8 March 2018 in Earth's Future
During the last few decades, the global agricultural production has risen and technology enhancement is still contributing to yield growth. However, population growth, water crisis, deforestation, and climate change threaten the global food security. An understanding of the variables that caused past changes in crop yields can help improve future crop prediction models. In this article, we present a comprehensive global analysis of the changes in the crop yields and how they relate to different large‐scale and regional climate variables, climate change variables and technology in a unified framework. A new multilevel model for yield prediction at the country level is developed and demonstrated. The structural relationships between average yield and climate attributes as well as trends are estimated simultaneously. All countries are modeled in a single multilevel model with partial pooling to automatically group and reduce estimation uncertainties. El Niño‐southern oscillation (ENSO), Palmer drought severity index (PDSI), geopotential height anomalies (GPH), historical carbon dioxide (CO2) concentration and country‐based time series of GDP per capita as an approximation of technology measurement are used as predictors to estimate annual agricultural crop yields for each country from 1961 to 2013. Results indicate that these variables can explain the variability in historical crop yields for most of the countries and the model performs well under out‐of‐sample verifications. While some countries were not generally affected by climatic factors, PDSI and GPH acted both positively and negatively in different regions for crop yields in many countries.
1 April 2019 in Science of The Total Environment
23 March 2020 in Scientific Reports
Diagnosing potential predictability of global crop yields in the near term is of utmost importance for ensuring food supply and preventing socio-economic consequences. Previous studies suggest that a substantial proportion of global wheat yield variability depends on local climate and larger-scale ocean-atmospheric patterns. The science is however at its infancy to address whether synergistic variability and volatility (major departure from the normal) of multi-national crop yields can be potentially predicted by larger-scale climate drivers. Here, using observed data on wheat yields for 85 producing countries and climate variability from 1961–2013, we diagnose that wheat yields vary synergistically across key producing nations and can also be concurrently volatile, as a function of shared larger-scale climate drivers. We use a statistical approach called robust Principal Component Analysis (rPCA), to decouple and quantify the leading modes (PC) of global wheat yield variability where the top four PCs explain nearly 33% of the total variance. Diagnostics of PC1 indicate previous year’s local Air Temperature variability being the primary influence and the tropical Pacific Ocean being the most dominating larger-scale climate stimulus. Results also demonstrate that world-wide yield volatility has become more common in the current most decades, associating with warmer northern Pacific and Atlantic oceans, leading mostly to global supply shortages. As the world warms and extreme weather events become more common, this diagnostic analysis provides convincing evidence that concurrent variability and world-wide volatility of wheat yields can potentially be predicted, which has major socio-economic and commercial importance at the global scale, underscoring the urgency of common options in managing climate risk.
1 April 2019 in Data in Brief
We present the output data of Robust Principal Component Analysis (RPCA) applied to global crop yield variability of maize, rice, sorghum and soybean (MRSS) as presented in the publication "Climate drives variability and joint variability of global crop yields" (Najafi et al., 2019). Global maps of the correlation between all the principal components (PCs) acquired from the low rank matrix (L) of MRSS and Palmer Drought Severity Index (PDSI), air temperature anomalies (ATa) and sea surface temperature anomalies (SSTa) are provided in this article. We present co-varying countries, impacted cropland areas across global countries, and 10 global regions by climate and the association between PCs and multiple atmospheric and oceanic indices. Moreover, the joint dependency between PCs of MRSS yields are presented using two different approaches.
10 February 2019 in arXiv
Drought is one of the most devastating environmental disasters. Analyzing historical changes in climate extremes is critical for mitigating its adverse impacts in the future. In the present study, the spatial and temporal characteristics of the global severe droughts using Palmer Drought Intensity Index (PDSI) from 1900 to 2014 are explored. K-means clustering is implemented to partition the extreme negative PDSI values. The global extreme droughts magnitude around the world from 1950 to 1980 were less intense compared to the other decades. In 2012, the largest areas around the world, especially Canada, experienced their most severe historical droughts. Results show that the most recent extreme droughts occurred in some regions such as the North of Canada, central regions of the US, Southwest of Europe and Southeast Asia. We found that after 1980, the spatial extent of the regions that experienced extreme drought have increased substantially.