Figure 1: Mean monsoon-season (June-September) precipitation (mm) for South Asia, 1901-2016. Land grid cells where precipitation was under 100 mm are left unshaded and are not included in the study domain.
Densely populated South Asia, a leading agricultural area, derives most of its water supply from intense summer rainfall (Figure 1). The rainy season is closely linked to the South Asia Summer Monsoon (SASM), in which moisture flows to the Indian subcontinent from the Arabian Sea and the Bay of Bengal, driven by the land-sea temperature contrast. Interannual variation in the amount and spatial distribution of summer precipitation in the SASM monsoon zone is considerable, with severe consequences for agricultural productivity and human well-being. Prediction of monsoon precipitation has therefore been pursued for over a century, using several predictors such as snow cover and air pressure patterns.
With average precipitation of 35 inches per four-month season over an area encompassing most of the Indian subcontinent, the South Asia summer monsoon is intense, only partly understood, and notoriously difficult to predict. A frequent visitor to the region, stretching from Nepal to Sri Lanka, Krakauer has devised a methodology that allows forecasts potentially up to a year in advance. Currently, most predictions are made about two months in advance of the South Asia monsoon season that runs from June to September, but it is not known how far ahead skillful forecasts might be possible.
"People usually use one or two predictors for forecasts," said the Grove School of Engineering associate professor who is also affiliated with the CCNY-based NOAA-CREST. Many of these predictors are one or another pattern of sea surface temperatures. His question was how do you find which patterns are important for forecasting the monsoon—the amount of rain and where it will be?" Unlike other forecasters who use only the sea surface temperature readings from neighboring waters, Krakauer looked at the predictive potential of all the common patterns in the sea surface temperature map. He developed prediction methods using global sea surface temperature and monsoon precipitation data from between 1901 to 1996, and tested the performance of his prediction methods on data from 1997-2017. The results show that two methods seem to do a good job of forecasting the monsoon. Sea surface temperatures at the beginning of the monsoon as well as four years back were analyzed.
His finding was that, generally, the closer to the beginning of the monsoon season, the more accurate forecasts that are based on sea surface temperature can be. But predictions with some accuracy can be made as far as a year in advance. Getting a better sense of how much water will be available is particularly important given that the rainfall is getting more intense in South Asia while the total amount remains constant, meaning that more rain is falling in a shorter period. This could be problematic for farmers in the region.
By considering linear and nonlinear forecast methods using SST modes as predictors, prediction skill for spatially distributed monsoon-period precipitation over South Asia was found to decay with a timescale of 5-12 months, but with residual skill at several-year lead times due to long-term climate trends. While these methods are not definitive and could likely be further improved, the present findings suggest that South Asia monsoon period precipitation can be predicted with longer lead time than the subseasonal to seasonal leads usually attempted now.
Nir Y Krakauer. Year-ahead predictability of South Asian Summer Monsoon precipitation. Environmental Research Letters, 2019; DOI: 10.1088/1748-9326/ab006a
To learn more about other climate-related stories occurring cross Asia, be sure to click here!
© 2018 Oceanographer Daneeja Mawren