Characterizing Water Flow in Vegetated Lysimeters with Stable Water Isotopes and Modeling.

We have used stable water isotopes (δ O, δ H) in combination with lumped-parameter modeling for characterizing unsaturated flow in two lysimeters vegetated with maize. The lysimeters contained undisturbed soil cores dominated by sandy gravel (Ly1) and clayey sandy silt (Ly2). Stable water isotopes were analyzed in precipitation and lysimeter outflow water over about 3 years. The mean transit time of water T and dispersion parameter P , obtained from modeling, were higher for the silt soil in Ly2 than for the gravel soil in Ly1 (T of 362 vs. 129 d, P of 0.7 vs. 0.12). The consideration of preferential flow (PF) paths could substantially improve the model curve fits, with 13 and 11% contribution of PF for Ly1 and Ly2 as best estimates. Different assumptions were compared to estimate the input function, that is, stable water isotope content in the recharging water. Using the isotopic composition of precipitation as input (no modification) resulted in reasonable model estimations. Best model fits for the entire observation were obtained by weighting the recharging isotopes according to average precipitation within periods of 3 and 6 months, in correspondence to changing vegetation phases and seasonal influences. Input functions that consider actual evapotranspiration could significantly improve modeling at some periods, however, this led to deviations between modeled and observed δ O at other periods. This may indicate the influence of variable flow, so that dividing the whole observation period into hydraulically characteristic sub-periods for lumped-parameter modeling (which implements steady-state flow) is recommended for possible further improvement.