Abstract The uncertainty of gridded precipitation data, which have many engineering and scientific applications, is investigated in this study by comparing two popular gridded products in Australia, i.e., AWAP (Australian Water Availability Project) and SILO, at local (25?km) and continent spatial scales with various annual, monthly and daily rainfall statistics. The results show that two datasets usually produce similar rainfall statistics over entire Australia but could be significantly different at local scale. For example, the long-term (1960???2016) mean annual rainfall differences between AWAP and SILO is about 2.6 for the entire Australia continent, but could be as large as 1118.9?mm/year (cell at 145.00E and 16.25S) or as low as -697.4?mm/year (cell at 146.25E and 43.25S). And the largest difference of daily rainfall between AWAP and SILO could be as high as 630.3?mm/day. The differences at local scale are so large that one of them must be wrong, and potentially both are inaccurate. The reported results can be used not only to enhance the confidence where two products show similar results, but also to quantify the ranges of uncertainties where two datasets are different. The uncertainty of gridded rainfall is usually large in the regions with limited weather stations, the mountain regions with different algorithms being used to deal with elevations, as well as the coastal regions where an extrapolation is needed to obtain coastal rainfall from inland observations.
展开▼
