Its forecasts aren’t much help given India’s diverse agro-climatic zones, but there’s only so far meteorology can go
YOGINDER K ALAGH
Chairman, Institute of Rural Management, Anand
“The Met models are still the best — not always completely accurate in their forecast, they are the best all the same. As far as I know, the counterfactual does not exist”
The debate is, in a way, a no-brainer. As former president of the Indian Econometric Society, I can definitely say that unless you are a charlatan or an astrologer, no prediction is absolutely reliable. The real question is a different one: if you are not a believer in astrology and havans determining the weather, does the India Meteorological Department’s (IMD’s) monsoon forecast reduce the area of your ignorance in the best possible manner available? That, I believe, is correct. I believed that even while in the Planning Commission as its agricultural member I was mandated by Rajiv Gandhi to fight the great droughts of 1987 and 1988 and the IMD’s models were, to put it mildly, of little help.
In those days, the IMD was working with what would, at best, be called an associational model using 16 variables, which were, according to statistical analysis, correlated with monsoon rainfall in India. These included the North Atlantic temperatures and, of course, the famous El Nino. Given the values of these variables, the average level of rainfall as well as an associated variance was predicted. Soon thereafter the great game started. Many “experts” claimed to do better. As a student of statistics, I tracked them and every time I went back to government and forced a serious review. There was no evidence that anybody else was doing better on a systematic basis.
Meteorology is like econometrics, even medicine — a stochastic science. Its models are not exact. On the basis of information and models developed from data, it forecasts. For the next six hours they may be quite precise, but not so much through time. We must develop the database and the tools to use it.
We, as a country, complain vociferously about the outcomes but are miserly when it comes to supporting the science, its database and those who manage it. After the MONEX, very little work concerning the physical phenomenon was done on the monsoon in spite of all the talk of it being our lifeline.
Developing the first agro-climatic model for agriculture in India in the late eighties, I argued for the medium-term weather prediction model. We had a good 48-hour forecast and by then the monsoon model as a whole, admittedly a simple one, but the farmer wanted to know what could happen in the next week or 10 days.
Rajiv Gandhi got the Americans to give us a supercomputer for that purpose, but soon they proposed restrictions no sovereign country like India would accept. We then developed our own PARAM; the rest is history. The IMD is now in the process of the operational use of the medium-term decentralised forecasts.
In the mid-nineties as its minister, I encouraged the IMD to go to Parliament with its forecasts and defended it there the way I do now. I encouraged it to experiment more and made the department part of the Indo-US Technology Forum. This later led to movement to the so-called structural models — while not quite fully there, the so called “ensemble” model is methodologically a definite improvement.
More recently, as I chaired a CII Expert Group on the monsoon, conviction dawned again that the Met models were still the best — not always completely accurate in forecast but the best all the same. To the best of my knowledge, the counterfactual does not exist.
It warmed the cockles of my heart to read that a Rs 400-crore Monsoon Mission has now been accepted as a 12th Plan programme by the Planning Commission. I hope this mission is not shelved like many other good ideas, given that Professor Rangarajan’s Committee has suggested that Plan programmes should be abolished.
Spare a thought for the statistician, meteorologist and others doing difficult jobs trying to push back the area of ignorance with the dim light of knowledge. It is a difficult job that needs support. For my meteorologist friends, don’t get into the game of showing the rains will improve. The farmer’s eye in the sky gets betrayed. Let the politician and bureaucrat do that. They say the trader will benefit from “bad” forecasts. We know from Dantwala and Pavaskar’s classic studies of markets that the trader has his own sources of information. For you, my friend, “Satyamev Jayate”.
P CHENGAL REDDY
Secretary General, Consortium of Indian Farmers Associations (CIFA)
“it is not correct to expect the IMD to give 365 days’ advance predictions for a large country with 130 million hectares of agricultural land and highly divergent agro-climatic conditions”
Long-term monsoon forecasting is not only a complex activity, but it is also an extremely difficult task. In the past 25 years, the predictions of the India Meteorological Department (IMD) about the accuracy of the monsoon proved correct only in 10 years.
But the bigger issue is that it is not correct to expect the IMD to give 365 days’ advance predictions for a large country with 130 million hectares of agricultural land and highly divergent agro-climatic conditions. Instead, the IMD’s monsoon forecast should be separate for agro-climatic zones. A general prediction for the entire nation is neither possible nor practical.
Being a large country, India experiences 10 to 15 per cent variations in rainfall every year. The rainfall has been deficient in six of the last 10 years. The IMD’s figures show that the rainfall was -1.7 per cent in 2008 and -21.8 per cent in 2009.
Region-wise rainfall was -20 per cent in the north-east and -10 per cent in the north-west in 2005. In 2006, it was -15 per cent in the north-east. In 2009, it was -36 per cent in the north-west, -20 per cent in central India and -23 per cent in the north-east. In 2010, it was -18 per cent in the north-east. This uneven pattern has been noticed at regular intervals for 50 years.
In May-June 2012, the IMD predicted a normal monsoon. Based on it, 10 million farmers in rain-fed areas sowed Bt cotton in Maharashtra, Andhra Pradesh and Gujarat. Now that the rains have failed, these farmers don’t know what to do.
What is expected of the IMD is a prediction for each agriculture zone. This will help farmers save high investments on risky crops like cotton, and opt for minor millet or fodder crops .
Short-term forecasting in a compact area of 10 kilometres with three to five days’ advance notice would also help farmers keep away pests and diseases and bring in timely harvests.
In India’s drought-prone areas, monsoon variations normally follow the pattern of one good year, one average year and two bad years. This is where the IMD can play an important role by suggesting water conservation measures, cropping patterns and so on. In the old days, farmers used to adjust to weather variations by planting multiple crops (Nava Dhanya) which met their basic food requirements and provided fodder for their cattle. In recent years, however, the commercialisation of agriculture has encouraged farmers towards mono-cropping; it is this trend that is causing serious problems in times of deficient or no rains.
The responsibility for the IMD’s poor performance lies squarely with the government. In 2006, the National Commission on Farmers, chaired by M S Swaminathan, made suggestions for upgrading and enlarging the IMD’s activities. In 2007, the government announced a national policy for farmers and prepared a plan to establish a project for agriculture output using space, agro-meteorology and land-based observations. The broad suggestion for the IMD was that it would issue regular bulletins at the sub-divisional level.
None of these recommendations has been implemented. The government allots a meagre Rs 350 crore a year for the IMD, which works out to Rs 3 per hectare. Contrast this with the billions allocated to space research such as sending a man to Mars, and developing Inter-Continental Ballistic Missiles. The government does not consider weather forecasting and its impact on millions of small farmers an important issue.
Although the IMD’s long-term monsoon predictions are not accurate, it has made remarkable progress in forecasting cyclones in recent years. Experience gained in the past 20 years by the state governments has helped keep in place a daily weather forecasting system that has saved the lives of fishermen and coastal populations.
The government’s priorities for the IMD’s key sectors require urgent attention. It is now bringing in communication technologies and nuclear energy technologies. Weather forecasting technologies are available across the world. The government must immediately focus on an accurate weather forecasting programme by getting the necessary technology and equipment. It should also develop manpower and use existing communication channels such as TV, radio and mobile phones to reach every Indian.
