资源环境科技发展态势分析平台

Maximum precipitation, which varies among locations and annually at a location of interest, is the base for sizing hydraulic structures. Understanding such spatiotemporal variations is necessary for accurate rainfall prediction but remains incomplete in most regions, including the state of Virginia. The objective of this study was to fill this information gap for Virginia using 15-min precipitation data collected at 57 gauges. In this regard, a linear regression method and a modified Mann-Kendall technique that can consider the climate-relevant non-stationarity were applied to detect possible step changes and temporal trends of maximum annual rainfall intensities at various time intervals, while spatial statistics, namely global Moran's I and local indicator of spatial association, were used to reveal the spatial autocorrelations and clusters. The results indicated that although no step changes were detected, statistically significant trends were detected for almost half of the selected gauges. The coastal plain experienced an increasing trend in rainfall intensity for durations of 24 h and longer, whereas the other areas of the state had a decreasing trend, with relatively more trends and larger decreasing rates in the west-central ridge-valley areas. The spatial autocorrelations of rainfall intensities in the state of Virginia were dependent on a spatial scale of interest. Given the diversity of the climatic patterns and physiographic features in the testbed, this study demonstrates innovations in understanding how extreme rainfall intensity would be changed from climate change beyond Virginia.