Effects of spatial and temporal weather data resolutions on streamflow modeling of a semi-arid basin, Northeast Brazil
Academic Article
Overview
Research
Identity
Additional Document Info
View All
Overview
abstract
2015, Chinese Society of Agricultural Engineering. All rights reserved. One major difficulty in the application of distributed hydrological models is the availability of data with sufficient quantity and quality to perform an adequate evaluation of a watershed and to capture its dynamics. The Soil & Water Assessment Tool (SWAT) was used in this study to analyze the hydrologic responses to different sources, spatial scales, and temporal resolutions of weather inputs for the semi-arid Jaguaribe watershed (73 000 km2) in northeastern Brazil. Four different simulations were conducted, based on four groups of weather and precipitation inputs: Group 1-SWAT Weather Generator based on monthly data from four airport weather stations and daily data based on 124 local rain gauges; Group 2-daily local data from 14 weather stations and 124 precipitation gauges; Group 3-Daily values from a global coupled forecast model (NOAAs Climate Forecast System Reanalysis-CFSR); and Group 4-CFSR data with 124 local precipitation gauges. The four simulations were evaluated using multiple statistical efficiency metrics for four stream flow gauges, using: Nash-Sutcliffe coefficient (NSE), determination coefficient (R2), the ratio of the root mean square to the standard deviation of the observed data (RSR), and the percent bias (PBIAS). The Group 4 simulation performed best overall (provided the best statistical values) with results ranked as good or very good on all four efficiency metrics suggesting that using CFSR data for weather parameters other than precipitation, coupled with precipitation data from local rain gauges, can provide reasonable hydrologic responses. The second best results were obtained with Group 1, which provided good results in three of four efficiency metrics. Group 2 performed worse overall than Groups 1 and 4, probably due to uncertainty related to daily measures and a large percentage of missing data. Groups 2 and 3 were unsatisfactory according to three or four of the efficiency metrics, indicating that the choice of weather data is very important.
