Traditional Remote Sensing Methods for Drought Monitoring
Satellite-based remote sensing provides a unique perspective on drought, providing spatially distributed information that can be used in tandem with traditional, in situ-based measurements to gain a more complete view of drought conditions across the landscape. The space-borne earth observation era began in 1960 with the launch of the Television Infrared Observation Satellite (TIROS) by the National Aeronautics and Space Administration (NASA), which was designed to determine the utility of satellite-based imagery for the study of the earth. Although TIROS was designed for meteorological and climatological observations, the value of the satellite-based observations of the earth's environment were realized through this effort and provided the foundation for subsequent development of satellite-based, land observation remote sensing instruments in the following decades.
Remotely sensed satellite imagery provides a "big picture" view of the spatial patterns and conditions of the earth's land and water surfaces and atmosphere. The digital image data acquired by these space-borne remote sensing instruments overcome several of the issues related to in situ-based observations highlighted in the previous section. Satellite imagery provides a spatially continuous series of spectral measurements across large geographic areas that are acquired in the form of pixel-based grids. The ground area measured in these image pixels varies by satellite-based sensor, ranging from several meters (e.g., 1-30 m) to several kilometers (e.g., 1-25 km). The complete spatial coverage of satellite imagery can fill in the spatial gaps within and between in situ-based observational networks and provide invaluable information in many parts of the world where such networks may be sparse or nonexistent. Another benefit is that satellite imagery is collected in an objective and quantitative manner, yielding spatially and temporally consistent datasets that are required for environmental monitoring activities such as drought detection and severity assessment. Most satellite-based sensors record the reflected or emitted signal of electromagnetic radiation (EMR) from multiple regions of the EM spectrum spanning the visible, infrared, and microwave wavelengths. EMR in different spectral regions is responsive to different environmental parameters and can collectively be used to estimate and assess different drought-related environmental conditions such as plant stress and soil moisture. As a result, satellite-based estimates of these types of environmental conditions provide a valuable source of internally consistent historical data for accurate anomaly detection required for drought monitoring.
Historically, the application of satellite remote sensing for operational drought monitoring has primarily involved the use of Normalized Difference Vegetation Index (NDVI) data from the National Oceanic and Atmospheric Administration (NOAA) advanced very high resolution radiometer (AVHRR). The NDVI, which was developed in the early 1970s by Rouse et al. (1974), is a simple mathematical transformation of data from two spectral bands commonly available on most satellite-based sensors, the visible red and near infrared (NIR). The visible region is sensitive to changes in plant chlorophyll content, while the NIR region responds to changes in the intercellular spaces of the spongy mesophyll layers of the plants' leaves. Based on these interactions, the NDVI was developed as a general indicator of the state and condition of vegetation, with index values increasing with the amount of healthy green photosynthetically active vegetation. A large body of research has shown that NDVI has a strong relationship with several biophysical vegetation characteristics (e.g., green leaf area and biomass) (Asrar et al. 1989; Baret and Guyot 1991) and temporal changes in index values are highly correlated with interannual climate variations (Peters et al. 1991; Yang et al. 1998; McVicar and Bierwith 2001; Ji and Peters 2003). As a result, negative deviations in NDVI values for a given time period during the growing season compared to the long-term historical average NDVI value for that same period is indicative of vegetation stressed from an event such as drought. This concept has formed the basis for many NDVI-based drought monitoring efforts throughout the world, which have primarily relied upon the analysis of historical NDVI time series data collected by the satellite-based AVHRR sensor. AVHRR has provided a near-daily global coverage of 8 km gridded NDVI products dating back to the 1980s (note: 1-km AVHRR NDVI data are available since 1989 over the continental United States). The long multidecade time series of AVHRR NDVI data has proved valuable for drought monitoring because NDVI anomaly measures calculated from this dataset reflect the degree of departure of "current" vegetation conditions from longer-term historical average NDVI values for that same date over a period of more than 25 years. The use of AVHRR NDVI- based measures for drought monitoring can be traced back more than 20 years to the early work of Hutchinson (1991); Tucker et al. (1986); Kogan (1990); Burgan et al. (1996); and Unganai and Kogan (1998). Key examples of current operational drought monitoring using AVHRR NDVI anomaly products include the US Agency for International Development (USAID) Famine Early Warning System (FEWS) and the United Nations Food and Agricultural Organization (FAO) Global Information and Early Warning System (GIEWS) on Food and Agriculture.
The Vegetation Health Index (VHI) (Kogan 1995), which builds upon the NDVI concept and incorporates a temperature component through the use remotely sensed thermal infrared (TIR) data, is another traditional remote sensing indicator used for drought monitoring. The VHI integrates two indices into its calculation: the NDVI-based Vegetation Condition Index (VCI; Kogan and Sullivan 1993) and the TIR-based Temperature Conditions Index (TCI; Kogan 1995). The VCI is based on the assumption that the historical maximum and minimum NDVI values represent the upper and lower bounds of possible vegetation at a specific location (i.e., areas within an image pixel), with anonymously low NDVI values indicative of vegetation stress. The complimentary TCI is based on a similar concept where the historical maximum and minimum TIR values represent the upper and lower bounds of thermal response of vegetation for a specific location. Higher TIR anomalies expressed in the TCI should correspond to drought-stressed vegetation because more energy is being partitioned into the sensible heat flux rather than the latent heat flux because there is less moisture available to be transpired and evaporated from the vegetation and soil background. Kogan (1995) unified the VCI and TCI into the VHI to provide a remotely sensed indicator representative of both the NDVI and thermal response of vegetation. The VHI has been derived globally since 2005 at 8- and 16-km spatial resolutions using AVHRR NDVI and TIR data inputs.