing weather related activities, specifically, the Indian monsoon. Moreover, the dependencies of climatic variables and predictors are not adequately studied in both temporal and spatial scale. This project aims at carrying out a detailed comprehensive literature review on the existing traditional and computational models for monsoon forecasting; and developing novel/hybrid computational models with improved way of analysing the climatic variables and predictors for a more precise forecasting of the Indian monsoon activities to help the policy-makers undertake well-planned and timely actions. |
Total Budget (INR): | 18,78,531 |
Organizations involved