Rainfall patterns in a small-scale watershed based on NexRad and ground-based datasets

Quantifying and predicting the ecosystems responses to changes in natural and anthropogenic stressors using environmental models require a realistic representation of probable rainfall in its most sensible spatial-temporal dimensions matching that of the phenomenon under investigation. As one of the most critical inputs in environmental models, rainfall data can significantly change the quality and reliability of the model predictions. Due to the lack of ground-based measurements with high spatial and temporal resolution, other methods like radar, have been recently used as an alternative source of rainfall data. However, little research has been conducted to evaluate the possible tradeoffs in using radar generated rainfall data as oppose to ground-measured rainfall. The main objective of this study was to analyze the ability of radar estimates (NexRad N1P) in representing the rainfall patterns in a small-scale watershed and generate high temporal and spatial resolution rainfall data when rainfall pattern was given. To achieve this objective, first, we compared the precipitation from NexRad N1P and ground-based measurements in the Little Washita River Experimental Watershed in Oklahoma to quantify the differences in their patterns and distributions. And second, we tested the ability of a rainfall pattern simulator software “Zeus” to generate high temporal and spatial resolution rainfall data when the rainfall pattern was given. The generated rainfall data was compared with the original rainfall data from both NexRad and ground-based stations. The comparison of the NexRad and ground-based measured rainfall revealed that the mean rainfall from radar in March, April, May, June, and August is closer to the rainfall recorded from the ground-based station with an average difference between of less than 25%. Rainfall recorded in the other months can easily be affected by extreme rainfall events, and the difference in the mean monthly rainfall can be higher than 40%. Also, the analysis showed that the NexRad has a tendency of recording the heavy and intense rainfall higher than the ground-based rainfall while the less intense rain less. The generated rainfall based on the NexRad data has lesser percent error than the ground-based when simulating the dry rainfall months (Jan, Feb, Mar, Apr, Sep, Oct, Nov and Dec). The results showed that significant differences were found between the NexRad and ground-based datasets that can significantly impact the response of environmental models that used them as inputs. This study enabled us to establish the rainfall patterns by using the NexRad when the density of ground-based stations is not sufficient to derive rainfall time series with high spatial and temporal resolution. Also, the “Zeus” software could help us generate the rainfall time series when other sources of high spatial and temporal resolution data are not available. The use of synthetically generated spatial-temporal rainfall patterns will enable us to explore the impacts of precipitation on hydrologic processes driven by changes in environmental stressors like land use and climate changes