Defining a best practice methodology for modeling the environmental performance of agriculture

Abstract Robust quantification of the environmental performance of agricultural management practices is critical both for ensuring regulatory compliance and for creating accountability in voluntary environmental markets and corporate sustainability commitments. Because environmental impacts cannot be measured under all conditions and on all farms, models are required. However, models must be used appropriately if predictions of environmental performance are to be reliable. To assist policymakers and stakeholders, we define a 7-step process for model selection and use, and present a case study applying this 7-step process to greenhouse gas emissions from corn ( Zea mays L.) fields in the USA. Based on this case study and other examples from the literature, we suggest that all models are limited by the data available to validate them for different combinations of cropping systems, management practices, site conditions, and types of environmental performance. Additionally, both statistical and process models are much more reliable for making predictions of environmental performance for multiple fields and years than for predictions of a single location and year. We suggest that using this 7-step process will help improve predictions of environmental performance for regulatory and voluntary purposes at local, state, and national scales.