A rigorous attribution of the demand side of drought: a case study in the Midwest US.

Our goal here is to answer the question, “what drives the demand side of drought?” We achieve this by decomposing atmospheric evaporative demand (Eo) anomalies during periods of drought into contributions from all of its drivers, using the US Midwest as a study region. In drought, anomalies in Eo are driven by anomalies in moisture availability, but Eo reacts quickly and so is a robust drought indicator. Thus, asking to what extent each meteorological driver determines evaporative demand in drought conditions is of value both academically and operationally.

We define drought as a sustained imbalance between the supply of moisture from the atmosphere to the surface (Precipitation) and the demand in the atmosphere for moisture from the surface, in favor of the demand. The demand arm is atmospheric evaporative demand (Eo; (sometimes referred to as “potential evaporation”); evapotranspiration (ET), the actual return flux to the atmosphere, is determined as the extent to which this demand can be met by the moisture available at the surface.

In this context, Eo can be thought of as the “thirst of the atmosphere.” It is a function of meteorological and radiative drivers at the surface: specifically temperature, solar radiation, wind speed, and humidity (and to a lesser degree, surface pressure). For Eo we use daily reference ET (ETo) from the Penman-Monteith equation, which provides a fully physical estimate that incorporates the effects of both advective and radiative forcing. We drive ETo by inputs from the North American Land Data Assimilation System phase-2 (NLDAS-2), which are distributed across CONUS at a spatial resolution of 0.125 degrees, and available from 1979 to the present.

Drought periods are determined using various spatially distributed drought-monitoring tools: specifically, the US Drought Monitor (USDM); the Evaporative Demand Drought Index (EDDI); the Standardized Precipitation Index (SPI); and soil moisture percentiles from the NLDAS-driven Noah land surface model.

We conduct a first-order analysis of the anomalies in Eo that exist during drought conditions. This technique assumes that the contributions from anomalies in all drivers sum to the anomaly in Eo; each driver’s contribution is the product of the sensitivity of Eo to, and the anomaly in, the driver. As our expression for Eo (i.e., Penman-Monteith ETo) is differentiable, the sensitivity to each driver can be derived explicitly by partial differentiation. Drivers’ anomalies are observed by querying the reanalysis during drought periods and deriving deviations from the drivers’ long-term means for the same periods across the entire reanalysis period.

Here we present the (i) general methodology for both the development of Eo and its decomposition and (ii) the results of the decomposition of drought-period Eo anomalies into the relative contributions from each driver across the Midwest Drought Early Warning System (DEWS) region.