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

Abstract

In this study, the performance of nine widely used and accessible crop growth simulation models (APES-ACE, CROPSYST, DAISY, DSSAT-CERES, FASSET, HERMES, MONICA, STICS and WOFOST) was compared during 44 growing seasons of spring barley (Hordeum vulgare L.) at seven sites in Northern and Central Europe. The aims of this model comparison were to examine how different process-based crop models perform at multiple sites across Europe when applied with minimal information for model calibration of spring barley at field scale, whether individual models perform better than the multi-model mean, and what the uncertainty ranges are in simulated grain yields. The reasons for differences among the models and how results for barley compare to winter wheat are discussed.

Regarding yield estimation, best performing based on the root mean square error (RMSE) were models HERMES, MONICA and WOFOST with lowest values of 1124, 1282 and 1325 (kg ha⁻¹), respectively. Applying the index of agreement (IA), models WOFOST, DAISY and HERMES scored best having highest values (0.632, 0.631 and 0.585, respectively). Most models systematically underestimated yields, whereby CROPSYST showed the highest deviation as indicated by the mean bias error (MBE) (−1159 kg ha⁻¹). While the wide range of simulated yields across all sites and years shows the high uncertainties in model estimates with only restricted calibration, mean predictions from the nine models agreed well with observations. Results of this paper also show that models that were more accurate in predicting phenology were not necessarily the ones better estimating grain yields. Total above-ground biomass estimates often did not follow the patterns of grain yield estimates and, thus, harvest indices were also different. Estimates of soil moisture dynamics varied greatly.

In comparison, even though the growing cycle for winter wheat is several months longer than for spring barley, using RMSE and IA as indicators, models performed slightly, but not significantly, better in predicting wheat yields. Errors in reproducing crop phenology were similar, which in conjunction with the shorter growth cycle of barley has higher effects on accuracy in yield prediction.

Introduction

Various model-based tools are used to support the decision making and planning in agriculture (Brouwer and van Ittersum, 2010, Ewert et al., 2011). Crop growth simulation models (hereafter referred to as crop models) are increasingly being applied, particularly in climate change-related agricultural impact assessments (Rosenzweig and Wilbanks, 2010, White et al., 2011).

Recently, there has been renewed interest and discussion about the need for improved understanding and reporting of the uncertainties related to crop growth and yield predictions (Rötter et al., 2011a, Ferrise et al., 2011, Børgesen and Olesen, 2011). Comparison of different modelling approaches and models can reveal the uncertainties involved. Variation of model results in model comparisons involves also the uncertainty related to model structure, which is probably the source of uncertainty most difficult to quantify. Model comparisons, when combined with experimental data of the compared variables, may also be used to test the performance of different models. However, comprehensive data sets that would allow such thorough comparisons (see, e.g. Groot and Verberne, 1991 or Kleemola et al., 1995), are scarce and in most cases have already been utilized or published for model calibration or validation. This situation calls for a concerted effort to exploit existing (unused) and develop new high quality data sets for different locations (agro-climatic conditions) and crops (Rötter et al., 2011a). Since the 1980s, there have been many studies on comparing different process-based crop models on their performance in predicting yield variability in response to climate and other factors (see, e.g. Kersebaum et al., 2007, Palosuo et al., 2011), including a very active period during the 1990s (Porter et al., 1993, Diekkrüger et al., 1995, Ewert et al., 2002, Goudriaan et al., 1994, Jamieson et al., 1998, Kabat et al., 1995, Wolf et al., 1996). Most of these comparisons have been made for wheat while other crops such as barley, received much less attention (Tubiello and Ewert, 2002; see, e.g. Eitzinger et al., 2004 for an exception).

Since proper understanding and modelling of crop responses to heat and drought stress becomes increasingly important in climate impact assessments (Semenov and Shewry, 2011, Lobell et al., 2012), we also looked into this issue. In a couple of studies in different parts of the world specific responses of barley to heat and drought stress have been investigated (e.g. Jamieson et al., 1995, Passarella et al., 2005). For the critical growth stages during and immediately after flowering (Savin and Nicolas, 1999), it has been found that significant yield reduction is experienced if threshold temperatures of 28–30 °C are exceeded. Yield-reducing effects depend, however, on the timing and intensity of events (Passarella et al., 2005). Moreover, there is considerable response diversity among barley cultivars (see, e.g. Hakala et al., 2012). For drought stress, Jamieson et al. (1995) found no clear thresholds, but rather the importance of timing of drought for reduction in final biomass of barley, whereby final biomass was especially sensitive to soil moisture deficit for the early drought treatments.

To analyze sources of crop model uncertainties in climate impact assessments for Europe, four crop model intercomparisons were set-up during 2009–2010 in the framework of COST action 734, seeking coverage of the most widely used and accessible crop simulation models: one comparison for winter wheat (Palosuo et al., 2011) and another one for spring barley (this study) across multiple sites in Europe with restricted calibration, one on the sensitivity of crop models to extreme weather conditions for maize and winter wheat (Eitzinger et al., in press), and one with a detailed calibration using comprehensive barley datasets from one Finnish location (Salo et al., companion paper, in preparation).

This paper presents the results of the spring barley (Hordeum vulgare L.) comparison across multiple sites in Europe. Barley is currently the third most important cereal in Northern and Central Europe after wheat and grain maize (EUROSTAT, 2011). Since spring barley has been much less considered in crop modelling than winter wheat, and assuming that accordingly wheat models were developed with more experimental data than those for barley, we hypothesized that the uncertainties in simulation results for barley are higher.

The specific objectives of this model intercomparison study were to examine (1) how different process-based crop models perform at multiple sites across Europe in estimating grain yield when applied with minimal information for model calibration of spring barley at field scale, (2) whether individual models perform better than the multi-model mean, and (3) what the uncertainty ranges are in simulated grain yields. Furthermore, an initial effort is made to discuss the reasons for differences among the models and investigate how results for barley compare to winter wheat (Palosuo et al., 2011).

We applied nine crop models altogether for 44 growing seasons of spring barley at seven different study sites in Europe: in the Czech Republic, Denmark, Finland and Slovakia.