Extended Probabilistic Forecasts of Heat Stress

Even if there were to be no changes in climate, heat stress lurks as one of the major instigators of weather-related mortality and morbidity. This is especially true in the less-developed world where infrastructures and health services are often stressed and mitigation strategies for avoiding heat common to developed countries (e.g. air conditioning, shelter, etc.) limited. But the more-developed world is not immune to heat stress and there are numerous examples of increased morbidity and mortality such as Western Europe in 2003, New York City in 2000, Russia in 2010 and Australia 2012. With global warming, exposure to higher temperatures may increase the levels of heat stress in both the less- and more-developed worlds, but currently vulnerability to heat stress is less well managed in lower resource settings. We consider heat stress from a very fundamental manner noting that the ability of a body to cool is a complicated function of both ambient temperature and humidity. Using the Indian monsoon as an example, there are two major periods of heat stress exhibiting humidity's importance to diurnal heat stress. The first is the time leading up to the monsoon rains when temperature, the main driver, is very high often > 45°C but humidity is low, at least allowing some recovery during the nighttime. The second is during the lulls (dry periods) in the monsoon where the temperature is near 40°C and the humidity is very high. During these periods there is no respite during the night and heat stress can often remain at critical levels during the entire 24 hour period.

However, heat stress is predictable out to 7-10 days in the future and predictions can facilitate management plans. We show an example of the use of statistically adjusted maximum temperature forecasts in Ahmedabad, Gujarat , in northwest India using the ECMWF VarEPS. These forecasts were provided to public health authorities allowing the issuance of warnings to the populace and increasing health services wherever possible. We also show examples of the extended version of the prediction scheme where the critical diurnal variability of heat stress, dependent on both humidity and temperature, is forecast out to 10-days that will be added in future to the forecasts to further refine heat management planning. Experimental results of probabilistic heat stress forecasts for 30-day lead-time are also shown.

The system developed and illustrated here was a collaborative effort supported by a consortium including the Georgia Institute of Technology, the Ahmedabad Municipal Corporation, the Indian Institute of Public Health-Ghandinagar and Public Health Foundation of India, the Natural Resources Defense Council, Emory University, and Mount Sinai School of Medicine. The effort was funded by the Climate Knowledge Development Network. While the work shown highlights the role of the Indian monsoon, it is applicable to all sectors of human society in lower and higher resource settings.