How forecasters got their monsoon predictions wrong, again
Forecasters, both government and private, know that any update on the monsoon can lift or dampen public sentiments, shape economic policy, and have acknowledged gaps in forecasting
India got its monsoon forecast wrong this year, consistently.
To begin with, India Meteorological Department (IMD)’s May 31 forecast for monsoon onset over Kerala had raised hopes of early showers across the country. But this was revised, at the last minute on May 30, to declare the onset of monsoon for June 3.
IMD, on June 11, forecast that monsoon would advance to remaining parts of the country outside south Rajasthan and Kutch during the next six to seven days. But the northern limit of monsoon has been stationary, passing through Diu, Surat, Nandurbar , Bhopal, Nowgong, Hamirpur, Barabanki, Bareilly, Saharanpur, Ambala and Amritsar since June 13.
Last week, IMD had forecast that monsoon would make onset over Delhi by June 15, around 12 days ahead of its normal onset date. But on Tuesday, June 15, IMD scientists said onset over Delhi may take another seven to 10 days because of a trough in the westerlies that had weakened the monsoon flow over northwest India.
Also Read | Conditions unfavourable for monsoon’s progress to parts of N-W India
Forecasters, both government and private, know that any update on the monsoon can lift or dampen public sentiments, shape economic policy, and have acknowledged gaps in forecasting. The fact that monsoon itself is a very complex system because of its interactions with various land and oceanic features makes overall forecasting during the season challenging, according to forecasters.
The explanation, the critique
On Wednesday, IMD, in its bulletin, said a western disturbance was affecting the Western Himalayan region and a cyclonic circulation is lying over east Uttar Pradesh. These large-scale atmospheric conditions are not favourable for further advance of monsoon in Rajasthan, remaining parts of Gujarat, Punjab, Haryana and Delhi. However, it said that there could be slow monsoon progress over parts of Uttar Pradesh.
Experts said forecasters should have identified an impending trough in the westerlies well in advance based on models and should have withheld the monsoon onset forecast for Delhi. Monsoon onset dates are particularly important for northwest India because of the relatively low average rainfall received in the region following a gruelling summer.
“Its not challenging to forecast or see a western disturbance approaching. They (IMD) probably did not anticipate the possibility. I am not sure what exactly may have happened,” said M Rajeevan, secretary, ministry of earth sciences who had said that monsoon could slow down in its progress towards the end of the month.
But forecasting heavy rainfall events during the monsoon season is a challenge. “Heavy or extremely heavy rainfall events during the monsoon are less predictable. Monsoon lows (low pressure area), monsoon depressions are relatively difficult to predict because of the complex nature of the monsoon. It’s not impossible. The shorter the period of forecast, accuracy is higher,” explained Rajeevan.
IMD forecasters may not have anticipated the change in track of monsoon due to strengthening of westerlies. “The low pressure area which had formed over Bay of Bengal last week and is responsible for the fast progress of monsoon up to northwest India suddenly changed track. Its effect remained more concentrated towards Uttar Pradesh, Jharkhand, Bihar etc and did not progress towards Rajasthan. These changes can be monitored two to three days in advance and IMD has alerted people accordingly in the change of onset dates,” said DS Pai, scientist and head, climate research and services, IMD Pune.
Additionally, while there are a set of statistical parameters that are followed for monsoon onset over Kerala, in other parts of the country, monsoon onset is rainfall driven. “We see if rain has been recorded and if there are chances of rain sustaining for a couple of days,” Pai added.
Inadequate focus on wind patterns in monsoon forecast may be a factor in forecasting lapses, said Akshay Deoras, an independent meteorologist and PhD student at the University of Reading in the United Kingdom.
“There is a specific set of criteria to declare the monsoon’s onset over Kerala, which includes a combination of the observed winds, intensity of clouds and rainfall. However, very often, rainfall is given more importance over winds to declare the monsoon’s progress in other parts of India. Traditionally, rainfall was given a lot of importance since upper-air observations were limited. However, we now have better upper-air observations as well as wind forecasts. Whilst the monsoon is progressing in June, we may still get rainfall due to non-monsoon systems such as western disturbances. In such cases, wind patterns help in identifying such events, resulting in improved predictions of the monsoon’s arrival in a region,” he said.
Deoras added that in case of Delhi’s forecast, the wind pattern should have been considered well in advance. “Wind patterns are considered to be more predictable than rainfall. For the last ten days, various weather models were showing a western disturbance blocking the monsoon’s progress over northwest India around mid-June. In such cases, a forecaster’s judgement is very important.”
Why others got it wrong
Skymet Weather, a private weather forecasting company announced monsoon onset over Kerala on May 30 while IMD announced onset on June 3. How could there be such a large disparity?
“When we announced onset over Kerala, our criteria for rainfall and monsoonal winds were met but not the criteria for outgoing longwave radiation (OLR). We still declared monsoon onset because often a copybook style onset is not possible,” said Mahesh Palawat, vice president, climate change and meteorology, Skymet Weather. OLR represents the total radiation going to space emitted by the atmosphere or extent of cloudiness.
IMD had said on May 28 that monsoon was likely to make an onset over Kerala on June 31, a day ahead of its normal date. But on May 30, IMD said that monsoon parameters are not being met yet. According to IMD’s monsoon onset parameters, after May 10, 60% of the available 14 stations enlisted — Minicoy, Amini, Thiruvananthapuram, Punalur, Kollam, Allapuzha, Kottayam, Kochi, Thrissur, Kozhikode, Thalassery, Kannur, Kudulu and Mangalore — would have to report rainfall of 2.5 mm or more for two consecutive days, depth of westerlies should be maintained upto 600 hPa, Outgoing longwave radiation (OLR) value should be below 200 wm-2 ( watt per sq metre).
“@Indiametdept will never manipulate data or make statements to justify their forecasts. They admit forecast failures with humility. In weather & monsoon forecasts, no one can be perfect. We are also accountable to Indian taxpayers. We show respect to our country,” tweeted M Rajeevan, secretary, ministry of earth sciences on May 30 after IMD’s sudden retraction of May 31 onset over Kerala.
Skymet Weather also did not anticipate the sudden lull in monsoon before it makes an onset over Delhi. “Models are more accurate for three to four days. Thereafter their accuracy reduces, which may be the case this time in forecasting onset over Delhi. The westerly component of winds was not prominent earlier. People in north India wait eagerly for monsoon rains, they wait desperately for rain to cool down the region so onset dates are important to people,” added Palawat.
The next challenge
The real test for forecasting is yet to come. During the monsoon season, there are sudden extremely heavy rainfall events that are difficult to forecast and cause localised disasters. Summer monsoon rain over India has declined by around six per cent from 1951 to 2015, with notable decreases over the Indo-Gangetic Plains and the Western Ghats according to Assessment of Climate Change over the Indian Region, a report by the ministry of earth sciences released last year.
There has been a shift in recent years toward more frequent dry spells and more intense wet spells during monsoon. Over central India, the frequency of extremely heavy rain with intensities exceeding 150 mm per day increased by about 75% during 1950–2015.
“Forecasting heavy rainfall events during the monsoon season still remains a challenge despite recent advancements in numerical weather prediction. Things become more challenging when the monsoon flow interacts with mountains as well as monsoon low-pressure systems,” said Deoras. He pointed out the challenges in regional weather predictions for a region such as Mumbai; the recent heavy rainfall event in the city on June 12 was not picked up by weather models well in advance.
India receives about 70% of its annual rainfall during the four-month monsoon season — this is crucial for the country’s farm-dependent economy and for rice, soybeans, and cotton cultivation. A normal monsoon this year will significantly help the agriculture sector. Good rains have been a prime reason for the farm sector’s resilience for two years despite the pandemic. India has over 150 million farmers and nearly half of Indians are dependent on a farm-based income. As much as 60% of India’s net-sown area does not have access to irrigation.
And that is why the stakes for getting Met forecast right is high. 2021, once again, offers lessons.