E 0- and ET-based Drought-Monitoring Tools
The availability of RS-derived data in both thermal and visible bands is maturing, as is the necessary climatology required to place analyses in statistical contexts, particularly from reanalyses. Further, the interactions between ET and E0 in drought are becoming better understood while awareness is increasing of the dangers of simplistic parameterizations of E0 as either an end in itself, or as a driver of ET from LSMs. This has led the drought and ET communities to begin to coalesce behind new approaches to drought monitoring that examine the flux of moisture from the earth's surface to the atmosphere—the demand side of the surface water balance (or imbalance, in the case of drought). Further, new tools are being developed that use RS observations and/or a better understanding of ET and E0 relations in drought. Here we examine some of these emerging tools and their promise.
As an example of an existing tool that has been shown to be fraying with age, the Palmer Drought Severity Index (PDSI; Palmer 1965) has long been commonly used in the United States in monitoring, generally on weekly or monthly timescales. It gained popularity due to its minimal data requirements: only Prcp and T are needed. In the United States, it is central to the suite of tools informing the US Drought Monitor's (USDM) assessment of developing drought conditions. It is also used in NOAA's operational PDSI. The PDSI derives the water balance using a simple hydrology bucket model with a two-layer soil column, where the difference between Prcp and the sum of Runoff and the ET results in a moisture anomaly then used to derive a non-dimensional drought index. In the bucket model, the maximum possible ET is the minimum of the available water in the soil column and E0.
Despite its popularity and longevity, the PDSI only poorly resolves the evaporative aspects of drought, due to its use of T-based E0, which Palmer (1965) originally estimated from the Thornthwaite (1948) equation and then by another T-based model (of unknown origin) further divorced from observed E0-T relations. While T-based E0 correlates well with humidity and net radiation on subannual timescales, drought-scale anomalies in observed E0 are often forced by drivers other than T (Equation 11.2), particularly across CONUS (Hobbins 2016). The PDSI is not suited for widespread application, particularly in more arid regions or at high latitudes or in cooler seasons in which neglected cold processes dominate (Sheffield et al. 2012). Further, the empirical parameters characterizing local climate and drought timing were only calibrated for the midwestern United States. These issues pertaining to its widespread use in time and space may be obscured to casual users seeking a simple off-the-shelf index with minimal data requirements. Indeed, the PDSI is often used in longterm drought analyses worldwide, to questionable effect (see Section 11.4).