Measurement and modelling evaporation for irrigated crops in north-west Mexico

The surface energy balance and near-surface weather variables were measured during the complete growth cycle of cotton and wheat crops planted in two successive seasons in 1995-1996 at a field site in the Yaqui Valley in Sonora, mexico. These data were analysed in terms of a set of simple evapotranspiration models, which included the Penman-Monteith equation applied at both the hourly and daily time-scales and four semi-empirical simplifications of that equation. The field data were used to investigate and specify seasonal variations in the surface parameters (crop coefficient or surface resistance) appropriate to these models. There was a marked difference between the effective value of surface parameters for the (cotton) row crop and the (wheat) continuous-cover crop during and immediately after periods of flood irrigation. This difference reflects the fact that wet soil and remnant pools of irrigation water are well exposed to the atmosphere for flood-irrigated row crops and that evapotranspiration rates are enhanced. It proved possible to incorporate a description of this phenomenon by including a simple, additional term in an otherwise traditional model of the seasonal variation in crop fractor or surface resistance. Once the parameters in the various models had been optimized against field data, all the models provided reasonable descriptions of the measured evapotranspiration. However, the Penman-Monteith equation applied at the hourly time-scale performed least well and required greater parameterization; hence, its use in this application is not justified. At the daily time-scale, use of the Penman-Monteith equation or Shuttleworth's reference crop evaporation (with a crop factor) is arguably preferable to the others in that, having more realism, it provides a description in which the (calibrated) surface parameters are a purer measure of crop control. However, in terms of descriptive ability after local calibration, the Penman, Priestley-Taylor, and Makkink models proved superior in this study. The good performance of the Makkink formulation is particularly encouraging because this equation requires only incoming solar radiation, which can be readily estimated from remotely sensed data.