Implications of hydrologic predictability for the development of a global drought information system

Abstract:

Drought information systems consist of two key elements: nowcasts; essentially maps that identify the locations of areas currently in drought and the severity thereof as related to an historical record; and drought forecasts, which predict the best understanding of the evolution of drought, along with the uncertainty of those predictions. Because drought nowcasts are only meaningful if current conditions are related to an historical period, stable quality controlled historical space-time records of key drought variables are essential. Generally, these include surface meteorological conditions, such as surface air temperature and precipitation, and other land surface variables, such as downward shortwave and longwave radiation, which typically are derived from more commonly measured variables, such as the daily temperature and temperature range. I discuss our experience in constructing such records, both across the continental U.S. where we have the luxury of relatively lengthy and high quality meteorological and climate records, and globally, where attaining statistical stability of constructed historical records is more challenging. In the case of agricultural (soil moisture) and hydrological (runoff) droughts, the key drought variables usually must be derived from land surface models (LSMs) forced by precipitation, surface air temperature, humidity, surface wind, and downward shortwave and longwave radiation. Clearly, the statistical stability of derived soil moisture and runoff depend on stability of the LSM forcings, and I discuss some experiences and issues in our development of a prototype global drought information system GDIS. With respect to drought forecasts, skill derives either from a) persistence in the land surface initial states, primarily soil moisture and snow; and b) skill in prediction of the LSM forcings (most importantly, precipitation). I review recent work that shows the relative influence of these two sources of drought forecast skill, as well as evidence of practical skill or lack thereof in seasonal precipitation forecasts, and their implications for drought forecast skill.