연구보고서
- 저자
- 김대하 박사
- 작성일
- 2019.06.13
- 조회
- 295
- 요약
- 목차
Drought has been defined as an abnormally prolonged deficiency in precipitation that leads to substantial hydrological imbalances in ecosystems and human societies. Although this definition includes the concept of water “deficiency,” many drought assessments have depended solely on precipitation anomalies and their chain effects on water cycles. To suit drought assessments to the definition of water deficiency, this study characterized historical droughts in the conterminous United States (CONUS) using evapotranspiration (ET) estimates from the generalized complementary relationship (GCR). Since the GCR explains the land-atmospheric feedback mechanism, it allowed estimations of the actual ET (ETa) and the wet-environment ET (ETw) using routine meteorological observations only. The ETa estimates were validated with streamflow observations and simulations from a land surface model. The Standardized Evapotranspiration Deficit Index (SEDI) across the CONUS for 1895 –2016 was then calculated by normalizing the ET deficit (ETw minus ETa). The SEDI was compared to the Standard Precipitation Index (SPI) at various time scales. The total water storage within catchments was also modeled with the GCR ETa and precipitation data using the linear reservoir theory in order to assess hydrological droughts. The results showed that the GCR provided unbiased ETa estimates with the Priestley–Taylor coefficient of 1.26, which is a reasonable value for saturated grass surfaces. The GCR ETa generally agreed with ETa from precipitation and streamflow observations and was highly consistent with the land surface model simulations. The SEDI was strongly correlated with the SPI at the strong land-atmosphere coupling hotspots. At longer time scales, the correlation between the SEDI and the SPI was higher due to the residence time of water vapor in the atmosphere. Furthermore, by comparing the SEDI and the Vegetation Health Index it was found that the former performed well in capturing vegetative (or agricultural) droughts at time scales of six months or longer. The linear reservoir modelling showed that the SPI at short time scales (e.g., two to four months) may explain variations of the water storage in catchments with wet climates and low surface slopes. This indicates that hydrological droughts could be related to relatively short-term precipitation. The water storage in (semi-)arid catchments, however, was not well captured by the linear reservoir theory due to its low connectivity to streamflow. This study suggests that ETa estimated from the GCR could be a good indicator of agricultural drought and could be used together with precipitation data monitor and forecast hydrological droughts at the catchment, regional, and national scales.