Tag Archives: AgMIP

spatial sampling

Whenever data is not available or simulating the entire region of interest is not feasible, sampling seems to be a good solution. How many sampling points are needed to accurately estimate the regional yield? What’s the error associated with that? What is the most suitable sampling strategy for the variable of interest, model, region or spatial variability? These questionas are adressed at different stages of AgMIP:

  • phase 1: estimates the performance of spatial sampling strategies accounting for climate variations
  • phase 2: estimates the performance of spatial sampling strategies accounting for soil and climate variations

The performance of sampling for instance climate or soil is estimated for a range of crops, crop models, production situations and variables (yield, biomass, LAI, soil moisture, evapotranspiration, phenology, NPP, NEE, soil organic carbon, N2O-emmissions, NO3-leaching, etc.).

Further reading

Spatial sampling of weather data for regional crop yield simulations

authors: Lenny van Bussel, Frank Ewert, Gang Zhao, Holger Hoffmann, Andreas Enders, Daniel Wallach, Julie Constantin, Hélène Raynal, Christian Klein, Christian Biernath, Florian Heinlein, Eckart Priesack, Fulu Tao, Reimund Rötter; Davide Cammarano, Senthold Asseng, Joshua Elliott, Michael Glotter, Claas Nendel, Kurt-Christian Kersebaum, Xenia Specka, Bruno Basso, Guillermo Baigorria, Consuelo Romero, James Chryssanthacopoulos.

abstract: Field-scale crop models are increasingly applied at spatio-temporal scales that range from regions to the globe and from decades up to 100 years. Sufficiently detailed data to capture the prevailing spatio-temporal heterogeneity in weather, soil, and management conditions as needed by crop models are rarely available. Effective sampling may overcome the problem of missing data but has rarely been investigated. In this study the effect of sampling weather data has been evaluated for simulating yields of winter wheat in a region in Germany over a 30-year period (1982-2011) using 12 process-based crop models. A stratified sampling was applied to compare the effect of different sizes of
spatially sampled weather data (10, 30, 50, 100, 500, 1000 and full coverage of 34,078 sampling points) on simulated wheat yields. Stratified sampling was further compared with random sampling.  Possible interactions between sample size and crop model were evaluated.

journal:  Agr For Met (under review).


Model and data scales are of concern to nearly all simulations of the environment, including cropping systems. Temporal and spatial resolutions of data and model or between linked models should match. However, data is often scarce and simulations are often conducted for estimating parameters or assessing impacts were data is not available. With an increasing number of simulation studies ranging from field to global scale, scaling models has consequently become top science pick. For an overview on scaling methods with regard to models, please see Ewert et al., 2011. Two main aspects of spatial scales are illustrated in the figure below: spatial data aggregation / resolution and the representation of spatial mean and variability, e.g. via sampling.


Set up of simulation experiments for evaluating deviations from a) high resolution simulations of b) different grid cell sizes and c) reduced numbers of sampling points. Details are described in van Bussel et al. (2015) and Hoffmann et al. (2015). With friendly permission of Ewert et al. (2015).

The scaling research presented on this site has been mainly driven in the past by the AgMIP and MACSUR scaling activities. These activities mainly conduct crop model ensemble studies in order to

  • adress the error which is made by applying field-scale crop models at larger scales
  • identify crop model specific model-data-interactions
  • identify error sources and ways of generalizing these errors
  • estimate the error which is made by choosing a specific spatial sampling strategy
  • develop spatial sampling stragies
  • identify spatial patternsa and corresponding drivers

The following wikis on input data aggregation and spatial sampling  give an overview on the current scaling activities.

If you are a contributing scientist, please register or login to get more details, read or write about the on-going work.

Frank Ewert, Martin K. van Ittersum, Thomas Heckelei, Olivier Therond, Irina Bezlepkina, Erling Andersen, 2011. Scale changes and model linking methods for integrated assessment of agri-environmental systems. Agriculture, Ecosystems & Environment 142, 6 – 17.

Frank Ewert, Lenny G.J. van Bussel, Gang Zhao, Holger Hoffmann, Thomas Gaiser, Xenia Specka, Claas Nendel, Kurt-Christian Kersebaum, Carmen Sosa, Elisabet Lewan,  Jagadeesh Yeluripati,  Matthias Kuhnert, Fulu Tao, Reimund P. Rötter, Julie Constantin, Helene Raynal, Daniel Wallach, Edmar Teixeira, Balasz Grosz, Michaela Bach, Luca Doro,  Pier P. Roggero, Zhigan Zhao, Enli Wang, Ralf Kiese, Edwin Haas, Henrik Eckersten,  Giacomo Trombi, Marco Bindi, Christian Klein, Christian Biernath, Florian Heinlein, Eckart Priesack, Davide Cammarano, Senthold Asseng, Joshua Elliott, Michael Glotter, Bruno Basso, Guillermo A. Baigorria, Consuelo C. Romero, Marco Moriondo, 2015. Uncertainties in Scaling up Crop Models for Large Area Climate-change Impact Assessments. In C. Rosenzweig and D. Hillel, editors. Handbook of Climate Change and Agroecosystems: The Agricultural Model Intercomparison and Improvement Project (AgMIP).


This site is about crop modelling  across scales.

The aims are:

  • to give scientists the chance to communicate, contribute and inform
  • to inform about on-going activities in the varying projects
  • to document simulations and data

This site is an initiative of the AgMIP and MACSUR scaling activities.