Estimation of regional groundwater recharge using data from a distributed soil moisture network

Calibrating one-dimensional soil-vegetation-atmosphere models against soil moisture measurements from stations arranged in a network representing the spatial variation in land use, precipitation, and soil texture has proved capable of providing reliable estimates of continuous groundwater recharge and actual evapotranspiration at land use and catchment scales. Regional groundwater recharge and actual evapotranspiration were estimated by calibrating the one-dimensional soil-vegetation-atmosphere transfer model Daisy against soil moisture measurements from 30 stations and at 3 depths located within a 1050 km ² subcatchment of the Danish hydrological observatory HOBE. Thirty models were constructed considering the local climate, soil texture, land use, and field practice. First estimates of the hydraulic parameters were obtained from textural data using a pedotransfer function. On the basis of sensitivity analysis, hydraulic conductivity ks and van Genuchten parameter n were found to be most sensitive, and these two parameters were therefore subject to calibration at each site using the parameter estimation code PEST. From the calibrated models, the regional variation of evapotranspiration and groundwater recharge was predicted and tested against local measurements, giving annual catchment scale values of 474 and 505 mm, respectively, for the period 2009 to 2011. These values corresponded well with comparable field observations. Various formulations of effective parameterizations were tested. Effective parameters of ks and n for forest, heath, and agriculture found by autocalibration against average soil moisture measurements of the three land cover types provided evapotranspiration and groundwater recharge estimates comparable to individual field observations (stream gauge and eddy covariance [EC] data).