Loading...
Precipitation correlation distance: characterizing the spatial correlation of precipitation in the contiguous United States
Yeo, Alexa
Yeo, Alexa
Citations
Altmetric:
Editor
Date
Date Issued
2024
Date Submitted
Keywords
Collections
Research Projects
Organizational Units
Journal Issue
Embargo Expires
Abstract
Precipitation uncertainties are one of the major causes of model errors in hydrology (Bárdossy et al., 2022). Direct precipitation measurements come from a limited number of gauged locations, which are used to inform precipitation estimates in areas without gauges. Precipitation variability increases uncertainties in precipitation measurements and, in a warming climate, the variability of precipitation is projected to increase (Pendergrass et al., 2017). Appropriate understanding of the spatiotemporal variability of climate-scale seasonal to annual precipitation and its changes could help improve gap-filling approaches used in gridded data product development. Here, I aim to better characterize the spatiotemporal variability of precipitation in the contiguous United States (CONUS) by introducing the precipitation correlation distance (PCD), which is the distance at which correlation between gauges no longer exists. I use daily precipitation data from the serially complete Earth (SC-Earth; Tang et al., 2021) dataset for just under 20,000 precipitation gauges from 1950 to 2019 across CONUS along with annual and monthly precipitation normals from the Parameter-elevation Regressions on Independent Slopes Model (PRISM; Daly et al., 2008) to determine annual and seasonal precipitation anomalies at each gauge. Over a grid of 64 km resolution, I use a spatial statistics modeling approach to determine the PCD, the range estimate of the kriging analysis, in a 400-km radius circle centered on each grid cell. I also compute lambda, the reciprocal of the signal to noise ratio, for each grid cell, representing our confidence in the distance estimates. Findings are presented in the form of the PCDs and lambda values mapped across CONUS. The PCD results are also broken down by region and a trend analysis was performed for each grid cell. Finally, I evaluated the extent to which PCD varies interannually with temperature. I find that the spatial extent of annual precipitation correlation varies. Complex terrain tends to have shorter PCDs than simple terrain. For seasonal precipitation, the spatial structure of correlation is most variable in summer. The Upper Midwest, Ohio Valley, South, and Southeast regions have the longest PCDs compared to the rest of the country, especially in winter. The Northwest, West, and Southwest regions have the shortest PCDs and there is little variation seasonally. The trend analysis reveals that PCD decreased over the 70-year study period in summer, fall, winter, and spring (p < 0.05) in 48%, 16%, 12%, and 11% of grid cells, respectively. PCD decreases were largest in the summer in the Northern Rockies and Plains, Northwest, and Upper Midwest regions. These results could help re-evaluate gap-filling approaches that assume constant PCD and inform network design of gauges in places with high spatial variation. These findings could also be used to evaluate climate models, as well as inform the development of gridded products to improve forcing datasets for hydrologic models and validation datasets for atmospheric models.
Associated Publications
Rights
Copyright of the original work is retained by the author.