Abstract: Working with ensembles of crop models is a recent but important development in crop modeling which promises to lead to better uncertainty estimates for model projections and predictions, better predictions using the ensemble mean or median, and closer collaboration within the modeling community. There are numerous open questions about the best way to create and analyze such ensembles. Much can be learned from the field of climate modeling, given its much longer experience with ensembles. We draw on that experience to identify questions and make propositions that should help make ensemble modeling with crop models more rigorous and informative. The propositions include defining criteria for acceptance of models in a crop MME, exploring criteria for evaluating the degree of relatedness of models in a MME, studying the effect of number of models in the ensemble, development of a statistical model of model sampling, creation of a repository for MME results, studies of possible differential weighting of models in an ensemble, creation of single model ensembles based on sampling from the uncertainty distribution of parameter values or inputs specifically oriented toward uncertainty estimation, the creation of super ensembles that sample more than one source of uncertainty, the analysis of super ensemble results to obtain information on total uncertainty and the separate contributions of different sources of uncertainty and finally further investigation of the use of the multi-model mean or median as a predictor.

Abstract: Ensembles of process-based crop models are increasingly used to simulate crop growth for scenarios of temperature and/or precipitation changes corresponding to different projections of atmospheric CO2 concentrations. This approach generates large datasets with thousands of simulated crop yield data. Such datasets potentially provide new information but it is difficult to summarize them in a useful way due to their structural complexities. An associated issue is that it is not straightforward to compare crops and to interpolate the results to alternative climate scenarios not initially included in the simulation protocols. Here we demonstrate that statistical models based on random-coefficient regressions are able to emulate ensembles of process-based crop models. An important advantage of the proposed statistical models is that they can interpolate between temperature levels and between CO2 concentration levels, and can thus be used to calculate temperature and [CO2] thresholds leading to yield loss or yield gain, without rerunning the original complex crop models. Our approach is illustrated with three yield datasets simulated by 19 maize models, 26 wheat models, and 13 rice models. Several statistical models are fitted to these datasets, and are then used to analyze the variability of the yield response to [CO2] and temperature. Based on our results, we show that, for wheat, a [CO2] increase is likely to outweigh the negative effect of a temperature increase of +2 degrees C in the considered sites. Compared to wheat, required levels of [CO2] increase are much higher for maize, and intermediate for rice. For all crops, uncertainties in simulating climate change impacts increase more with temperature than with elevated [CO2].

Abstract: Highlights
• 14 criteria for use of crop models in assessments of impacts, adaptation and risk • Working with stakeholders to identify timing of risks is key to risk assessments. • Multiple methods needed to critically assess the use of climate model output • Increasing transparency and inter-comparability needed in risk assessments
Abstract
Crop models are used for an increasingly broad range of applications, with a commensurate proliferation of methods. Careful framing of research questions and development of targeted and appropriate methods are therefore increasingly important. In conjunction with the other authors in this special issue, we have developed a set of criteria for use of crop models in assessments of impacts, adaptation and risk. Our analysis drew on the other papers in this special issue, and on our experience in the UK Climate Change Risk Assessment 2017 and the MACSUR, AgMIP and ISIMIP projects. The criteria were used to assess how improvements could be made to the framing of climate change risks, and to outline the good practice and new developments that are needed to improve risk assessment. Key areas of good practice include: i. the development, running and documentation of crop models, with attention given to issues of spatial scale and complexity; ii. the methods used to form crop-climate ensembles, which can be based on model skill and/or spread; iii. the methods used to assess adaptation, which need broadening to account for technological development and to reflect the full range options available. The analysis highlights the limitations of focussing only on projections of future impacts and adaptation options using pre-determined time slices. Whilst this long-standing approach may remain an essential component of risk assessments, we identify three further key components: 1. Working with stakeholders to identify the timing of risks. What are the key vulnerabilities of food systems and what does crop-climate modelling tell us about when those systems are at risk? 2. Use of multiple methods that critically assess the use of climate model output and avoid any presumption that analyses should begin and end with gridded output. 3. Increasing transparency and inter-comparability in risk assessments. Whilst studies frequently produce ranges that quantify uncertainty, the assumptions underlying these ranges are not always clear. We suggest that the contingency of results upon assumptions is made explicit via a common uncertainty reporting format; and/or that studies are assessed against a set of criteria, such as those presented in this paper.