Quantifying the impact of input data uncertainties in regional scale crop-climate modeling

Watson, James and Challinor, Andrew (2011). Quantifying the impact of input data uncertainties in regional scale crop-climate modeling. In: American Geophysical Union (AGU) Fall Meeting 2011, San Francisco, CA, USA, (). 5-9 December, 2011.

Author Watson, James
Challinor, Andrew
Title of paper Quantifying the impact of input data uncertainties in regional scale crop-climate modeling
Conference name American Geophysical Union (AGU) Fall Meeting 2011
Conference location San Francisco, CA, USA
Conference dates 5-9 December, 2011
Publication Year 2011
Sub-type Published abstract
Open Access Status
Language eng
Formatted Abstract/Summary
Understanding the effects that future climate conditions will have on global food security is a key challenge for the 21st century. Global climate models (GCMs) provide a range of possible future scenarios. Regional-scale crop models such as GLAM use the output of these climate models, along with yield calibration data, to project possible future crop production scenarios. However, the skill of such regional-scale crop models is limited by the quality of the calibration and weather data used to constrain them. The skill of climate models in reproducing surface properties such as mean temperature and daily rainfall patterns is critical for the simulation of crop yield, as is the ability for the available yield calibration data to capture productivity trends for the given region. However, the impact of input data errors on the skill of regional scale crop models has not been systematically quantified. In this study, we assess the performance of the GLAM crop model when errors are introduced into calibration and weather inputs, using a well-understood historical (1966-1989) scenario of groundnut production in the Gujarat region of India. We analyze the effect of seasonal and climatic perturbations in crop yield and temperature inputs by applying perturbations uniformly across seasonal and climatic (i.e., across the entire 24 year period) timescales. Also, we compare daily, seasonal and climatic perturbations of rainfall inputs. To investigate the importance of temporal information in each of these inputs, we also analyze the effect of shuffling input values within each season, shuffling years while keeping within-year values intact, and shuffling values across the entire time period. By comparing the impact that simulated errors and data shuffling have on GLAM's skill for each input type, we find that the relative impact of seasonal and climatic perturbations is highly dependent on the input variable being considered, and that there are significant differences in model performance depending on which timescale input values are shuffled. These results indicate the relative importance of accuracy and temporal information for yield, rainfall and temperature inputs when modeling at the regional scale, and provide a basis for assessing whether, at this scale, given crop model inputs are fit for purpose.
Q-Index Code EX
Q-Index Status Provisional Code
Institutional Status Non-UQ
Additional Notes Abstract #GC11A-0902

Document type: Conference Paper
Collection: Queensland Alliance for Agriculture and Food Innovation
 
Versions
Version Filter Type
Citation counts: Google Scholar Search Google Scholar
Created: Tue, 10 Mar 2015, 13:49:17 EST by James Watson on behalf of Centre for Plant Science