Probabilistic drought forecasting using copula and satellite rainfall based PERSIANN‐CDR and MSWEP datasets

• Published in: International journal of climatology 2022-10, Vol.42 (12), p.6441-6458
• Main Authors: Alijanian, Mohammadali, Rakhshandehroo, Gholam Reza, Dehghani, Maryam, Mishra, Ashok
• Format: Article
• Language: English
• Subjects: Artificial neural networks; Basins; Climatic data; copula; Datasets; Drought; drought class transition; Drought forecasting; Ensemble precipitation; Forecasting; Hydrologic data; Mathematical models; Modelling; MSWEP; Neural networks; PERSIANN‐CDR; Precipitation; Precipitation estimation; Probability theory; Rain; Rainfall; Rainfall data; Remote sensing; Satellites; Spring; Spring (season); Statistical analysis
• ISSN: 0899-8418; 1097-0088
• DOI: 10.1002/joc.7600

In situ rainfall data plays a vital role in drought assessment. However, adequate in situ data are not available in many parts of the world, and they do not provide the proper spatial coverage for drought assessment. With the advacements in satellite rainfall estimates (SREs), it is possible to monitor droughts in ungauged basins. However, the applications of SREs for drought forecasting are not widely explored due to the inherent uncertainties associated with these products.In this study, we evaluated two long‐term SREs for drought forecasting in the Zayandehrood basin, a critical region in the central plateau of Iran. The performance of two SREs, including Precipitation Estimation from Remotely Sensed Information using Artificial Neural Networks‐Climate Data Record (PERSIANN‐CDR), and Multi‐Source Weighted‐Ensemble Precipitation (MSWEP) are compared with observations during 1983–2015. The overall results indicate that utilizing MSWEP data in the forecasting model can slightly overestimate the probability of spring drought based on winter drought (with the highest error of 8.5%). In comparison, the PERSIANN‐CDR underestimated the probabilities (with the lowest error being −44%). The performance of copula models and SREs can vary based on the thresholds for drought severity. For example, the performance of MSWEP datasets for predicting moderate to severe droughts during the Spring season is closer to the predicted values by gauge datasets. It is concluded that the MSWEP may be considered more reliable in drought forecasting than the PERSIANN‐CDR. Our results highlight the potential application of copula‐based forecasting models for seasonal drought forecasting using SREs datasets. Such models can be implemented for global‐scale drought predictions, especially in ungagged basins. This research is an attempt to find the evaluation of PERSIANN‐CDR and MSWEP in drought forecasting. Zayandehrood basin which is located on central Iran and suffers from severe droughts is selected as a case study. Two different probabilistic approaches are developed by copula joint functions and their results are compared. Results show that MSWEP shows better performance in drought forecasting.