Analysis of spatial variability of near-surface soilmoisture to increase rainfall-runoff modellingaccuracy in SW Hungary

Abstract

Abstract Between September 5, 2008 and September 5,2009, near-surface soil moisture time series were collectedin the northern part of a 1.7 km2 watershed in SWHungaryat 14 monitoring locations using a portable TDR-300 soilmoisture sensor. The objectives of this study are to increasethe accuracy of soil moisture measurement at watershedscale, to improve flood forecasting accuracy, and to optimizesoil moisture sensor density. According to our results, in 10 of 13 cases, a strong correlationexists between the measured soil moisture dataof Station 5 and all other monitoring stations; Station 5is considered representative for the entire watershed. Logically,the selection of the location of the representativemeasurement point(s) is essential for obtaining representativeand accurate soil moisture values for the given watershed.This could be done by (i) employing monitoringstations of higher number at the exploratory phase of themonitoring, (ii) mapping soil physical properties at watershedscale, and (iii) running cross-relational statisticalanalyses on the obtained data. Our findings indicate that increasing the number of soilmoisture data points available for interpolation increasesthe accuracy of watershed-scale soil moisture estimation.The data set used for interpolation (and estimation ofmean antecedent soil moisture values) could be improved(thus, having a higher number of data points) by selectingpoints of similar properties to the measurement pointsfrom the DEM and soil databases. By using a higher numberof data points for interpolation, both interpolation accuracyand spatial resolution have increased for the measuredsoil moisture values for the Pósa Valley.