Search Results
Impact of extreme weather conditions on European crop production in 2018
Extreme weather increases the risk of large-scale crop failure. The mechanisms involved are complex and intertwined, hence undermining the identification of simple adaptation levers to help improve the resilience of agricultural production. Based on more than 82 000 yield data reported at the regional level in 17 European countries, we assess how climate affected the yields of nine crop species. Using machine learning models, we analyzed historical yield data since 1901 and then focus on 2018, which has experienced a multiplicity and a diversity of atypical extreme climatic conditions. Machine learning models explain up to 65% of historical yield anomalies. We find that both extremes in temperature and precipitation are associated with negative yield anomalies, but with varying impacts in different parts of Europe. In 2018, Northern and Eastern Europe experienced multiple and simultaneous crop failures - among the highest observed in recent decades. These yield losses were associated with extremely low rainfalls in combination with high temperatures between March and August 2018. However, the higher than usual yields recorded in Southern Europe - caused by favourable spring rainfall conditions - nearly offset the large decrease in Northern European crop production. Our results outline the importance of considering single and compound climate extremes to analyse the causes of yield losses in Europe. We found no clear upward or downward trend in the frequency of extreme yield losses for any of the considered crops between 1990 and 2018. This article is part of the theme issue 'Impacts of the 2018 severe drought and heatwave in Europe: from site to continental scale'. © 2020 The Authors.
The effects of climate extremes on global agricultural yields
Climate extremes, such as droughts or heat waves, can lead to harvest failures and threaten the livelihoods of agricultural producers and the food security of communities worldwide. Improving our understanding of their impacts on crop yields is crucial to enhance the resilience of the global food system. This study analyses, to our knowledge for the first time, the impacts of climate extremes on yield anomalies of maize, soybeans, rice and spring wheat at the global scale using sub-national yield data and applying a machine-learning algorithm. We find that growing season climate factors—including mean climate as well as climate extremes—explain 20%–49% of the variance of yield anomalies (the range describes the differences between crop types), with 18%–43% of the explained variance attributable to climate extremes, depending on crop type. Temperature-related extremes show a stronger association with yield anomalies than precipitation-related factors, while irrigation partly mitigates negative effects of high temperature extremes. We developed a composite indicator to identify hotspot regions that are critical for global production and particularly susceptible to the effects of climate extremes. These regions include North America for maize, spring wheat and soy production, Asia in the case of maize and rice production as well as Europe for spring wheat production. Our study highlights the importance of considering climate extremes for agricultural predictions and adaptation planning and provides an overview of critical regions that are most susceptible to variations in growing season climate and climate extremes.
Estimating near-surface air temperature across Israel using a machine learning based hybrid approach
Rising global temperatures over the last decades have increased heat exposure among populations worldwide. An accurate estimate of the resulting impacts on human health demands temporally explicit and spatially resolved monitoring of near-surface air temperature (Ta). Neither ground-based nor satellite-borne observations can achieve this individually, but the combination of the two provides synergistic opportunities. In this study, we propose a two-stage machine learning-based hybrid model to estimate 1 × 1 km2 gridded intra-daily Ta from surface skin temperature (Ts) across the complex terrain of Israel during 2004–2016. We first applied a random forest (RF) regression model to impute missing Ts from the Moderate Resolution Imaging Spectroradiometer (MODIS) Aqua and Terra satellites, integrating Ts from the geostationary Spinning Enhanced Visible and InfraRed Imager (SEVIRI) satellite and synoptic variables from European Centre for Medium-Range Weather Forecasts' (ECMWF) ERA5 reanalysis data sets. The imputed Ts are in turn fed into the Stage 2 RF-based model to estimate Ta at the satellite overpass hours of each day. We evaluated the model's performance applying out-of-sample fivefold cross validation. Both stages of the hybrid model perform very well with out-of-sample fivefold cross validated R2 of 0.99 and 0.96, MAE of 0.42°C and 1.12°C, and RMSE of 0.65°C and 1.58°C (Stage 1: imputation of Ts, and Stage 2: estimation of Ta from Ts, respectively). The newly proposed model provides excellent computationally efficient estimation of near-surface air temperature at high resolution in both space and time, which helps further minimize exposure misclassification in epidemiological studies. © 2020 The Authors. International Journal of Climatology published by John Wiley & Sons Ltd on behalf of the Royal Meteorological Society.
How the Selection of Training Data and Modeling Approach Affects the Estimation of Ammonia Emissions from a Naturally Ventilated Dairy Barn—Classical Statistics versus Machine Learning
Environmental protection efforts can only be effective in the long term with a reliable quantification of pollutant gas emissions as a first step to mitigation. Measurement and analysis strategies must permit the accurate extrapolation of emission values. We systematically analyzed the added value of applying modern machine learning methods in the process of monitoring emissions from naturally ventilated livestock buildings to the atmosphere. We considered almost 40 weeks of hourly emission values from a naturally ventilated dairy cattle barn in Northern Germany. We compared model predictions using 27 different scenarios of temporal sampling, multiple measures of model accuracy, and eight different regression approaches. The error of the predicted emission values with the tested measurement protocols was, on average, well below 20%. The sensitivity of the prediction to the selected training dataset was worse for the ordinary multilinear regression. Gradient boosting and random forests provided the most accurate and robust emission value predictions, accompanied by the second-smallest model errors. Most of the highly ranked scenarios involved six measurement periods, while the scenario with the best overall performance was: One measurement period in summer and three in the transition periods, each lasting for 14 days.
Combining statistical and machine learning methods to explore German students’ attitudes towards ICT in PISA
In our age of big data and growing computational power, versatility in data analysis is important. This study presents a flexible way to combine statistics and machine learning for data analysis of a large-scale educational survey. The authors used statistical and machine learning methods to explore German students’ attitudes towards information and communication technology (ICT) in relation to mathematical and scientific literacy measured by the Programme for International Student Assessment (PISA) in 2015 and 2018. Implementations of the random forest (RF) algorithm were applied to impute missing data and to predict students’ proficiency levels in mathematics and science. Hierarchical linear models (HLM) were built to explore relationships between attitudes towards ICT and mathematical and scientific literacy with the focus on the nested structure of the data. ICT autonomy was an important variable in RF models, and associations between this attitude and literacy scores in HLM were significant and positive, while for other ICT attitudes the associations were negative (ICT in social interaction) or non-significant (ICT competence and ICT interest). The need for further research on ICT autonomy is discussed, and benefits of combining statistical and machine learning approaches are outlined.