hide
Free keywords:
-
Abstract:
Streamflow prediction is crucial for better water resources management, however, it is still challenging in cold-climate mountainous regions due to sparse ground-based precipitation (GBP) and complex snow-melting processes. Using satellite-based precipitation (SBP) products in hydrological modelling could be beneficial, however, these products suffer from significant errors in these regions. This study compares the product- and hydro-validation of PERSIANN family SBPs (PERSIANN, PERSIANN-CDR, PERSIANN-CCS, PERSIANN-PDIR) in two upper snowfed catchments (Karasu, Murat) of Euphrates River Basin, Türkiye. The first basin (Karasu, 1125-3487 m, 10,275 km2) represents a relatively enriched gauge network, and the second basin (Murat, 1559-3508 m, 5,910 km2) is poorly gauged and has data gaps due to its remote location. The capability of SBPs under different time scales (annual, monthly, and daily) within four seasons and various precipitation thresholds is assessed. Daily hydrological simulations are conducted using the conceptual TUW model and data-driven Multi-layer Perceptron (MLP) model. The simulations are performed in two schemes: calibrated with GBP only and by each SBP individually. According to the results, each SBP has a particular skill for different conditions, while PERSIANN-CDR data gives the best correlation with GBP for different seasons and is followed by the PERSIANN-PDIR-Now, PERSIANN-CCS, and PERSIANN data. Both hydrological models demonstrate better performance in simulation streamflow when the model is calibrated by each SBP separately. Applying post-bias corrections to SBP leads to significant improvements when the models are calibrated with GBP only. MLP can be an alternative to the TUW model, especially for the poorly gauged Murat Basin.