Assessing the use of standardized groundwater index for quantifying groundwater drought over the conterminous US

Abstract Groundwater drought is a drought phenomenon caused by the decrease in groundwater level or groundwater storage under natural and anthropogenic disturbances (e.g., climate variability/change and groundwater abstraction). Due to the complexity of groundwater flow systems and the difficulty in obtaining direct observational data related to groundwater resources, it remains challenging to characterize groundwater drought quantitatively. To this end, standardized groundwater level index (SGI), which was modified from the standardized precipitation index (SPI), was used to quantify groundwater drought with the long-term groundwater level data (1981-2010) obtained from the Climate Response Network wells across the conterminous United States. The best fitting function of groundwater level distributions was lognormal which was selected from various distribution functions based on the Anderson-Darling (AD) test among 100 wells. Four monitoring wells located in Georgia, Massachusetts, Oklahoma and Washington were selected as the main research objects. The trend of change in groundwater level was divided into two categories: the wells in Georgia and Oklahoma showed an initial decrease followed by an increase while the wells in Massachusetts and Washington showed a continuous decline. Groundwater drought varied significantly in different areas due to the complexity of geographical location, agricultural irrigation, population and other natural environment and human activities. Start and end time of drought conditions and the severity of drought and flooding at different time scales in the same area also varied, therefore it is necessary to describe different groundwater droughts at a reasonable time scale. In this paper, the difference between SGI and SPI at different time scales was so obvious that the cross-correlation analysis was used to find the law of lag time. The cross-correlation coefficients increased with the increasing time scales, and the average correlation values between SPI and SGI of wells in Georgia, Massachusetts, Oklahoma and Washington were -0.568, - 0.634, - 0.667 and 0.496, respectively. With the time scale increased, the lag time in Georgia and Oklahoma also increased, but the lag time in Massachusetts and Washington region was almost unchanged, and there was almost no lag time in Massachusetts region. Through the study of the hydrogeological conditions such as vadose zone, lithology, soil and the influence of human activities, the hysteresis effect can be analysed, which can provide a new alternative to the conventional method and important information for future groundwater resource management.