Accuracy assessment of potential alternatives to observed precipitation for extremes over complex topography

Several complex topographic regions of the world, in addition to their susceptibility to recent climate change impacts, lack proper precipitation observation network to monitor precipitation extremes. This study assesses the accuracy of model-based (ERA5) and satellite-based (GPM IMERG, CHIRPS, and PERSIANN) precipitation products as potential alternatives to ground station-based network for monitoring of such extremes over a complex topographic region of Pakistan. The two-staged assessment is conducted through an initial time series and statistical evaluation of the products, by using station-based observed data as reference, followed by their accuracy assessment in capturing wet and dry precipitation extremes (based on moving sums thresholds) over the locations of 5 stations for 2010–2018. The initial evaluation results show that ERA5 outperformed the three satellite-based products with a lower RMSE (49.24 mm/mon) and a higher monthly correlation (0.79) with the observed data. The observed monthly seasonality was best captured by GPM despite it having an overall underestimating tendency. All four products tend to under- (over) estimate wet (dry) precipitation extremes. ERA5 (CHIRPS) performed best (worst) in capturing the observed wet extremes while the statistical performance of GPM IMERG was lower than ERA5. The study concludes that the considered products, especially ERA5, could be recommended as potential alternatives to the observed extremes after certain improvements in their systematic bias and random errors.