Combining Remote Sensing and Crop Models: It's Like Baking a Cake.

The concept of "within-season calibration" has proved to be an effective way to bring information on crop growth from remote sensing into simulation models. Key to this approach is the process of "parameterization". Parameterization has been used in a wide variety of models, and represents a method for implicitly incorporating the detailed effects of environmental phenomena into model processes at spatial or temporal scales that are greater than the scales at which the environmental phenomena actually occur. A good example is the use of an "extinction coefficient" in the Beer's Law description of light absorption by a plant canopy. The extinction coefficient implicitly incorporated the detailed effects of leaf geometry, canopy architecture, leaf optical properties, and other factors into a single parameter that can be practically applied to estimating canopy light absorption. Parameterization also exhibits another characteristic that I have termed the "folding-in phenomenon." This descibes the ability of parameters to absorb the effects of environmental phenomena so that the phenomena themselves do not have to be explicitly modeled. For example, the value of a parameter that determines leaf growth in a model can absorb the effects of phenomena like water stress (perhaps through within-season calibration based on remotely sensed observations of canopy growth) without an explicit simulation of water stress processes in the model. This allows models to be simpler in their construction yet retain reasonable accuracy. The folding-in phenomenon is analogous to baking a cake because the individual qualities of the ingredients of a cake give the cake its final flavor although the individual qualities of the ingredients are lost when they are folded into the batter. Numerous projects by Paul Doraiswamy and this presenter have made good use of these modeling practices in simulating agricultural crops and natural vegetation.