3 Drought Monitoring Using Reservoir Data Collected via Satellite Remote Sensing
Huilin Gao1, Gang Zhao1, Yao Li1, and Shuai Zhang2
1 Zachry Department of Civil and Environmental Engineering, Texas A&M University, College Station, Texas, USA
2 Department of Geological Sciences, University of North Carolina at Chapel Hill, Chapel Hill, North Carolina, USA
ABSTRACT
Drought can significantly impair water availability, agricultural productivity, ecosystem health, and the economy. The advent of satellite remote sensing has meant that reservoirs can be observed from space, which offers a unique promise for monitoring hydrological drought. Thus, the overarching goal of this chapter is to review and explore how these remotely sensed reservoir data (elevation, area, and storage) can be used for drought monitoring and decision making. Although reservoir storage is deemed to be the best indicator of drought severity, such data are only available for a small portion of reservoirs globally, mainly limited by the availability of altimetry measurements. Reservoir area data, which have better spatial and temporal coverage than elevation/storage data, can be used to derive a drought index suitable for monitoring purposes at local and regional scales. A new surface‐area‐based hydrological drought index has been introduced and compared with the meteorological drought index. The skills of hydrological‐modeling‐based drought monitors can be enhanced by incorporating remotely sensed reservoir information. Furthermore, new and future satellite missions, such as Ice Cloud and Land Elevation Satellite 2 and Surface Water and Ocean Topography, will make the global monitoring of reservoirs storage feasible.
3.1. INTRODUCTION
Drought, which is caused by a lack of precipitation over an extended period in a large region, can significantly impair water availability, agricultural productivity, ecosystem health, and the economy (Mishra & Singh, 2010). It has become a pressing global issue given the fast growing population (Gleick, 2003) and the fact that more than 38% of world’s population live in dryland regions (Reynolds et al., 2007). Under global warming, both observations and model simulations have shown an increasing trend of aridity (Dai, 2013; Sheffield & Wood, 2008; Trenberth et al., 2014).
Droughts are typically classified into four types: meteorological, agricultural, hydrological, and socioeconomic (Wilhite & Glantz, 1985). Meteorological drought is defined based on the amount and duration of a precipitation shortfall. Agricultural drought usually focuses on quantifying the decline of soil moisture (by examining its magnitude and duration) or vegetation water stress (by comparing with normal conditions) during a dry period. Hydrological drought is used to quantify the extreme events through the “lens” of water resources, by estimating the deficits of surface water and ground water. With elements of meteorological, agricultural, and hydrological droughts, socioeconomic drought is associated with the imbalance between the supply and demand of water resources by using water as the principle economic “good” (AMS, 2004).
Many drought indices have been designed for quantifying the severity of different types of drought events (Heim Jr, 2002; Mishra & Singh, 2010). Representative meteorological drought indices include (just to name a few) the Standardized Precipitation Index (SPI; McKee, 1995; McKee et al., 1993), the Rainfall Anomaly Index (RAI; Van Rooy, 1965), the Palmer Drought Severity Index (PDSI; Palmer, 1965), the deciles (Gibbs & Maher, 1967), and the National Rainfall Index (NRI; Gommes & Petrassi, 1994). Some agriculture drought indices primarily examine soil moisture values, such as the Soil Moisture Deficit Index (SMDI; Narasimhan & Srinivasan, 2005) and the Soil Moisture Index (SMI; Hunt et al., 2009), while others focus on quantifying crop or vegetation stress, such as the crop moisture index (CMI; Palmer, 1968) and the Evaporative Drought Index (EDI; Yao et al., 2010). There are a variety of hydrological drought indices, ranging from the Standardized Runoff Index (SRI; Shukla & Wood, 2008), the Streamflow Drought Index (SDI; Vicente‐Serrano et al., 2011), to the Surface Water Supply Index (SWSI; Shafer, 1982), which integrates reservoir storage, streamflow, and precipitation. Because the quantification of socioeconomic drought is intriguingly complex, it is difficult to define a generalized drought index to represent the many aspects of demand and supply and their interactions.
A number of drought monitors have been developed at regional and global scales. For meteorological drought, some representative ones include: the U.S. Drought Monitor (Svoboda et al., 2002), the Standardized Precipitation–Evapotranspiration Index (SPEI) Global Drought Monitor (Vicente‐Serrano et al., 2010), and the Global Drought Information System (Heim Jr & Brewer, 2012). Most of the agricultural drought monitors are based on soil moisture outputs from hydrological models, such as the North America Land Data Assimilation System (NLDAS) Drought Monitor (Xia et al., 2014), the African Flood and Drought Monitor (Sheffield et al., 2014), and the Experimental Surface Water Monitor for the Continental United States (Mo, 2008). Taking advantage of satellite observations, agricultural droughts also can be monitored via remotely sensed soil moisture (Mishra et al., 2017) and evapotranspiration (Mu et al., 2013). For quantifying hydrological drought, a SRI based indicator has been calculated for the continental United States from model‐simulated runoff (Shukla & Wood, 2008). Meanwhile, satellite observations collected by the Gravity Recovery and Climate Experiment (GRACE) have offered an overall perspective about terrestrial water storage anomalies, in which signals from surface water, groundwater, soil moisture, and snow water are lumped across the globe from 2002 to 2017 (B. F. Thomas et al., 2017).
While these drought monitors have provided valuable information from various perspectives, there is a shortage of regional/global drought monitors that can support local scale decision‐making in a coherent manner. Meteorological drought is the “driver” (of each of the other drought categories), and we do not have a means to control these events. Actions can be taken, however, to mitigate the severity and impacts of agricultural drought and hydrological drought, and to help reduce the losses of socioeconomic drought. In this chain of relationships, reservoirs play a key role. Over the past century, more than 7320 reservoirs (with a surface area of 0.1 km2 or larger) have been constructed globally (Lehner et al., 2011). These reservoirs have redefined water resources management by damming and releasing river water to optimally support water supply (municipal, industrial, and agricultural), flood control, hydropower generation, and recreation purposes. On one hand, water is saved in reservoirs for anthropogenic usage during droughts. On the other hand, reservoir inflow and storage are directly affected by the severity of hydrological drought. This paradox is an even larger challenge for arid and semiarid regions where reservoir evaporative losses are significant (Friedrich et al., 2018). Despite the fact that in situ reservoir data (e.g., elevation, storage) are collected regularly, such information is rarely published or shared among regions with conflicts of interest (Zhang et al., 2014). Furthermore, consistent long‐term reservoir evaporation rate records are even harder to measure at a regional scale (Zhao & Gao, 2019a).
The advent of remote sensing has allowed for reservoir elevation, area, and storage using data collected from multiple sensors. Such near‐real‐time information, where available, can be directly used for supporting decision‐making under drought conditions. Thus, the overarching goal of this chapter is to review and explore how these remotely sensed reservoir data could be used for drought monitoring and decision‐making.
3.2. DROUGHT MONITORING USING REMOTELY SENSED RESERVOIR DATA
3.2.1. Reservoir Elevation
Reservoir elevation is the key factor for determining flow regulation rules. To optimize reservoir operation efficiency, a reservoir is typically divided into several “pools”: flood control, conservation, and inactive (Zhao et al., 2016). Water stored in the conservation pool is primarily used for agricultural irrigation, municipal uses, industrial supply, and hydropower generation. Because all of the operation rules are based on the reservoir stage