Integrating entropy and copulas for precipitation gauging network optimization based on information balancing strategy

This paper assessed five types of entropy based optimization criteria (H-T, H-C, H-T1-C&H-T2-C, TI) for precipitation networks under the impact of data discretization, with the use of daily precipitation in 2007-2016 from rainfall monitoring network in the southwest hilly region in the Taihu Lake Basin. We analyzed station ranking discrepancy under different criteria and discretization methods, the effect of index weights in multi-objective criteria, and the inter-annual variance of ranking. Results showed station ranking obtained with H-C was the most representative; H-T2-C had the largest deviation from other criteria, the highest sensitivity to index weight, and the largest inter-annual variance. We further assessed the Pareto solutions under different criteria. Results showed H-C and H-T1-C criteria shared the most common Pareto solutions, featuring high information content and low redundancy. The occurrence frequency of stations increased as the station number rose, while the frequency difference among stations decreased.