An Experimental Study of Small-Scale Variability of Rainfall

Spatial and temporal variability of rainfall has been studied employing six dual rain gauge sites at NASA Wallops Flight Facility, Wallops Island, Virginia (37.84°N, 75.48°W). The rain gauge sites were in a line where the separation distance ranged from 0.4 to 5 km and the sampling area, 5 km2, matched with TRMM and GPM-core satellite footprints. The database included two years of continuous gauge observations which allowed to determine the rainfall variability between seasons, six-month periods, and two years. The database was also grouped based on physical characteristics (tropical cyclones, and frontal events) and spatial homogeneity. The weather maps were used to determine the frontal and typical cyclone events, the spatial variability was relatively less significant in physically conditioned events than the climatologically classified events. This was due to the fact that the rainfall was mostly observed in all six sites during frontal or tropical system passage and the variability between these sites was not as significant as rain observed at a few sites in unorganized convection. The latter events were included in climatologically classified events. The spatial homogeneity was determined based on the coefficient of variation. The homogeneous events showed less spatial variability than the highly variable events.

The spatial variability at different climatological and physical conditioned events results in a wide variety of decorrelation distances (d0), ranging from 5 to 40 km. In this study, the decorrelation distance is calculated from the best fit to the correlation (r) which is expressed as a function of distance (d) in the form of r = r0 exp(-d/d0)s, where r0 is the Pearson correlation between the collocated gauges and s is the shape parameter. The d0 is inversely proportional and sensitive to the shape parameter. Previous studies reported that 5-minute accumulations might be considered as gauge instantaneous rainfall. The averaging gauge accumulations for longer periods result in higher correlations and longer decorrelation distances. The gauge time averaging has originally been done to reduce the sampling errors when gauge measurements are employed to validate the larger volumes satellite rain estimates. Considering temporal sampling, polar orbiting satellites like TRMM pass through at a given point within its coverage twice a day, while GPM program plans to have a three-hourly sample through its constellation satellites. The monthly rainfall error in a footprint scale is then 73% and 36% for three-hourly and twice-daily passages, respectively. The partial beam filling is another source of error in satellite rain estimate. The partial beam filling arises from non-uniform and partial coverage of rainfall within a footprint. We separated these two factors and noticed that the gauge site at one of the outer edges reported highest amount and occurrence than other gauge sites. With respect to previous similar studies, Wallops Island offers a diverse climatology with weather systems from remnants of tropical storms to east coast cyclogenesis. Although there is no climatological monthly rainfall trend, the diversity in monthly rainfall is evident between two-months long dry and wet periods. Regardless, Wallops is one of the best choices to conduct this kind of study as it received 1960 mm of rain in 7087 samples during two-year observation period.