A recent deep-learning model finds that till 2027, major parts of the country will be affected by droughts
Representative photo from iStock It is nearly akin to the clichéd “memory vs forgetfulness” syndrome. Just before the onset of its weekly updates by the meteorological department, forecasting the monsoon and reconciling with its actual performance have been an intimate official affair that the country keenly follows.
Of late, the definition of a normal monsoon has lost its relevance as week-by-week and month-by-month forecasts of the monsoon have fallen flat. This demonstrates how erratic the season has become.
Take, for instance, the current monsoon season — from a very late onset in June to an excess in July to a record-breaking spell of pause in August resulting in lowest rainfall since 1901 for the month, this season perfectly defines how unpredictable it has become in a changing climate.
However clichéd it is, monsoon performance is all about gauging the country’s economic well being. The most dreaded impact of a seasonal failure in monsoon is the drought that directly affects agriculture on which the majority of people depend for survival. But, drought forecasting in India has not evolved to a level that can early warn people so that they can prepare to mitigate or adapt.
Declaration of drought is an official, post-seasonal activity based on a defined level of rainfall. However, drought evolves as the monsoon season progresses and each of its four months — June, July, August and September — has a bearing on whether a year will be a drought, particularly the one that impacts agriculture.
Besides, each month’s rainfall progress leads to certain type of weather and hydrological and agricultural crisis. For instance, the low rainfall in August — that contributes around 22 per cent of total seasonal rainfall — can lead to higher use of reservoir storage accumulated during June-July and also more extraction of groundwater. This will also ultimately impacts food production. If considered as a unit, the country is already in a hydrological drought situation and inching towards an agricultural one.
Most of the current initiatives on forecasting drought revolve around forecasting weather in short-term or bringing in periodic rainfall data to point at potential drought in a season. In short, we don’t have a forecasting method to foresee drought in the future even though it will be an essential economic tool in the changed climate.
In February this year, scientists from the SRM Institute of Science and Technology, India, and Skolkovo Institute of Science and Technology, Russia attempted a long-term forecasting of droughts in India.
They deployed the deep learning machine model called “long short-term memory” to analyse rainfall and drought conditions of 117 years (“…using the average monthly forecast from 1901 to 2000 and the model is tested with the average monthly forecast from 2001 to 2017”).
This model picks up past data trends that are used to forecast future trends for 2018-2027.
Its findings show that July 2022 to April 2023 was a “mild wet” period and next April 2027 to October 2027 will be the same. It means all the other periods in this forecast range will be of “mild drought”.
Till 2027, this study says that the states of Karnataka, Gujarat and Uttarakhand will be affected by severe drought while Arunachal Pradesh, Rajasthan, Haryana, Kerala, Maharashtra, Punjab, Tamil Nadu and Telangana are affected by moderate drought.
Bihar, Chhattisgarh, Jammu and Kashmir, Jharkhand, Odisha, West Bengal, Telangana and Sikkim will receive good rainfall but undergo drought every alternate year.
What is the use of this forecast? Drought is known as the disaster that one can see from a long distance thus giving opportunities to mitigate. When we have a long-term forecast of drought, the government can direct its drought mitigation activities like availing irrigation, creating the right crop cycle and also prescribing crops suitable for low rainfall. India has long experience of managing drought but it is time it evolves its long-term forecasting mechanism which, as demonstrated, is a possibility.
We are a voice to you; you have been a support to us. Together we build journalism that is independent, credible and fearless. You can further help us by making a donation. This will mean a lot for our ability to bring you news, perspectives and analysis from the ground so that we can make change together.
.png)
Comments are moderated and will be published only after the site moderator’s approval. Please use a genuine email ID and provide your name. Selected comments may also be used in the ‘Letters’ section of the Down To Earth print edition.