How Multiple Science-Based Approaches Will be Used for the Future ET Puzzle.

Both ET itself and its connections to the environment are non-linear and governed by complicated and coupled biophysical processes, each of whose relative importance varies with conditions. In addition, ET models inevitably have some level of simplification and/or parameterization. It is unrealistic to expect any one model to successfully simulate ET over the large range of surface and atmospheric conditions encountered. Rather it is suggested here that the path forward involves developing and validating a suite of models, each with differing attention to various mechanisms. This is analogous to weather forecasting, where the forecaster studies the output of a group of models and decides which ones are expressing the most reliable predictions.

 Complex mechanistic models are difficult to implement; empirical approaches are not robust and have outlived their usefulness. The key is to also consider Occam’s razor, and carefully decide which processes need be considered at the minimum. Since ET can involve among other things, soil water, plant physiology and non-linear interactions with the atmosphere, and the coupling with the atmospheric boundary layer and synoptic weather patterns, variations in the attention to mechanisms make a model more or less suited for certain conditions. Three tasks loom before us. First to develop an improved set of adequately mechanistic models for the broad range of surfaces and climate conditions. Second, to properly validate the models with reliable independent estimates over a range of places and conditions. Finally to choose a set of models run in an ensemble-like fashion, much like weather forecast models are run today. In any case, people who understand the science underlying the process and models will be required to run them and more importantly, interpret the results for the users.