A methodological framework for modelling the impacts of climate change on agricultural crop production at national and farm scales: Predicting maize (Zea mays L.) yields in Mexico

This thesis proposes a universally applicable methodology for the spatially-explicit assessment and communication of the impacts of climate change (CC) on crop production and food security at national and sub-national scales, supported by general circulation models (GCMs) and crop growth models (CGMs). Currently, global projections of major climatic variables extending to the full 21st century are derived from GCMs at very coarse resolutions, while CGMs provide site-specific estimates of crop yields with data from GCMs as inputs. Locally-valid spatial predictions of agricultural impacts resulting from CC are derived here by addressing this mismatch of scales through a paradigm of loose coupling GCMs to CGMs, followed by yield calibration and downscaling. The methodology is elucidated with Mexico as a case-study, maize (Zea mays L.) as a staple crop species, the IPCC SRES scenario Al as a projected path of global development, the GFDL-CM2.1 GCM as a climate predictor, the AquaCrop CGM as a simulator of yield production, site specific calibrations as an adjustment for natural yield responses to CC, geostatistical / deterministic spatial-interpolation procedures as a method of downscaling for continuous predictions, and a customized web tool as the vehicle for the efficient dissemination of information (www.mycaveat.com).;Keywords: methodology, climate change, agricultural predictive modelling, downscaling, calibration, national and farm scale, maize, Mexico.