* drwang@purdue.edu

Dynamic process-based plant models are computerized simulations of plant growth, development, and yield that use measurements of environmental factors and physiological processes as input data to generate predictions. However, using process-based models in wheat phenotyping programs still face challenges when parameterizing across multiple genotypes, such as (2) The need for large amounts of detailed datasets as parameters to run the simulations, many of which require destructive and disruptive sampling methods to obtain. (2)

The lack of standardized approaches needed to parameterize across extensive collections of genetically distinct, but often related, individuals that are typical of breeding populations. In this research, we developed a methodology for determining the most practical combination of remote sensing, proximal sensing, and direct measurement data to parametrize a process-based crop model for contrasting 14 wheat genotypes.

For the model parametrization, we collected a comprehensive dataset of remote, proximal, and direct measurements of a panel of wheat genotypes under three environments: well-watered, drought, and high temperature. Data were collected over two growing seasons from an experiment established at the CIMMYT's research station located in Northwestern Mexico. To facilitate data collection, we developed a framework that reduces the time needed for data collection and data processing.

Then, a decision tree approach was used to estimate key crop state variables from the available data sources, which were then used to parameterize the crop model using inversion methods. This methodology facilitated the integration of remote sensing and crop modeling to enhance wheat breeding programs and crop performance prediction.