Perils of predicting precipitation
Despite all the sophisticated analyses performed on super computers, our ability to predict the behaviour of the monsoon has not improved at all
Rains have brought tears to the eyes of people in two distant parts of the world for two different reasons: Their overabundance in the Northeastern United States in the wake of an even weakened Hurricane Ida and their absence in most of India. Both have been unanticipated and we have had explanations such as global climate change in one case and unexpected behaviour of phenomena such as Madden-julian Oscillation (MJO) and Indian Ocean Dipole (IOD) in the other (see Sanjeeb Mukherjee, Business Standard, September 3, 2021). What they signify is how fraught with uncertainty predicting precipitation is, notwithstanding sophisticated multi-variate models and super computers to process them.
I have been writing about these issues for long. In the drought of 2009, I had a series of such articles in this paper. I discovered MJO on my own and was the first columnist to mention it as an explanatory factor then. Essentially, it means our monsoon moves eastward. I called it as China having stolen our rains, a phrase that found much currency then. I found then, as I do now, our concern with predicting season-long rainfall as illogical and serving little purpose. But despite sound reasoning, I find that our interest in such forecasts has grown over time. Earlier, only the Indian Meteorological Department (IMD) made monsoon predictions; now a slew of private agencies, too, make them. Our fascination with these almost matches that with Budget numbers, which the common people understand as little as they do the monsoon figures. So here I go again.
Five decades ago, the father
of Indian climatology, the late Professor P R Pisharoty of the Physical Research Laboratory, Ahmedabad, said in a lecture that longer period predictions mean little, as weather itself makes up its mind no more than three weeks in advance. I was intrigued by this remark. Over the next several weeks of conversations with him, I learnt that what passes for long-term monsoon forecasts is merely associative in nature.
There is no causeand-effect relationship involved in such predictions. Notice that even now, MJO and IOD as also the more common El Niño phenomenon (or its counter, La Niña) are brought out as causes only after the fact. Their occurrence is not known for sure before the narrow window of time Pisharoty had talked about and cannot be truly used as a predictive factor. We can only say that if MJO occurs, or if El Niño is active, it may impact the monsoon thus. Even then, we are not in
a position to correlate quantitatively the result — the effect on precipitation — with the presumed cause.
The short point is that despite all the sophisticated analyses performed on super computers, our ability to predict the behaviour of the monsoon has not improved at all. Pretty much the same situation confronted the more sophisticated weather analysts in the United States as it was completely unable to forecast the movement of the hurricane Katrina in 2005 (much to the discomfiture of the then President George W Bush) and now her sister Ida, especially as she veered around and hit the Eastern seaboard states.
The other point that has troubled me is that the forecasts are about the averages of the entire monsoon and the whole country. Of late, IMD has begun providing its breakups by the five major divisions and by calendar months. But I believe that such regions and time periods are still too big to be of much use to end-users and policy makers. The concept of averages has implicit in it the possibility of compensating departures from the mean figure. For this to be valid in case of rains, we need to believe that the areas or periods of shortages would be made up by surpluses elsewhere or at other times. That is simply not possible, unless we have large storages and distribution networks in place. That is the case now only with the very large Narmada basin, its dams and canals. (That is why, living in Gujarat, I follow keenly the rains in the Narmada catchment area, for what is now our lifeline.) Elsewhere, this is an impossibility. Floods in Konkan will just run-off to the sea; the resulting average precipitation even in the western half of Maharashtra does not bring succour to the thirsty lands or people of the rainshadow areas east of the Sahyadri mountains. For a continental country whose agriculture and people depend on a four-month long-monsoon for all their moisture and drinking water needs, averages could be a cruel joke.
The Ida debacle in New York City should also silence our media critics about Mumbai’s unpreparedness for monsoons. If the urbs prima of the world’s richest country cannot handle 70-80 mm of rain in one day — we saw pictures of torrents of water rushing down the subway entrance stairs — can we really expect our own megalopolis by the sea, overcrowded and underfunded, to be prepared for inundation of 200-300 mm of precipitation in a day? The harassed Mumbai civic administration would say, give us 7080 mm of rain any day and we will handle it like breeze!