Predicting monsoon averting famine

Gilbert Thomas Walker was famed as a brilliant mathematician, evident by his lectureship at the illustrious Trinity College. His thesis related to electricity and magnetism was read in The Royal Society. His qualifications ranged from mathematics to music (he was an accomplished flautist); from water colour painting to ice skating and mountaineering; from ornithology to study of the dynamics of African folk toy-boomerang—all except meteorology. So, when in 1903, John Eliot chose him to be his successor to the post of chief meteorological reporter to the government of India and director-general of Indian observatories, Walker would have appeared a most unlikely candidate. We have no way of knowing what Eliot saw in Walker, but it was worth a gamble. Walker not only made an indelible mark in Indian monsoon research, but also changed the way meteorology was practiced.

It is not the famed English obsession with weather that made the colonial administration in India to give importance to meteorology, but politics. Weather and weather prediction was politically emotive. The colonial rule witnessed recurrent famines—there were 18 famines between 1875 and 1900 and the Indian Famine Commission of 1880 had come to the conclusion that poor rains resulting from failure of monsoon is the root cause of the famines. Thus, the Famine Commission solicited the colony’s first meteorological reporter, Henry F Blandford, in forecasting the variations in the monsoon rains.

The famines of 1877 and 1899 were horrendous. Coupled with failure of monsoon they resulted in widespread starvation and large number of deaths. It was also the period of the birth of Indian Nationalism. Perhaps in the past no one paid much attention to the famines, content to remain passive, blaming fate for the misfortune; but the incipient national movement projected the colonial government’s ruinous economic policies as the root cause of starvation deaths. Political agitations by the nationalists impelled the colonial government to act; meteorology was reorganised and the Indian Meteorology Department was established.

Seeking to predict Indian summer monsoon, Blandford initially speculated that the local factors such as amount of snow in the Himalayas, the summer heat or winter cold in central Asia and Tibet, and storms in Persia influenced its quantity and quality. He said by collecting these data and studying for relationships, we can find a way to predict the monsoon. However his forecasts were not successful; so were the monsoon prediction model adopted by Eliot, his successor. Repeated failures in prediction not only resulted in failure to prevent starvation deaths, but also severely embarrassed colonial government.

It is in this context Walker was invited to take the position as the chief meteorological reporter to the government of India. His task was daunting; he had to somehow make forecasts about the monsoon using the data being collected from various weather stations, both in India and world over. He had to succeed where two of his predecessors had utterly failed.

Photo: Sayantoni Palchoudhuri

Blandford had observed a curious fact—during the monsoon failure of 1877 and 1878, the atmospheric pressure over the countries surrounding the Indian Ocean had been abnormally high. To compare this abnormal situation, Blandford requested information on atmospheric pressure conditions at this time from other meteorological observers around the world. When the request from Blandford landed at his table, Charles Todd of South Australia was intrigued. Australia also had experienced in 1877 and 1888. Todd concluded that “there can be little or no doubt that severe droughts occur as a rule simultaneously over the two countries.” This tendency for Indian and Australian droughts to occur at the same time was one of the many “teleconnections” that the meteorologists were speculating at that time.

But doubts persisted; were these “teleconnections”—relationship across different parts of the globe— mere coincidences? Walkers’ statistical approach provided an unambiguous answer. By using the tools of statistics, specifically “regression” and “correlation”, Walker ventured to make sense of tonnes of raw weather data collected all over the world by the British imperialists. Further, he ventured to do things a conventional meteorologist would have balked at; he argued, if Blandford’s local factors are not responsible, then perhaps distant factors influenced the outcome of the monsoon. In his perception, meteorology was not seeking grand theory, but was a colossal problem of statistics.

In addition to usual weather parameters such as rainfall, pressure and temperature, he also gathered unconventional data such as river flood stages, mountain snowpack depths, lake levels and sunspot activity for his analysis. He obtained 40 years past records of these data from whatever weather stations he could mobilise. Until then meteorologists would average the year’s rainfall or temperature and get a “figure” to work with. But Walker changed this. He realised seasons are too important to be lost sight of and hence averaged the data over seasons. Thus for every year he obtained four values for each parameter. For example, he took rainfall, a variable and averaged it over a season for a particular station and obtained four-time series variable for each station. In like manner he got average of all other factors for every season and grouped them. By using the statistical tool of coefficient of correlation, he attempted to find if any two phenomena were varying in unison over a series of years across the weather stations located at random places. He also tried to find if one factor in one location could be impacting another factor in another season Essentially, his method implied that if the correlation between two phenomena was 1.0 then both were strongly related and so by observing one could predict another. On the other hand, if the correlation is 0 then both phenomena are unrelated. But if the correlation coefficient is -1.0 then it means both phenomena are related in a seesaw pattern; occurrence of one would imply non-occurrence of another.

To undertake this huge task of comparing immense information, the data were to be first complied, compared and analysed. The tedious tasks were performed by a battery of “native” human computers, recruited in hundreds. These assistants computed correlation coefficients for every set of two phenomena, however seemingly they appear to be unrelated. It was in this process of searching the proverbial needle in the haystack that Walker hit upon unique, but puzzling correlations. The barometer readings from the weather stations of Tahiti in eastern Pacific and the Darwin, Australia in western Pacific showed a remarkable -1.0 correlation. That is when barometer pressure increased in the east, it usually fell in the west, and vice versa. Based on this “seesaw” effect of atmospheric pressure, he coined a term “Southern Oscillation” to describe it. His statistical analysis showed that occasional failure of the monsoons in India often coincided with low pressure over Tahiti, high pressure over Darwin and relaxed trade winds over the Pacific. This condition also affected the rainfall in Africa and the temperatures in western Canada. Walker noted that these phenomena, spread around the world, were interrelated.

The Southern Oscillation was significant for Indian monsoons. Southern Oscillation, Walker said, “Implied high pressure in the Pacific and South America is associated with low pressure in land round the Indian Ocean, with low temperatures in tropical regions and the centre of North America, and with abundant rain in India, Java, and Australia, and high Nile floods, while the rainfall is scanty in Chile.” If the Southern Oscillation was abnormal, then it implied poor Indian summer monsoon. The Southern Oscillation occurred about six months before the onset of Indian summer monsoon. “Here was the key to forecasting the monsoon”, he said.

Walker was, however, unable to provide any physical mechanism governing these oscillations. He speculated ocean circulation and temperature may play a role but could not demonstrate it on account of paucity of data. In his time, he was not that much appreciated, for his results were seen as outlandish. That weather in India could be influenced by occurrences half way across the world mystified contemporary meteorologists.

After about 50 years of the discovery of the Southern Oscillation, the physical mechanism that causes it was discovered by Norwegian meteorologist J Bjerknes in 1969. Periodic warming of the east Pacific Ocean called El Nino, and the complex relation it has to the sea surface temperature and the atmosphere above the equatorial Pacific, Bjerknes showed, resulted in the Southern Oscillation. He also found that due to these intricate interactions, air in the eastern equatorial Pacific Ocean sinks, while the air in the western equatorial ocean rises up resulting in a circulation of air over Pacific Ocean. This circulation was named after Walker.

Any fluctuation in this circulation has a significant bearing on the Indian summer monsoon rainfall.