Uncertainties of Water Fluxes in Soil–Vegetation–Atmosphere Transfer Models: Inverting Surface Soil Moisture and Evapotranspiration Retrieved from Remote Sensing

摘要

Effective hydraulic parameters of soil–vegetation–atmosphere transfer (SVAT) models can be derived by inverting observed surface soil moisture, θobs, and evapotranspiration, ETobs, retrieved from remote sensing. We investigated the uncertainties in simulating the water fluxes of contrasting hydroclimatic scenarios for which it was assumed that θobs had a RMSE of 0.04 m3 m−3 (Δθobs) and ETobs had a relative error of 20% (ΔETobs). The correlation of the uncertainties in the simulated water fluxes (ΔWFsim) with Δθobs and ΔETobs was derived with the proposed Uncertainty Simulator Algorithm. The results show that ΔWFsim is influenced by climate and increases when the climate is drier. The uncertainty in estimated root-zone θ was found to be correlated with Δθobs. The prediction of evaporation contained large uncertainties and was correlated with the actual/potential evapotranspiration ratio. The uncertainties in transpiration under dry climates were high and were correlated with ΔETobs; however, the uncertainty under wet climates was insignificant. The uncertainties in groundwater recharge under dry climates were large but were reduced under wet climates. Furthermore, uncertainties in groundwater recharge were correlated with ΔETobs but not with Δθobs. In general, the ΔWFsim increases as (i) climate gets drier, (ii) texture gets coarser, or (iii) roots grow deeper. The uncertainty in recharge is explained by soil moisture and transpiration decoupling. Soil moisture decoupling occurs when the information provided by surface θ is no longer representative of root-zone θ. Transpiration decoupling occurs when there is substantially more water storage at depth. We propose methodology to reduce the nonuniqueness of the inverted hydraulic parameters.