Session 2.8: Information Technologies and Data
For a complete list of all of the workshop abstracts click here (PDF).
If you are presenting a poster, please view the Poster Specifications here
Session 2.8: Oral Presentation
Title: An integrated interdisciplinary modelling system of climate change impacts on agriculture in support of adaptation planning: MOSAICC
Authors: Hideki Kanamaru and M. Evangelisti FAO
Abstract: Food and Agriculture Organization of the United Nations has been developing a server-based integrated system of tools and models called MOSAICC to evaluate climate change impacts on the agriculture sectors (crops, water resources, forests, and economy). MOSAICC is designed to respond to the needs of developing countries which may benefit from institutional and individual capacity development in producing relevant information for national climate change adaptation planning. This innovative system design facilitates a participatory environment where researchers with different expertise can work together efficiently. The utilization of the Web technologies avoids complex installations on the user’s computer and makes maintenance and update of the system on the centralized server simple. MOSAICC is based on open-source technology and models thus it is transferrable to countries free-of-charge. The easy-to-use interface and data sharing/exchange, through central database, support interdisciplinary cooperation among local experts as the chain of simulations and data are transparent to users. Visualization and communication of the results in graphs and maps on the Web are also integral parts of the system. The implementation strategy of MOSAICC in countries emphasizes stakeholder involvement in a technical working group throughout the study in order to produce information that is truly necessary in the country. National experts, who are members of the working group, design the study, perform simulations using MOSAICC, and publish the results to inform stakeholders. In the process, the experts learn the theory, methodology, and the use of models in a series of capacity development workshops.
Session 2.8: Oral Presentation
Title: The Akkerweb platform: models and data to support precision farming
Authors: Frits K. van Evert1, T. Been1, H. N. C. Berghuijs1, A. J. Haverkort1, C. Kempenaar1, G J. T. Kessel1, E. J. J. Meurs1, L. P.G. Molendijk1, A. A. Pronk1, D. A. van der Schans1, W. C. A. van Geel1, and J. A. Booij1 1 Wageningen University, Netherlands
Abstract: It is believed that precision farming (PF) will contribute to increased profitability of farming, to a reduction of the environmental impact of agriculture and, ultimately, to increased global food security. PF is based on the concept of observing and responding to spatial and temporal variability in crops and soils, both between and within fields. Thus, a first challenge in the implementation of PF is collecting and storing large amounts of geo-referenced data; a second challenge is to utilize this data to generate recommendations that can be used by farmers. Akkerweb (http://www.akkerweb.nl) is a web-based portal that allows for safe and easy storage of spatial and temporal soil, crop, climate and management data. Akkerweb provides a mechanism to deliver model-based recommendations, such as variable rate application (VRA) of seeds, fertilizers and crop protection agents. Akkerweb is an initiative of Wageningen UR and the Dutch farmers’ cooperative Agrifirm. We describe first the process of ingesting data from farms, soil analysis labs, unmanned aerial vehicles (UAVs), and satellites, as well as a prototype of
automatic data capture. We then detail how modelling is used to generate recommendations to farmers for most of the important decision points in growing potatoes, namely nematode control, application of sidedress nitrogen, late blight control, and potato haulm killing. These recommendations allow potato farmers to reduce the use of nitrogen fertilizer by up to 15% and of crop protection agents by 25% relative to current practice.
Session 2.8: Oral Presentation
Title: Enhancing Discoverability and Re-use of CGIAR’s Agricultural Data: Challenges and Progress
Authors: Medha Devare and the Open Access and Data Management Communities of Practice CGIAR
Abstract: CGIAR’s 15 Centers and other entities involved in agricultural research and development are charged with tackling complex challenges at a variety of scales, but research outputs are too often not easily discoverable or reusable. CGIAR is attempting to enhance discovery and reuse of its data for models and other tools through the Open Access and Open Data initiative and the development of a platform to harness the power of big data and ICTs. The OA/OD initiative focuses on creating a culture of data sharing, and provides support for harmonized data through the development of common metadata, ontologies, and strengthened collaboration and coordination around tools and approaches. This foundational work will render CGIAR outputs interoperable, ensuring they are discoverable via integrated and contextualized views across Centers and programs, type (e.g., publications, data, etc.), and discipline (e.g., genetic/genomic; agronomy; breeding; socioeconomic, and other sectors). The contents of most repositories at CGIAR Centers are not generally easily discoverable or inter-linked (e.g., agronomic trial data with socioeconomic or adoption data in the same geography). In the absence of such interoperability-mediated integration, “open” is of limited utility. The overall objective, then, is to make CGIAR’s trove of research data and associated information accessible for indexing and interlinking by a robust, demand-driven cyberinfrastructure for agriculture, ensuring that research outputs are Findable, Accessible, Interoperable and Re-usable (FAIR) for simulation, analytics, and visualization tools to enhance innovation and impact. This presentation will review the challenges to sharing and mining CGIAR data effectively, and the progress towards addressing these.
Session 2.8: Oral Presentation
Title: DataMill: a new application to interface researchers’ database with crop models
Authors: Myriam Adam1, S. Auzoux2, R. Loison2, and F. Affholder2 1 CIRAD, UMR AGAP, ICRISAT West Central Africa, Burkina Faso, 2 CIRAD, UR AIDA, France
Abstract: DataMill is an application built to improve access and re-use of agronomic data for crop modelling. DataMill extracts data from a standardized database and then translates these data to compatible model-ready formats for multiple crop models. DataMill is made of: (i) layouts of each input/output file from selected crop models, (ii) a Microsoft Access database (DataMill_DB) that structures all the variables of input/output files, (iii) and a Visual Basic executable file that converts the data from DataMill_DB into model input/output files in native format. The next step is to input researchers’ data into DataMill_DB through queries. This step will be crucial and depends on how each researcher is structuring and storing its data. DataMill facilitates the creation of model input files, one of the major bottlenecks in the use of crop models. Currently, DataMill is working for SARRA-H and DSSAT formats input files and is in development for APSIM and STICS. The concept of DataMill and an example of its use will be presented at the conference.
Session 2.8: Oral Presentation
Title: Interlinked data and models using a semantic approach: an example of the RECORD platform in the context of the ANAEE-France project
Authors: Hélène Raynal, A. Chanzy, C. Pichot , M. El hadramy , E. Casellas, F. Lafolie, and D. Maurice INRA
Abstract: RECORD is a modelling and software simulation platform dedicated to the study of agro-ecosystems. It is part of the AnaEE-France infrastructure which is a national research infrastructure for the study of continental ecosystems. This infrastructure brings together modelling platforms and databases for long term experiments, and aims at developing interoperability between them. For about two years now, RECORD and the other partners of AnaEE-France have worked together to develop this interoperability using the approach of linked data. In computing, linked data is a method of publishing structured data so that it can be interlinked and become more useful through semantic queries. A thesaurus and an ontology are being developed. The ontology is based on the generic conceptual model proposed by the OBOE ontology. Semantic tools are used. The presentation is intended to illustrate how this approach can be useful in the context of Open Data for Agricultural Modeling. The example of the specific work realized by the RECORD platform will be given. This work aims i) to facilitate the development of simulation applications combining models and experimental data and ii) to help modelers in finding models within the RECORD models library.
Session 2.8: Oral Presentation
Title: Application of AgMIP data interoperability standards to the US National Agricultural Research Data Network for Harmonized Data (NARDN-HD)
Authors: Cheryl H. Porter1 , J. W. Jones1 , G. Hoogenboom1 , C. Rosenzweig2 , C. Villalobos1 , and M. Zhang1 ,
1 University of Florida, USA, 2 NASA, GISS and Columbia University, USA
Abstract: Thereis a major gap between the potential value of agricultural research data and current outcomes that are mainly in the form of scientific papers. Vastly greater value could be obtained if datasets were combined across time, locations and management conditions. Most agricultural research data exist in small to medium files on personal computers and hand written notes. The data describe widely diverse systems, hindering development of common databases. Increasingly, funding agencies are mandating that research data be made available at the end of the project. However, the tools to support researchers in making their data uploadable, findable, accessible, interoperable and re-useable are not available in the agricultural sector. It has, therefore, become clear that a national effort is needed in the US to support researchers complying with federal open data mandates. The NARDN-HD project proposes to apply the data interoperability standards developed for AgMIP ensemble modeling activities to allow harmonization, archiving, retrieval and aggregation of data from experiments conducted at Agricultural Experiment Stations at US Land Grant Institutions and other USDA-funded projects. This is part of a broader initiative promoting open agricultural data by the USDA National Agricultural Library.
The NARDN-HD system, when implemented, will facilitate advancement of science using data intensive research methods for model development, statistical analyses, and meta-analyses; and will open up opportunities for new scientific discoveries via use of big data analytics spanning multiple sectors. These open data archives will also allow for
improved transparency and reproducibility of research findings to funders.
52. Poster Presentation: Session 2.8
Title: Challenges of integrating expert reasoning in DSS evaluation
Authors: Romain Bourget1 , A. Bsaibes1 , A. Caffarra1 , G. Garin1 , A. Guaus1 , P. Hublart1 , V. Houlès1 , and P. Stoop1
1 SAS iTK, France
Abstract: iTK develops software solutions to support the decisions of vine-growers and field technicians. These tools help users optimize their return on investment (RoI) and sustainability for vine irrigation (DSS Vintel®) and pest protection (DSS Bay+ Movida®). They use process-based models that describe the dynamics of soil water content and vine
water needs, or disease contaminations and severity. To evaluate these models, field constraints limit the amount of available data. Classically, these data refer to observations of soil water content (SWC) or disease severity at different dates. In a simplistic approach, direct statistical criteria (e.g. RMSE on SWC or severity) can assess the accuracy of the model on these variables. But the core question remains: “Are the decisions taken with the tool better than without it, in terms of RoI?” Indeed, multiple variables, other than those easily observed can impact decisions. For example, fungicide should be applied when contaminations occur, while only severity is easily observable. Moreover, criteria as RMSE highlight differences on variables at specific observation dates and do not consider dynamical aspects that are crucial in the decision-making process. Finally, user expertise is also ignored by such methods of evaluation. In this study, we will present examples of how to integrate expert reasoning in model evaluation (i.e. feedback from field technicians). However, because human experts cannot be included in large-scale, automatic routines of parameterization, we will discuss other methods that can be used to formalize and summarize such expertise by statistical criteria: expert-based data weighting, fuzzy logic, Pareto efficiency.
53. Poster Presentation: Session 2.8
Title: Calibration of CropSyst combining available climate information, remote sensing and data and minimum yield data
Authors: Francisco J. Meza and W. Davila
Centro de Cambio Global. Pontificia Universidad Catolica de Chile
Abstract: Crop simulation models (CSM) are a fundamental tool to understand the behavior of complex systems, particularly under uncertain and dynamic climate conditions. One of the most severe limitation for its wide use is the difficulty to achieve a proper calibration due to the extensive date requirements (both in terms of quantity and quality). The massification of automatic weather stations, that are run following reasonable protocols for quality control and the access of environmental data form satellites has opened new avenues for the development of methods to calibrate CSM and use them as tools to predicts impacts of climate variability and change as well as to carry on ex-ante assessments of adaptation strategies. Here we present an experiment run using CropSyst
that was calibrated with minimum yield data provided by ANASAC seed company and
the use of remote sensing data to calibrate phenological changes in Maize. Results are compared against independent data form the following two seasons showing a reasonably high coefficient of determination (0.85) in terms of biomass accumulation and water consumption.
54. Poster Presentation: Session 2.8
Title: How to use global sensitivity analysis to improve the user experience of a crop model commercialized in a Web Application? Feedbacks from the conception of CropWin®-Corn
Authors: Pierre Moreau, K. Bezzou, R. Bourget, A. Guaus, A. Pinet, N. Saint-Geours, and P. Stoop
iTK, France
Abstract: To support farmer in managing nutrient and water inputs for corn and soybean, iTK develops and markets Decision Support Systems (DSS) integrated in Web Applications (CropWin®-Corn, CropWin®-Soybean). These tools aim at helping farmer to improve their
profitability by pro-actively managing their cultural practices. The DSS are based on crop models that dynamically simulate, in soils and plants, the processes of carbon, nitrogen, potassium and water cycles on a daily basis to predict crop development and yield gap. The most limiting factors are daily displayed to the farmers. One key challenge of such application development is to limit the number of site-specific and hybrid-specific parameters that are asked to the end-users. This study shows how Global Sensitivity Analysis (GSA) can be an efficient tool to overcome this challenge and to improve user
experience. The aim of GSA is to determine how sensitive the outputs of a crop model are, with respect to the elements of the model which are subject to uncertainty or variability: input variables, equations, parameters. Morris and time-dependent Sobol’ methods were used on the corn model on 3 sites under 5 climates and 2 management practices to test the influence of site and hybrid parameters on predicted yields. Involving the end-users throughout the GSA process allowed defining the choice of parameters
to be tested and defining the experimental design. We will present in more details how we use GSA to identify the minimum set of parameters asked to the end-users in CropWin®-Corn, and to highlight hybrid-specific information.
55. Poster Presentation: Session 2.8
Title: How weather uncertainty impacts the predicted yield in CropWin-Soybean?
Authors: Amélie Pinet, R. Bourget, P. Moreau, N.
Saint-Geours, and P. Stoop
iTK, France
Abstract: iTK develops and markets software (CropWin®-Corn, CropWin®-Soybean) to support and improve farmer decision making. Our tools are designed so that the farmers can improve their profitability by pro-actively adapting their cultural practices: irrigation and
fertilization. Our software are based on crop models that dynamically simulate
processes of carbon, nitrogen, potassium and water balance on daily basis to predict crop growth and yield gap. One key challenge of developing software for farmers is the use of appropriate input data. Indeed, input data accuracy together with an adequate representation of plant physiology processes and choice of model parameters are the key factors for a reliable simulation. One of the most important inputs of our software is the weather: current weather data drive the simulation of the daily growth, while seasonal forecasts are used to predict the yield at the end of the growing season. Some farmers have their own weather stations but most of them do not gather all the required
weather data (temperature, rain, solar radiations, relative humidity and wind speed). Thus our models are run using gridded weather data. The uncertainty in weather data is due to both the spatial resolution of gridded data and the use of seasonal forecasts. This uncertainty has a crucial impact on the accuracy and the robustness of our software. In this study, we will present (i) the magnitude of differences between observed, gridded data and seasonal forecast and (ii) the impact of uncertainty in weather data on the yield predicted for three contrasted climates.
56. Poster Presentation: Session 2.8
Title: CASANDRA: Web based platform to assess
impacts and define adaption strategies to climate change.
Authors: Alfredo L. Rolla1,2 , E. R. Guevara3 , S.G. Meira3
1 CIMA (CONICET – UBA), 2 UMI IFAECI (CNRS, CONICET, UBA), 3 INTA
Abstract: Crop models are used to assess or estimate yield among other important variables such as phenology, dry matter, LAI, soil water content, soil and water management, as well as to evaluate different climate or management scenarios at a particular site. On the other hand, problems involving regional climate change include various spatial scales, and would benefit from simulations over large areas using high-resolution climate data spatially distributed (CMIP5, reanalysis, etc.). The geospatial web platform “CASANDRA” was developed to solve the spatial scale problem combining different site specific crop models (DSSAT, APSIM*, STICS*) with different climate scenarios and typical regional management. This way, it is possible to visualize impacts at both spatial and temporal level and also evaluate different regional adaptation strategies. The platform was design to be easy to use as well as to have the possibility to generate output information to be share with modern devices as smart-phones and tablets, using open source web mapping technologies (OpenLayers) to show results. CASANDRA was design by the concept of “platform independent”, “open source” and “cloud development” so it can be use on Windows, Linux, OSX, Android, etc. CASANDRA performs calculations incorporating the concept of homogeneous area, which is simply the intersection of different layers of information as climate, soil, management/cultivar with regions of interest. The platform uses open source technologies as (MySQL, R, php, JavaScript, Open Layers) and use algorithms of optimization and parallelization to speed up calculations. The platform was used for the “Third National Communication of Climate Change in Argentina for Impact and Adaption in Agriculture”.