Optimizing rain-gauge network design for enhanced estimate rainfall in ungauged plain and mountainous regions

Abstract

Nowadays, due to the insufcient number of stations for input in hydrological models, it is not possible to simulate runof and foods for points without stations. Therefore, to solve this problem, rain gauge networks should be designed for these station points. The present study was carried out to design an optimal rain gauge network for estimating rainfall in plain and mountainous areas with an area of 39 square kilometers in northeastern Iran. The R software was utilized to identify extreme events within the study area. Four events (representing all of the identifed events) were selected and modeled using the weather research and forecasting model under four schemes, namely Purdue-Lin (Lin), WRF Single-Moment class 3,5 (WSM3, WSM5), and WRF Double-Moment class 6 (WDM6). According to the modeling result, 91 rain gauge stations were required to cover the study area, with the rain gauge stations designed at diferent distances from the synoptic station. The microphysical Lin scheme produced better accuracy levels for the rainfall modeling, as indicated by false alarm ratio and threat score of 17 and 0.91 for the rain-gauge stations located in the mountainous areas and 9 and 0.83 for the rain-gauge stations located in the plain areas, respectively. The rain-gauge stations in the mountainous and plain areas tend to give more acceptable accuracies within 6 and 8 km of the main synoptic station, respectively, as indicated by the False Alarm Ratio (FAR) of 0.30 and 0, respectively. Generally, Outputs of the rain-gauge network design coupled with future-period (24 h) forecasts enable authorities to become aware of extreme rainfall events, and the rainfall data derived from this network can be used as input to hydrological models in non-gauged regions.