This use cases addresses data challenges with respect to the calibration and application of crop models.

Background

Crop simulation models have become important tools in agricultural research and crop systems analysis. They play an increasing role in research on decision-making for automation in crop management from planting to harvesting. These crop models require data from diverse sources with different accessibility, size, aggregation, units, quality, temporal and spatial scales and formats to operate, including field records, soil surveys, weather stations, climate change scenarios, on-time field sensors, remote data from drones and satellites for data assimilation and seasonal and market weather forecasts. Some of the required data are generated based on qualitative information from different sources and converted into quantities, such as cultivar parameters. One important aspect hereby refers to the accessibility of data sources: some of these data are publicly available (e.g., weather data), while some are collected in ongoing research and are not published yet (e.g., some field sensor data), and others have restricted access rights depending on regulations. The integration of all model input data requires experts in a range of disciplines, such as agronomy, soil science, crop science, breeding, biogeochemical, hydrological, ecological, pathology, agricultural economy, meteorology, climate science and informatics for locating, accessing, and transferring these data, converting file formats, scales and units, quality checks and filling missing information. The crop models also generate large amounts of data that need to be quality checked, documented, made available to derive decisions for robots and drones conducting future field operations, prepared for long-term archiving, and made accessible to the research community in agriculture and other fields (e.g., earth systems-, climate impact science) for other studies.

Objectives

This use cases will identify data requirements and define a generic framework for a seamless integration of data with crop models in close collaboration with use cases 2 and 3. We will outline and develop a prototype for a seamless workflow to apply crop model inputs, crop model simulations and crop model outputs for parameterization of the DSSAT-Potato model as part of research for automation of a potato growth simulation from planting to harvest. The expected results will enable a comprehensive and continuous use of data sources (according to the respective data access rights; Measures 3.6 and 4.2) needed for operating the DSSAT-Potato model and crop models in general. The developed framework and prototype will guide the development of seamless infrastructures to integrate a range of data and simulation models in the agricultural research communities and hence in FAIRagro as a whole. This UC will integrate the FAIRagro data principles into a scientific workflow infrastructure and consider existing data infrastructure [e.g., SRADI (Smart Rural Area Data Infrastructure) at TUM] to enhance the quality of the existing data and enable interoperability with other data infrastructures.

Actions

1. Development of workflows for crop model applications
2. Automating data processing and storage for crop models
3. Automated plausibility checks for inputs and outputs of crop models

Progress & next steps

UC Update: Streamlining Data Integration for High-Throughput Crop Modelling

The UC team led by Leroy et al. has developed a workflow within the csmtools framework to streamline data integration for high-throughput crop modelling. The approach enables seamless linkage of experimental, environmental, and management data, facilitating automated and reproducible model simulations at scale.

Read the detailed Update and Progress Report by Benjamin Leroy and Anne Sennhenn from May 2024.