Periodic Reporting for period 1 - CLIM4CROP (Climate monitoring and seasonal forecast for global crop production)
Periodo di rendicontazione: 2018-09-09 al 2020-09-08
The occurrence of and changes in climatic extremes constitute a great concern to food security. Estimating the expected crop yield variability a few months in advance would allow reducing socio-economic impacts through short-term adaptation and response to climate variability and change. However, seasonal prediction of climate-driven impacts on agriculture is still in its infancy.
The overall scientific goal of the project entitled “Climate monitoring and seasonal forecast for global crop production” (CLIM4CROP project H2020-MSCA-IF-2016-740073) was to explore how best to exploit seasonal forecasts for crop management decision making on a global scale making use of the latest advances in climate and crop sciences.
With this project, we have improved the knowledge on the limits in global datasets of climate observations relevant to agriculture and developed innovative statistical models to analyse the climate-driven impact on crops. In addition, the implementation of the project along with the advanced training gained at BSC have allowed the MSCA fellow Dr. Marco Turco to reach one of the most important milestones in his career: i.e. scientific independence. Indeed, the project PREDFIRE proposed by the fellow has been granted from the Spanish Ministry of Science, Innovation and Universities, that allows the fellow to become an autonomous scientist. The MSCA fellowship has undoubtedly impacted in attaining this job position.
The overall scientific goal of the project entitled “Climate monitoring and seasonal forecast for global crop production” (CLIM4CROP project H2020-MSCA-IF-2016-740073) was to explore how best to exploit seasonal forecasts for crop management decision making on a global scale making use of the latest advances in climate and crop sciences.
With this project, we have improved the knowledge on the limits in global datasets of climate observations relevant to agriculture and developed innovative statistical models to analyse the climate-driven impact on crops. In addition, the implementation of the project along with the advanced training gained at BSC have allowed the MSCA fellow Dr. Marco Turco to reach one of the most important milestones in his career: i.e. scientific independence. Indeed, the project PREDFIRE proposed by the fellow has been granted from the Spanish Ministry of Science, Innovation and Universities, that allows the fellow to become an autonomous scientist. The MSCA fellowship has undoubtedly impacted in attaining this job position.
The work carried out by the beneficiary (Dr. Marco Turco under the supervision of Prof. Francisco Javier Doblas-Reyes; Barcelona Supercomputing Center - BSC) mainly consisted of (i) compilation of all the available precipitation data that are currently publicly available providing data for the last three decades, and being updated every month, i.e. that are available in near-real time; (ii) creation of new drought probabilistic monitoring tool in which an ensemble of observations-based datasets is used to obtain the real-time estimates with associated uncertainty; and (iii) development of global climate-crop model for maize, rice, soybean and wheat. These works have led to a series of positive and innovative results: this included publications submitted to peer-reviewed journals and – most importantly – paved the way for new lines of research with multiple collaborations worldwide (i.e. networking) for BSC.
Accurate and timely drought information is essential to move from post-crisis to pre-impact drought risk management. A number of drought datasets is already available. They cover the last three decades and provide data in near-real time (using different sources), but they are all “deterministic” (i.e. single realisation), and results partly differ between them. Within CLIM4CROP we first evaluated the quality of long-term and continuous climate data for timely meteorological drought monitoring considering the Standardized Precipitation Index. Then we developed a new global land gridded dataset, operationally updated every month, to monitoring DROught by using a Probabilistic approach (DROP) inspired by the multi-model approach in weather and climate prediction. We presented a monitoring tool in which an ensemble of observations-based datasets is used to obtain the best near-real time estimate with its associated uncertainty. This approach makes the most of the available information and brings it to the end-users. The high-quality and probabilistic information provided by DROP is useful for monitoring applications, and may help to develop global policy decisions on adaptation priorities in alleviating drought impacts, especially in countries where meteorological monitoring is still challenging. Currently, the paper describing the DROP dataset is under review in the Bulletin of the American Meteorological Society; i.e. fulfilling the dissemination task of the project.
The links between crops and heat/water stress drivers have been analysed to better understand their interactions and develop statistical models. These models will be used to explore the seasonal predictability for crop yields. We have used historical yield data of the four principal crops worldwide (maize, rice, soybean and wheat) from the Global Dataset of Historical Yield (Iizumi 2017), a recent developed gridded dataset never used before for this kind of study. Importantly, we established a collaboration with the developer of this dataset, Dr. Iizumi, an internationally recognized agro-meteorologist.
The historical data on crop and potential climate predictors have been used to calibrate parsimonious regression models, providing a computationally inexpensive alternative to process-based models that usually require as input several variables at very fine scale where seasonal forecasts are not yet skilful. The fellow was trained by staff of the JRC in the development of these models, follow and extend the models developed by Zampieri et al. (2017, 2018, 2019a) and Ceglar et al. (2018).
The novelties of the empirical model developed relies mainly in (i) taking into account the potential effect of antecedent (i.e. within the growing season) climate conditions on crop yield and (ii) calibrate the model to achieve out-of-sample crop predictions from the knowledge of the predictor data outside the period used to train the model, adopting a leave-one-out cross-validation method. That is, the calibration of the crop-climate models and their evaluation are done by using cross-validation in order to evaluate the predictions as if they were done operationally. Specifically, merging observational information (for the months previous to the harvest months) with seasonal forecasts for the rest of the crop growth calendar (e.g. anthesis and harvesting stages) is a special feature of our approach that can contributes to increase crop predictability, making the most of the best information available to the users. This is especially useful over areas where the performance of the dynamical forecast systems is still affected by significant errors (e.g. Europe). Preliminary results suggest that the climate-crop lagged relationships provide a substantial contribution to the development of a seasonal forecast system allowing more efficient crop management.
CLIM4CROP was planned in such a way as to offer a unique opportunity to reinforce and expand the multi-sectorial experience of the fellow, linking his background in meteorology and climate change studies and the theoretical and practical requirements for climate prediction applications in agriculture. Hence, this MSCA has also addressed the training objectives described in the proposal on seasonal prediction, where the expert guidance of the supervisor at BSC (F.J. Doblas-Reyes) has been essential for the successful implementation of the work plan. Also, the training offered by the collaborators of the JRC on crop yield impacts has allowed to learning about agriculture production and food security, and to widen his competences in the statistical aspects regarding climate-impact models. Also, scientific challenges in the context of CLIM4CROP, such as highly cooperative research, have ensured the two-way transfer of derived new knowledge strengthening the skills of all participants.
The links between crops and heat/water stress drivers have been analysed to better understand their interactions and develop statistical models. These models will be used to explore the seasonal predictability for crop yields. We have used historical yield data of the four principal crops worldwide (maize, rice, soybean and wheat) from the Global Dataset of Historical Yield (Iizumi 2017), a recent developed gridded dataset never used before for this kind of study. Importantly, we established a collaboration with the developer of this dataset, Dr. Iizumi, an internationally recognized agro-meteorologist.
The historical data on crop and potential climate predictors have been used to calibrate parsimonious regression models, providing a computationally inexpensive alternative to process-based models that usually require as input several variables at very fine scale where seasonal forecasts are not yet skilful. The fellow was trained by staff of the JRC in the development of these models, follow and extend the models developed by Zampieri et al. (2017, 2018, 2019a) and Ceglar et al. (2018).
The novelties of the empirical model developed relies mainly in (i) taking into account the potential effect of antecedent (i.e. within the growing season) climate conditions on crop yield and (ii) calibrate the model to achieve out-of-sample crop predictions from the knowledge of the predictor data outside the period used to train the model, adopting a leave-one-out cross-validation method. That is, the calibration of the crop-climate models and their evaluation are done by using cross-validation in order to evaluate the predictions as if they were done operationally. Specifically, merging observational information (for the months previous to the harvest months) with seasonal forecasts for the rest of the crop growth calendar (e.g. anthesis and harvesting stages) is a special feature of our approach that can contributes to increase crop predictability, making the most of the best information available to the users. This is especially useful over areas where the performance of the dynamical forecast systems is still affected by significant errors (e.g. Europe). Preliminary results suggest that the climate-crop lagged relationships provide a substantial contribution to the development of a seasonal forecast system allowing more efficient crop management.
CLIM4CROP was planned in such a way as to offer a unique opportunity to reinforce and expand the multi-sectorial experience of the fellow, linking his background in meteorology and climate change studies and the theoretical and practical requirements for climate prediction applications in agriculture. Hence, this MSCA has also addressed the training objectives described in the proposal on seasonal prediction, where the expert guidance of the supervisor at BSC (F.J. Doblas-Reyes) has been essential for the successful implementation of the work plan. Also, the training offered by the collaborators of the JRC on crop yield impacts has allowed to learning about agriculture production and food security, and to widen his competences in the statistical aspects regarding climate-impact models. Also, scientific challenges in the context of CLIM4CROP, such as highly cooperative research, have ensured the two-way transfer of derived new knowledge strengthening the skills of all participants.