Elsevier

Agricultural and Forest Meteorology

Volume 200, 15 January 2015, Pages 156-171
Agricultural and Forest Meteorology

Demand for multi-scale weather data for regional crop modeling

https://doi.org/10.1016/j.agrformet.2014.09.026Get rights and content

Highlights

  • Spatial resolution of weather data is a key issue for regional crop modeling.

  • We propose a method to choose spatial resolution of weather data for crop modeling.

  • An example at national scale of Germany was used to demonstrate the proposed method.

  • The bias caused by data aggregation was restricted to a specified threshold.

  • The multi-resolution simulation captured more spatial details with fewer simulations.

Abstract

A spatial resolution needs to be determined prior to using models to simulate crop yields at a regional scale, but a dilemma exists in compromising between different demands. A fine spatial resolution demands extensive computation load for input data assembly, model runs, and output analysis. A coarse spatial resolution could result in loss of spatial detail in variability. This paper studied the impact of spatial resolution, data aggregation and spatial heterogeneity of weather data on simulations of crop yields, thus providing guidelines for choosing a proper spatial resolution for simulations of crop yields at regional scale. Using a process-based crop model SIMPLACE 〈LINTUL2〉 and daily weather data at 1 km resolution we simulated a continuous rainfed winter wheat cropping system at the national scale of Germany. Then we aggregated the weather data to four resolutions from 10 to 100 km, repeated the simulation, compared them with the 1 km results, and correlated the difference with the intra-pixel heterogeneity quantified by an ensemble of four semivariogram models. Aggregation of weather data had small effects over regions with a flat terrain located in northern Germany, but large effects over southern regions with a complex topography. The spatial distribution of yield bias at different spatial resolutions was consistent with the intra-pixel spatial heterogeneity of the terrain and a log–log linear relationship between them was established. By using this relationship we demonstrated the way to optimize the model resolution to minimize both the number of simulation runs and the expected loss of spatial detail in variability due to aggregation effects. We concluded that a high spatial resolution is desired for regions with high spatial environmental heterogeneity, and vice versa. This calls for the development of multi-scale approaches in regional and global crop modeling. The obtained results require substantiation for other production situations, crops, output variables and for different crop models.

Introduction

Crop models are becoming essential tools to study the interactions between plant, climate, soil and management practices in agricultural systems at different scales (Fischer et al., 2005, van Ittersum et al., 2008) to support policy and decision making for food security and land-use related management (De Vries et al., 1997, Rabbinge and Van Diepen, 2000). To apply crop models at regional scale, a level of homogeneity of spatial units or resolution needs to be assumed (Harrison et al., 2000), since the spatial heterogeneity of environmental conditions is ubiquitous (Wu, 2004). Once the resolution is determined, the input data such as climate and soil conditions, and management practices need to be aggregated or disaggregated to the same resolution to drive the crop models (Priya and Shibasaki, 2001). At a certain spatial extent, the resolution determines the detail in spatial variability that can be captured as well as the computational load. A fine resolution could capture more spatial detail, but it raises the demand for data storage, computational costs and efforts for results analysis (Zhao et al., 2013a). A coarse resolution requires less computing resources, but loses information in spatial variability.

The layers of coarse resolution are inherent or can be produced from layers of fine resolution by spatial aggregation (van Bussel et al., 2011). Spatial aggregation could be conducted by different methods such as spatial mean, median, and majority (Bian and Butler, 1999). No matter which method is adopted, loss of the spatial variability is inevitable (Li and Reynolds, 1995), because the resultant data uses one value to represent many others (Bian and Butler, 1999, Orcutt et al., 1968). The magnitude of the spatial variability loss varies with the features of the original data and the resolution differences. If the resultant area (also referred to as pixel) is homogeneous or covers a range of smaller pixels with similar values, the information loss would be minor. In contrast, the loss would be enormous if the covered smaller pixels are extremely heterogeneous. Here we term the variability and structure of the smaller pixels covered by the coarser resolution pixel as intra-pixel spatial heterogeneity. The loss of the intra-pixel spatial heterogeneity could be propagated by the crop model, thus affecting the simulated results such as crop yields (Guisan et al., 2007).

The effects of spatial resolution of weather data on simulated crop yields or crop phenology have been extensively studied (Angulo et al., 2013b, Folberth et al., 2012, Mearns et al., 1999, Mummery and Battaglia, 2002, van Bussel et al., 2011). These studies evaluated the effects by comparing the results simulated by input data at different spatial resolutions and found relatively small effects of weather data aggregation (Hansen and Jones, 2000, Nendel et al., 2013). Aggregation effects on mean values or medians are diminished when the response of the crop model to the change of weather data is close to linear or the non-linear response is not significant. This could result in a conclusion that minor aggregation effects are caused by input data aggregation.

The aim of spatially explicit modeling is capturing the spatial heterogeneity, which should be a priority in determining the spatial resolution (Di Vittorio and Miller, 2014). However, to our best knowledge no method has been developed to guide the choice of spatial resolution for regional scale simulation, because quantification of spatial heterogeneity of long-term time series of daily weather data is difficult. Weather data consisting of a set of variables, e.g. temperature, radiation, precipitation, and a high temporal frequency, often daily, is required for process-based crop models (Asseng et al., 2013, Balkovič et al., 2013). Within the spatial extent of a pixel (e.g. 10 km), the long-term weather condition is greatly influenced by the local topographic features (Ashcroft and Gollan, 2012). Topographic conditions are frequently used to correct the layers of weather data (Hancock and Hutchinson, 2006), thus long-term weather conditions are highly correlated with terrain features. Using the spatial heterogeneity of the topography to surrogate the weather conditions has the potential to bypass the difficulties in quantifying the spatial heterogeneity of different layers of weather variables (Kerr, 1993, Seoane et al., 2004). However, this has not been tested yet to inform about the required spatial resolution for simulating crop yield at regional scale using crop models. This paper aims to improve the understanding of the required spatial resolution of weather data for the simulation of regional crop yields and how the proper resolution can be determined.

Section snippets

General workflow of analysis

The general workflow of the simulation analysis is presented in Fig. 1. We produced 31-year weather data at 1 km resolution for the whole Germany and aggregated weather data into resolutions of 10, 25, 50, and 100 km. Using these data, we simulated a 30-year continuous cropping system of winter wheat, with a process-based crop model LINTUL2 (Light INTerception and UtiLization) (van Ittersum et al., 2003, van Oijen and Lefelaar, 2008) under rainfed condition. Three regions, respectively,

Model evaluation

The model overestimated the yields for the majority of the study area, especially in southern regions (Fig. 4b and c). The spatial patterns between observed and simulated yield were similar. Both the linear regression (r2) and correlation test (rho) showed that the model reproduced the temporal evolution quite well in the north-eastern regions, but relatively poor in the northern and south-western regions (Fig. 4d–g). The temporal variation of simulated yield was consistent with the historical

Model performance

The aim of this study was to evaluate the effect of weather data aggregation on crop yields simulated by a dynamic crop growth model. Specific emphasis was on the effects of aggregation on the spatial variability of simulated yields. Although we did not aim for a model evaluation of regional yield simulations, some demonstration of model performance for regional application was required to justify its use in the present study. The considered model LINTUL2 has been successfully applied in many

Conclusion

Regional scale application of crop models developed at field scale is becoming a common practice in impact assessment studies of climate variability and change. This paper evaluated the impact of weather data aggregation on simulation of rainfed crop yields at the national scale of Germany. It advances from early studies on aggregation by specifically addressing the effect of aggregation on spatial yield variability. Different resolutions of weather data simulated similar spatial patterns of

Acknowledgements

This study was jointly supported by the BMBF/BMELV project on “Modeling European Agriculture with Climate Change for Food Security (MACSUR)” (grant no. 2812ERA115) and the DFG projects on “Multi-scale modeling of the impacts of climate change and climate variability on winter wheat in Germany” (grant no. EW 119/5-1) and the Transregio (TRR32/2) on “Patterns in soil-vegetation-atmosphere systems: monitoring, modeling and data assimilation”. The comments from two anonymous reviewers greatly

References (65)

  • I. Farré et al.

    Analysis of maize growth for different irrigation strategies in northeastern Spain

    Eur. J. Agron.

    (2000)
  • G. Fiorese et al.

    A GIS-based approach to evaluate biomass potential from energy crops at regional scale

    Environ. Modell. Softw.

    (2010)
  • C. Folberth et al.

    Impact of input data resolution and extent of harvested areas on crop yield estimates in large-scale agricultural modeling for maize in the USA

    Ecol. Modell.

    (2012)
  • T. Gaiser

    Modeling biopore effects on root growth and biomass production on soils with pronounced sub-soil clay accumulation

    Ecol. Modell.

    (2013)
  • P. Hancock et al.

    Spatial interpolation of large climate data sets using bivariate thin plate smoothing splines

    Environ. Modell. Softw.

    (2006)
  • J.W. Hansen et al.

    Scaling-up crop models for climate variability applications

    Agric. Syst.

    (2000)
  • P.A. Harrison et al.

    Scaling-up the AFRCWHEAT2 model to assess phenological development for wheat in Europe

    Agric. For. Meteorol.

    (2000)
  • R.J. Hijmans et al.

    A quantitative and constraint-specific method to assess the potential impact of new agricultural technology: the case of frost resistant potato for the Altiplano (Peru and Bolivia)

    Agric. Syst.

    (2003)
  • J. Liu

    A GIS-based tool for modelling large-scale crop-water relations

    Environ. Modell. Softw.

    (2009)
  • D. Mummery et al.

    Data input quality and resolution effects on regional and local scale Eucalyptus globulus productivity predictions in north-east Tasmania

    Ecol. Modell.

    (2002)
  • C. Nendel

    Simulating regional winter wheat yields using input data of different spatial resolution

    Field Crops Res.

    (2013)
  • J.E. Olesen et al.

    Comparison of scales of climate and soil data for aggregating simulated yields of winter wheat in Denmark

    Agricult. Ecosys. Environ

    (2000)
  • T. Palosuo

    Simulation of winter wheat yield and its variability in different climates of Europe: a comparison of eight crop growth models

    Eur. J. Agron.

    (2011)
  • E.J. Pebesma

    Multivariable geostatistics in S: the gstat package

    Comput. Geosci.

    (2004)
  • S. Priya et al.

    National spatial crop yield simulation using GIS-based crop production model

    Ecol. Modell.

    (2001)
  • R. Rabbinge et al.

    Changes in agriculture and land use in Europe

    Eur. J. Agron.

    (2000)
  • R.P. Rötter

    Simulation of spring barley yield in different climatic zones of Northern and Central Europe: a comparison of nine crop models

    Field Crops Res.

    (2012)
  • J. Seoane et al.

    Competing roles for landscape, vegetation, topography and climate in predictive models of bird distribution

    Ecol. Modell.

    (2004)
  • S. Siebert et al.

    Spatio-temporal patterns of phenological development in Germany in relation to temperature and day length

    Agric. For. Meteorol.

    (2012)
  • L.G.J. van Bussel et al.

    Effects of data aggregation on simulations of crop phenology

    Agricult. Ecosys. Environ

    (2011)
  • M.K. van Ittersum

    Integrated assessment of agricultural systems—a component-based framework for the European Union (SEAMLESS)

    Agric. Syst.

    (2008)
  • M.K. van Ittersum

    On approaches and applications of the Wageningen crop models

    Eur. J. Agron.

    (2003)
  • Cited by (75)

    • Framework to guide modeling single and multiple abiotic stresses in arable crops

      2022, Agriculture, Ecosystems and Environment
      Citation Excerpt :

      At the level of a landscape or region this will increase, due to soil heterogeneity, as well as differences in topography, and crop management in terms of sowing date, fertilization, crop and variety selection (Eyshi Rezaei et al., 2017). Errors introduced in simulation with coarser resolution input data for landscape scale analysis is expected to be relatively greater when impacts of extreme events are considered due to their non-linear nature (Zhao et al., 2015). Further, under actual production conditions, combinations of biotic and abiotic stressor are also expected to intensify with climate change and warrant more investigation (Savary et al., 2018; Suzuki et al., 2014).

    View all citing articles on Scopus
    View full text