AgMIP: Developing An Intercomparison Protocol for Global Gridded Biophysical Simulations.

Building on site-based model intercomparisons for individual crops, AgMIP is developing a protocol for intercomparisons of global gridded biophysical simulations using multiple crop models, databases, and systems, initially as part of the Intesectoral Model Intercomparison Project (ISI-MIP). The goal is to improve characterization of climate change effects on world food production and food security. The global gridded AgMIP protocol starts with completion of a matrix of inputs and processes by  participating global crop models. Documenting differences in terms of model inputs, ways of estimating initial conditions, and management assumptions can result in major differences in the model outputs. Biophysical processes embedded in global economic models will also be documented. The aim is to standardize as many of these to the extent possible (e.g., planting dates, N fertilizer, etc.) so as to enable useful comparisons of climate change scenario results.

Crop-soil model characteristics and processes in the AgMIP matrix include timestep, leaf area development, light interception, light utilization, yield formation, crop phenology, root distribution, water and heat stress, water dynamics, evapotranspiration, soil C/N model, CO₂ effects, and cultivar coefficients. The matrix also documents model inputs, their sources, and assumptions relative to source of weather data, spatial scale of inputs, land masses covered; source of soil property inputs, source of planting details, nitrogen fertilizer amounts and dates of application, crop residue management, and initial soil water, nitrate, ammonia, carbon and organic matter, carbon pools, and crop residues. It is also important to document how biophysical process and results are linked to global economic models. For example, there are inputs and assumptions that the economic modelers specify, such as how biophysical model outputs are processed to provide input to the economic models (if at all). This includes whether absolute yield levels are used or if relative yields under scenarios vs. baseline are used. Documentation of structure, inputs, and assumptions will help to interpret results across multiple global biophysical models, with the role of scale and grid size explicitly considered.