Estimation of evapotranspiration of a salt marsh in southern South America with coupled Penman-Monteith and surface resistance models

Abstract

One of the most recommended method to estimate evapotranspiration (ET) of vegetated surfaces with different soil moisture conditions is the Penman-Monteith equation (PM). Canopy and soil conditions are parameterized through the surface resistance or conductance, while the contribution of the canopy to ET is measured by the canopy resistance. The study of natural ecosystems has gained interest because of its importance in water and carbon cycles. However, unlike monocultures, natural environments are composed of a mixture of species that make the estimation of ET with PM troublesome. This feature makes them suitable for ET estimation considering the contribution of both, the canopy and the soil represented by the surface resistance (rs), or the contribution of the canopy, represented by the canopy resistance (rc). This work aims to model the surface and canopy resistances using conventional meteorological, biological and pedological variables observed at a salt marsh used for livestock production in Buenos Aires province, Argentina. Twelve models (M1 to M12) based on the net solar radiation (Rn), air temperature (Ta), air relative humidity (RH), surface wind velocity (U), dew point departure (Dp), aerodynamic resistance (ra), leaf area index (LAI) and volumetric soil water content (ϑs) were obtained using two different regression methodologies. Surface resistances during daytime were calculated inverting the PM equation with ET fluxes measured with the eddy covariance method. PM-derived rs varied between 20 and 1000 s m−1, with a median of 137 s m−1. From 1620 observations, 468 were used for model calibration while 1152 for model validation. M5 and M11 with Rn, RH, ra, LAI predictor variables were the best models with 80.8 s m−1 root mean square error, 0.51 determination coefficient, 0.69 and 0.65 index of agreement, respectively. The modelled resistances allowed the estimation of latent heat fluxes with a root mean quadratic error varying from 60.7 to 69.5 W m-2. These results show the possibility to achieve rs from a minimum set of variables easily measured in the field which in turn, allows to estimate the ET of salt marsh ecosystems with scarce meteorological information.