Preprints
https://doi.org/10.5194/egusphere-2023-3002
https://doi.org/10.5194/egusphere-2023-3002
21 Feb 2024
 | 21 Feb 2024
Status: this preprint is open for discussion.

A data-driven framework for assessing climatic impact-drivers in the context of food security

Marcos Roberto Benso, Roberto Fray Silva, Gabriela Gesualdo Chiquito, Antonio Mauro Saraiva, Alexandre Cláudio Botazzo Delbem, Patricia Angélica Alves Marques, and Eduardo Mario Mendiondo

Abstract. Understanding how physical climate-related hazards affect food production requires transforming climate data into relevant information for regional risk assessment. Data-driven methods can bridge this gap; however, more development must be done to create interpretable models, emphasizing regions lacking data availability. The main objective of this article was to evaluate the impact of climate risks on food security. We adopted the climatic impact-driver (CID) approach proposed by Working Group I (WGI) in the Sixth Assessment Report of the Intergovernmental Panel on Climate Change (IPCC). In this work, we used the CID framework to select the most relevant indices that drive crop yield losses and identify important thresholds for the indices. When these thresholds are exceeded, the impact probability increases. We then examine the impact of two CID types (heat and cold, and wet and dry) represented by indices of climate extremes considering the impact on different crop yield datasets, focusing on maize and soybeans in the central agro-producing municipalities in Brazil. We used the random forest model in a bootstrapping experiment to select the most relevant climate indices. Then, we applied the Shapley Additive Explanations (SHAP) with the XGBoost model explanatory analysis to identify the indices thresholds that caused impacts. We found that the mean precipitation is a highly relevant CID. However, there is a window in which crops are more vulnerable to precipitation deficit. For soybeans, in many regions of Brazil, precipitation below 80 mm/month in December, January, and February represents an increasing risk of crop yield losses. This is the end of the growing season for those regions. In the case of maize, there is a similar pattern with precipitation below 100 mm/month in April and May. Indices of extremes are relevant to represent crop yield variability. Nevertheless, including climate means remains highly relevant and recommended for studying the impact of climate risk on agriculture. Our findings contribute to a growing body of knowledge critical for informed decision-making, policy development, and adaptive strategies in response to climate change and its impact on agriculture.

Marcos Roberto Benso, Roberto Fray Silva, Gabriela Gesualdo Chiquito, Antonio Mauro Saraiva, Alexandre Cláudio Botazzo Delbem, Patricia Angélica Alves Marques, and Eduardo Mario Mendiondo

Status: open (extended)

Comment types: AC – author | RC – referee | CC – community | EC – editor | CEC – chief editor | : Report abuse
  • RC1: 'Comment on egusphere-2023-3002', Anonymous Referee #1, 27 Feb 2024 reply
Marcos Roberto Benso, Roberto Fray Silva, Gabriela Gesualdo Chiquito, Antonio Mauro Saraiva, Alexandre Cláudio Botazzo Delbem, Patricia Angélica Alves Marques, and Eduardo Mario Mendiondo
Marcos Roberto Benso, Roberto Fray Silva, Gabriela Gesualdo Chiquito, Antonio Mauro Saraiva, Alexandre Cláudio Botazzo Delbem, Patricia Angélica Alves Marques, and Eduardo Mario Mendiondo

Viewed

Total article views: 253 (including HTML, PDF, and XML)
HTML PDF XML Total Supplement BibTeX EndNote
177 66 10 253 26 6 6
  • HTML: 177
  • PDF: 66
  • XML: 10
  • Total: 253
  • Supplement: 26
  • BibTeX: 6
  • EndNote: 6
Views and downloads (calculated since 21 Feb 2024)
Cumulative views and downloads (calculated since 21 Feb 2024)

Viewed (geographical distribution)

Total article views: 265 (including HTML, PDF, and XML) Thereof 265 with geography defined and 0 with unknown origin.
Country # Views %
  • 1
1
 
 
 
 
Latest update: 12 Apr 2024
Download
Short summary
The production of food is susceptible to several climate hazards such as droughts, excessive rainfall, and heat waves. In this paper, we present a methodology that uses artificial intelligence for assessing the impact of climate risks on food production. Our methodology helps us to automatically select the most relevant indices and critical thresholds of these indices that when surpassed can increase the danger of crop yield loss.