Historically, the development of satellite-based drought monitoring tools has focused on assessing general vegetation health conditions through the analysis of vegetation indices (VIs) such as the NDVI and VHI, which were discussed previously. While these VIs have proved valuable for this application, as demonstrated by their widespread use in monitoring systems such as FEWS, they still have several challenges for adequately characterizing drought-related vegetation stress. These VI-based methods rely on comparing the departure of the VI values to the historical average VI value for a given time period during the year, with drought stress represented by below- average VI values. As a result, a VI value threshold must be established that signifies a drought-stress vegetation signal within the range of negative VI anomaly values. In addition, other thresholds must be established to classify different drought severity levels. For example, does a 25 percent negative VI anomaly distinguish between drought and nondrought conditions and, if so, what negative percent VI values represent different drought severity levels (e.g., 25-35 percent = moderate, 35-50 percent = severe, and >50 percent = extreme)? Selection of such thresholds is often arbitrary and can be challenging given that they can vary by vegetation type, geographic region, and season. In addition, other environmental factors such as flooding, fire, frost, pest infestation, plant viruses, and land use/land cover change can result in negative VI anomalies that mimic a drought signal. As a result, traditional VI-based anomaly products can be misinterpreted as drought if analyzed in isolation. More recently, the remote sensing community has placed an emphasis on developing composite drought indicators (CDIs) that integrate various types of information, including VIs, into a single indicator that is representative of drought-specific vegetation conditions.
A prime example of a remote sensing-based CDI approach is the vegetation Drought Response Index (VegDRI), which integrates satellite-based vegetation condition observations, climate-based drought index data, and other environmental information to characterize drought stress on vegetation (Brown et al. 2008; Wardlow et al. 2012). VegDRI builds upon the traditional VI-based approach using satellite-based NDVI anomaly information as a general indicator of vegetation health, which is analyzed in concert with climate-based indicators that reflect the dryness conditions (i.e., SPI) for the same time period. Collectively, drought stress would be manifested as both below-average NDVI conditions and abnormally dry conditions in data inputs for VegDRI, with the NDVI anomaly decreasing and the climatic dryness conditions increasing as drought conditions intensified. VegDRI also incorporates several environmental characteristics of the landscape (land use/land cover, soils, topography, and ecological setting) that can influence climate-vegetation interactions at a given location. An empirical-based regression tree analysis method is used to analyze a historical record of satellite, climate, and environmental data to build the VegDRI models. VegDRI characterizes the severity of drought stress on vegetation using a modified version of the PDSI classification scheme (Palmer 1968). More technical details about the VegDRI methodology are presented by Brown et al. (2008) and Wardlow et al. (2012).
A VegDRI map for June 11, 2012, is presented in Figure 10.1, showing the widespread severe to extreme drought conditions that covered parts of the Rocky Mountains region and western United States at that time. VegDRI has three drought severity classes (moderate, severe, and extreme), as well as a predrought stress class reflecting areas of potential drought emergence. VegDRI maps have been operationally produced over the CONUS
The VegDRI map (a) shows the general seasonal drought conditions on June 11, 2012, over the continental United States, with large areas of moderate to severe drought depicted across the western and parts of the southeastern United States. The QuickDRI map (b) highlights shorter-term drought stress intensification over many parts of the drought-stricken western United States, but also reveals the rapid, shorter-term intensification of drought conditions that were beginning to occur across the US Corn Belt region (highlighted in a black oval), which were not detected by the longer-term VegDRI or in other key tools such as the USDM until later June or early July. This example illustrates that QuickDRI represents an early warning alarm tool of rapidly changing drought conditions that can complement existing drought monitoring tools such as VegDRI and the USDM.
since 2008, with a continuous time series of historical 1-km spatial resolution maps dating back to 1989. VegDRI information is routinely used by the USDM authors, US Bureau of Land Management (BLM) rangeland assessment programs, NWS drought bulletins, and several state drought task forces in the western United States. The VegDRI concept has gained interest internationally, with a similar VegDRI tool being developed in Canada for the North American Drought Monitor (NADM) and modified versions of the tool in China, Czech Republic, and India. VegDRI has proved to be a useful agricultural drought indicator that reflects longer- term seasonal conditions on the scale of several months (Brown et al. 2008; Tadesse et al. 2015).
The Quick Drought Response Index (QuickDRI) (B. D. Wardlow [The Quick Drought Response Index (QuickDRI)] pers. comm., January 20, 2017) is another CDI that has recently been developed to characterize shorter-term drought intensification on the order of a few weeks to a month. QuickDRI uses a similar modeling approach to VegDRI to integrate several new extended time-series remote sensing datasets and climate indictors that are sensitive to shorter-term changes in environmental conditions that influence drought stress. These variables include the thermal-based Evaporative Stress Index (ESI) (Anderson et al. 2007, 2011) representing the ET component of the hydrologic cycle (see Section 10.6), modeled root-zone soil moisture data from the North American Land Data Assimilation System-2 (NLDAS-2) (Xia et al. 2012) representing subsurface moisture conditions, and the climate-based Standardized Precipitation Evapotranspiration Index (SPEI) (Vicente-Serrano et al. 2010) and SPI (McKee et al. 1995) representing precipitation and air temperature conditions. Additional input variables include the Standardized Vegetation Index (SVI) (Peters et al. 2002) derived from time-series NDVI data to represent general vegetation health and the same set of environmental variables used in VegDRI. The same regression-tree-based analysis technique used for VegDRI was adopted for QuickDRI model development with models based on anomaly data for the ESI, soil moisture, and climate inputs representative of conditions on a 1-month time step. As a result, QuickDRI is designed to monitor the level of drought intensification over a monthly time period to serve as an "alarm" indicator of rapidly emerging drought conditions that are not detected by the longer seasonal VegDRI. Figure 10.1 shows the rapidly intensifying drought signal over the US Corn Belt region that is captured by QuickDRI for the flash drought conditions that occurred across that area during the early summer. By comparison, the longer-term seasonal QuickDRI showed most of the same area in normal or predrought conditions and did not represent the emerging severe to extreme drought conditions until mid- to late July. The completion of an operational QuickDRI tool for the CONUS is planned for summer 2017.
The Combined Drought Indicator (Sepulcre et al. 2012) was developed for operational monitoring across Europe that combines anomaly information from a climate indicator (i.e., SPI), modeled soil moisture (i.e., output from
LISFLOOD model [De Roo et al. 2000]), and remotely sensed vegetation conditions (i.e., fraction of Absorbed Photosynthetically Active Radiation [fAPAR]) observed from the Medium Resolution Imaging Spectrometer (MERIS). The index is based on the historical analysis of the relationship between precipitation deficits expressed by the SPI and the response of soil moisture and fAPAR and inherent time-lag relationships among these variables. The drought categories classified by the Combined Drought Indicator include a watch and warning category and two alert categories that are classified based on whether there is a precipitation and/or soil moisture deficit and the vegetation stress that is detected in the fAPAR data. The European Drought Observatory (EDO) operational program produces a 0.25-degree spatial resolution Combined Drought Indicator map over Europe on a 10-day time step.