Modelling global impacts of climate variability and trend on maize yield during 1980–2010

Xiaomeng Yin,Guoyong Leng

DOI: https://doi.org/10.1002/joc.6792

2020-09-01

International Journal of Climatology

Abstract:<p>Enhanced understanding of historical climate impacts on crop yield is critical for adaptation and mitigations within the context of global warming. Previous impact assessments rely on statistical or process‐based models, each with its own strength and weakness. To date, a global‐scale comparison between process‐based and statistical models in assessing climate impacts on yield variability and trends is lacking. Here, we leverage both statistical and process models to estimate global impacts of climate variability and trend on maize yield between 1980–2010. To allow for reasonable comparability, we develop a novel emulator to mimic each of eight global process‐based maize models, based on which the common set of climate scenarios are used for driving both statistical and process models. Results show that climate variability controls 42% of global maize yield variations in statistical model, while large discrepancy is found in process‐based models with R² ranging from 0.22 to 0.61, especially at the country scale. Both statistical and most process models suggest that historical climate trend has led to a yield loss by 1.51%–3.80% during the period 1980–1990. As for the period 1991–2000, however, the observed negative climate impacts are only captured by two process models. In contrast to the positive climate impact in statistical model, most of process‐based models simulated negative climate effects for the period 2001–2010, due to large yield sensitivity to maximum temperature in process models. This study highlights the large discrepancies not only among the eight process‐based models but also between statistical and process‐based models in simulating yield response to climate variation and trends, which has great implications for projecting future crop yields under climate change.</p><p>This article is protected by copyright. All rights reserved.</p>

meteorology & atmospheric sciences