Predicting and monitoring drought in the humid tropics: a case study on Sri Lanka

This dissertation develops an operational tool for predicting and monitoring drought applicable to the humid tropics. Using Sri Lanka as a case example, it examines whether droughts in the humid tropics are predictable on an operational basis, and investigates how moisture stress may be monitored as a season unfurls. Droughts in Sri Lanka occur when rainfall during the main cultivation season – the Maha (October-March) – fails. Such droughts profoundly impact rice production. From 1951-2008, there were 4 extreme [Standardized Precipitation Index (SPI) <-2.0], 1 severe (-1.9<SPI<-1.5), 5 moderate (-1.49<SPI<-1.0) and 4 mild (-0.99<SPI<-0.5) droughts. Maha droughts can be operationally predicted by forecasting the failure of the two rainfall regimes during the season. The contemporaneous westerly zonal wind at 850hPa (U850), over 60⁰E-105⁰E and 5⁰S-15⁰N, controls the strength of the October-November convective rains – with rain failure associated with anomalously strong U850. The contemporaneous northerly vertical shear of the mean meridional wind ( ), over 80⁰E-90⁰E and 0⁰N-20⁰N, controls the strength of the December-February northeast monsoon rains – with rain failure associated with an anomalously weak Vs. Drought forecast skill was assessed for 1981-2002 using predicted fields of U850 issued in September, and Vs, issued in November, from the NCEP Climate Forecast System and the ECHAM4.5 forced with two scenarios of prescribed sea surface temperature (SST) anomalies. October-November rain failure can be predicted with good skill over the rice cultivation areas in the central and southeastern regions using forecast U850 from the two versions of the ECHAM4.5. December-February rain failure can be predicted with good skill in the rice cultivation areas in the eastern, central and north central regions with forecast Vs from the ECHAM4.5 forced with constructed analogues of SST anomalies. The utility of the Vegetation Temperature Condition Index (VTCI) – calculated with Terra-MODIS Land Surface Temperature and Normalized Difference Vegetation Index products – as an indicator of abnormally wet or dry conditions was tested for the Maha season. Results show that the VTCI is a suitable metric for the near-real time monitoring of Maha drought because it captures the onset and progression of moisture stress as the season unfurls and complements the seasonal rainfall forecast.Ph.D.Includes bibliographical referencesIncludes vitaby Dinali Nelun Fernand