Spatially distributing monthly reference evapotranspiration and pan evaporation considering topographic influences

Many hydrological models engage spatially distributed measures of 'potential evapotranspiration' (ETpot). The reliability and utility of the physically based Penman-Monteith approach to generate ETpot has been recently advocated. Assuming land-surface conditions, spatial surfaces of reference evapotranspiration (ETO) can be generated taking into account the topographic influence of forcing meteorological variables. This was performed in this paper by spatially interpolating maximum (T-max) and minimum (T-min) air temperatures, wind speed (u) and vapor pressure (ea), using a spline model with a linear sub-model dependency on elevation, and modelling the radiation environment, taking topography (i.e., elevation, slope and aspect) into account, prior to calculating ETO at each grid-cell. In accordance with previous research, resultant lapse rates showed a strong seasonal pattern; values were steeper in summer than winter and those for Tma, were steeper than for T-min. Monthly mean T-max lapse rates varied from -3.01 degrees C km(-1) in winter to -7.69 degrees C km(-1) in summer, with T-min lapse rates ranging from -2.79 degrees C km(-1) in winter, to -6.64 degrees C km(-1) in summer. Monthly climatotogies of the near-surface elevation-dependence (NSED) for u and ea also showed strong seasonal values. NSED of u varied from 2.01 ms(-1) km(-1) in winter reducing to 0.75 ms(-1) km(-1) in summer. The NSED for e(a) ranged from -0.08 kPa km(-1) in winter to -0.64 kPa km(-1) in summer. For a 252-month sequence from 1980 through 2000, spatial surfaces of ET0 with a 100 m resolution for the 113,000 km(2) study site located in the Loess Plateau, China were generated using an 'interpolate-then-catculate' approach. Resultant ET0 values varied from about 20 mm month(-1) in winter to over 150 mm month(-1) in summer. In order to assess the reliability of these ETO surfaces, pan evaporation (Ep,,,) was also spatially interpolated and from these a set of pan coefficient (K-pan - a unitless ratio defined as ET0/E-pan) surfaces were calculated. Spatio-temporally averaged K-pan values for the study site varied from 0.44 in April to 0.65 in late summer. K-pan values were in agreement with another study using a Chinese 20 cm diameter micro-pan, and, as expected, were tower than other values documented using a Class A pan. The influence of topography, especially aspect, was seen on the resultant ETO and K-pan, but not E-pan, surfaces. Sensitivity analysis showed that results were particularly stable in the hydrologically active portion of the year extending from March to October, inclusive. This study demonstrated that high spatial resolution monthly surfaces of ETO can be spatially modelled while taking into account the influence of topography on the forcing variables. (C) 2007 Elsevier B.V. All rights reserved.