M. Mohapatra, director general, IMD, accepts the misses over the past few years but says the department has been working to improve its forecasts, especially for the monsoon.
He explains that forecasting the monsoon in India is trickier compared to other weather systems.
“The Indian monsoon is a large-scale process and a lot of local reasons play a role in the final rainfall outcomes,” says Mohapatra.
The Met chief adds that while the monsoon prediction itself is challenging, issuing long-range forecasts for such a dynamic season is a double whammy for weathermen.
“The accuracy of medium and short-range forecasts, which are usually issued five to three days in advance, is more reliable,” he says.
Also Read: The El Niño weather phenomenon is weakening. How this will impact Indian monsoon in 2024
A look at IMD’s performance
The long-range forecasts issued by the Met over the past three years have largely been accurate. The same, however, cannot be said about the IMD’s past performance.
In 2021 and 2022, the Indian forecasters accurately predicted a “normal” monsoon but narrowly missed the mark with the 2023 forecast. Long-range forecasts issued between 2011 and 2020 showed that only two of the 10 predictions were accurate — for 2011 and 2017. In one of the years, the forecast majorly missed the mark with a deviation of over 10 percentage points.
Going back another decade, the weathermen made four accurate monsoon forecasts but missed the mark for other years by a long shot.
In two years, 2002 and 2009, the margin of deviation between long-range forecasts and final rainfall received was nearly 20 percentage points. In 2004, the deviation was as much as 13 percentage points.
Problem areas
Curiously, many experts believe that models from the US and the UK are often not the most reliable in the Indian context. These are designed for forecasting a more straightforward temperate climate and are not meant for unpredictable tropical climate conditions like India’s.
Despite calibration, there is a certain error margin that these foreign models carry.
Madhavan Rajeevan, former secretary, ministry of earth sciences (MoES), says even the most world-class forecasting models have limitations and are only a part of the larger process of forecasting. The interpretations made by forecasters ar a more important part of issuing accurate forecasts, he says.
“We can have the best technology in the world, but unless we have trained and experienced forecasters managing these models, the predictions will have problems. But I have to mention that IMD’s forecasting has improved significantly over the years. There is scope for more,” says Rajeevan.
Some experts also emphasise that the success or failure of a forecast cannot be limited to a single number. The focus of the weathermen should be to provide location-wise forecasts, with spatial maps detailing distribution.
To explain this, the IMD this year has forecast an “above normal” monsoon for India. The rainfall recording for the entire country is predicted to be 106 percent of the long-period average. Beyond the error margin of 5 percent set by the IMD, any deviation would be considered a miss.
Akshay Deoras, research scientist at National Centre for Atmospheric Science, University of Reading, UK, says spatial maps are a better presentation of monsoon forecasts for people. These maps, ideally quantifying how much rainfall is likely over which areas in a monsoon season, would be better consumed by stakeholders and would be easier to understand.
“In addition to issuing the monsoon forecast as pure numbers, emphasis needs to be on issuing region-specific forecasts of seasonal monsoon rainfall and conducting a robust assessment of the forecast accuracy. The forecast for the entire country could be ‘above normal’ rainfall, but that does not always mean every state would (have above normal rain). Besides, current maps issued with the long-range forecasts show only probabilities of rainfall categories instead of quantifying rainfall, which is important if a region is expected to receive above normal or below normal rainfall,” he said.
Plans to improve forecasts
A senior official from the IMD told ThePrint that lately, the department has also been using indigenous models from the Indian Institute of Tropical Meteorology (IITM) to improve forecasts.
In May last year, Union minister of earth sciences Kiren Rijiju announced plans to acquire 18 petaflop supercomputers in IITM, Pune, and National Centre for Medium Range Weather Forecasting, Noida. These supercomputers, he said, would help the IMD provide more localised and high-resolution forecasts.
Speaking to The Print, MoES secretary M. Ravichandran said the network of observatories was also being expanded. More data would ensure that the forecasts are more accurate.
“There are certain parts of the country such as the North-East where there aren’t enough observatories. We are working to improve that. In the coming years you will see improvements,” he said.
Issuing long-range monsoon forecasts might be a tough task but in India it is also essential. It is also imperative that regular, updated forecasts are released as the season progresses.
Mahesh Palawat, vice-president (meteorology and climate change), at Skymet Weather, a private weather forecasting service, explained that the April forecast is only to give out rough trends for the season because when models start giving out more developed indications, around May-end, most monsoon-dependent activities have already begun.
“If you only issue a forecast in May, it will be more accurate, but it will be of no use. Currently, the IMD has the most advanced forecasting technology in the country. No private forecaster is a match,” he said.
What goes into predicting the monsoon
Domain experts explained that three major factors are studied to assess monsoon behaviour in India.
The first is to analyse El Nino and La Nina conditions, which are two phases of the El Nino Southern Oscillation (ENSO) phenomenon that occurs in the equatorial Pacific Ocean. The El Nino phase is marked by warmer temperatures, while La Nina refers to a cooling period.
The other factors that impact the southwest monsoon patterns are the Indian Ocean Dipole (IOD) — a climate pattern affecting the Indian Ocean — and the assessment of snowfall in the Eurasian region. A favourable IOD results in good monsoon rain, but snowfall in Eurasia and the Northern Hemisphere is known to inversely impact monsoon rains in India.
This time, the IMD forecast has spelt out that El Nino conditions will cease by the first half of the monsoon season. This will improve the chances of high rainfall by the second half of the monsoon (August-September). The recording of less snow in Eurasia also hints at a greater possibility of a favourable monsoon in India.
A second senior official from IMD, who did not wish to be identified, told ThePrint that two long-range forecasts are issued for the monsoon season every year. The April forecast only spells out initial trends. By mid-May, a clearer picture of the season is given out.
“When the monsoon trends are examined in March-April, there are certain spring barriers that can occur. That is why we caveat this forecast with having around 60 percent accuracy. Since India is an agriculture-dependent country, everyone wants to know the monsoon trend. All aspects of the country’s functioning are dependent on the monsoon, including government policies, interventions and the stock market,” the official said.
(Edited by Nida Fatima Siddiqui)
Also Read: India saw 329 heatwave days in 2 yrs, media coverage lukewarm — 2 reports highlight dual challenge