McKee et al. (1993) were the first to recognize the value of examining drought at numerous timescales, implementing this concept in the now-popular
Standardized Precipitation Index (SPI). Development of the SPI was a major breakthrough in the drought-monitoring community and allowed users to see that a drought could be occurring in the short term (e.g., 1-3 month Prcp deficits) even while long-term conditions were wet (e.g., a 24-48 month Prcp surplus). The primary limitation of SPI is that it only considers Prcp and ignores other atmospheric drivers of drought. To improve upon SPI, the Standardized Precipitation Evapotranspiration Index (SPEI) (Vicente- Serrano et al. 2010) was developed with the original goal of having a multiscalar drought index that could account for a warming climate. To accomplish this, SPEI uses a simple water balance (Prcp - E0) as the accumulating variable. The T-based Thornthwaite (1948) E0 approach was initially used but, as with PDSI, caution must be taken when using any T-based E0. Beguerta et al. (2014) tested several different E0 approaches in computing global SPEI, and recommended the fully physical Penman-Monteith model if data are available.
A signal feature of E0 is that it increases in droughts initiated across the hydroclimatic spectrum (i.e., in energy- and water-limited hydroclimates) and across timescales (i.e., in sustained and flash droughts), as noted in Section 11.2.3. This robust signal lies at the heart of the Evaporative Demand Drought Index (EDDI; Hobbins et al. 2016; McEvoy et al. 2016a), an emerging drought-monitoring and early warning tool. EDDI works by ranking E0 depths (using Penman-Monteith ET0 for E0) accumulated over a given timescale relative to same-period depths drawn from a climatology. Periods ranking higher (lower) than the median indicate drier (wetter) than normal conditions. The rank is converted to percentiles of the standard normal distribution, which are then categorized and mapped. It is multiscalar in time and can operate at the spatial resolution of the drivers of E0. Users report that multiple timescales are needed for a convergence-of-evidence approach (e.g., Nolan Doesken, Colorado State Climatologist, pers. comm.,), as dynamics specific to drying impacts on sectors within a region operate at various timescales. EDDI shows promise as an early warning indicator (Figure 11.4) of hydrological drought, and for ongoing monitoring of agricultural drought, both in dryland farming and rangeland environments. Ongoing research will reveal whether the strong relations between forest physiology and E0 permit the index to improve fire weather risk prediction. As EDDI relies solely on the radiative and meteorological forcings of atmospheric E0 and their feedbacks with the state of the land surface, it requires no SM, Prcp, or land surface data, enabling EDDI to operate in ungauged areas. While EDDI is simple to estimate, one must use a fully physical E0 to properly reflect the ET-E0 interrelations and land surface drying anomalies. Fortunately, requisite drivers (Equation 11.2) are available across CONUS and globally.
While the use of E0 focuses on relating atmospheric demand to developing drought conditions (e.g., EDDI) and highlighting the potential for vegetation stress, other indicators are necessary to estimate a direct response of the land surface to drought and to estimate the onset of actual
Temporal evolution of USDM drought categories, 2-week Prcp totals, and 2-week EDDI, ESI, and ESI RCI during a "flash" drought in the midwestern United States, from June 2 to July 28, 2012. (Adapted from Otkin et al. 2014.)
vegetation stress. This has led to the development of the Evaporative Stress Index (ESI) (Anderson et al. 2011a), which is based on RS-derived estimates of ET retrieved via energy balance principles using observations of LST. The ESI represents standardized anomalies in the ratio of ET to E0, and normalization by E0 serves to minimize variability in ET due to seasonal variations in available energy and vegetation cover, further refining focus on the relationships between SM and ET. As an indicator of actual ET, the ESI requires no information regarding Prcp or SM storage—the current available moisture to vegetation is deduced directly from the LST. This signal also inherently accounts for both Prcp- and non-Prcp-related sources and sinks of plant-available moisture (e.g., irrigation, tile drainage, vegetation tied to groundwater reserves; Hain et al. 2015), which can modify the vegetation response to Prcp anomalies. Rapid onset of vegetation and/or water stress can occur when extreme atmospheric anomalies persist for an extended period of time (e.g., several weeks) over a given location (Otkin et al. 2013, 2014, 2016; see example from 2012 in Figure 11.4). Therefore, the development of ESI Rapid Change Indices (ESI RCI) (Otkin et al. 2014) based on weekly changes in the ESI have been developed and evaluated over CONUS. The ESI RCI is designed to capture the accumulated rate of moisture stress change occurring over the full duration of a rapidly changing event. The use of the ESI has been increasing in the drought-monitoring community following the development of an operational system, the Geostationary Operational Environmental Satellite (GOES) EvapoTranspiration and Drought product system (GET-D), which provides daily, operational ESI datasets over all of North America at an 8-km spatial resolution. Additional background information on ESI and ESI RCI is provided in Chapter 10.