Why do crop models diverge substantially in climate impact projections? A comprehensive analysis based on eight barley crop models

https://doi.org/10.1016/j.agrformet.2019.107851Get rights and content

Highlights

  • Eight barley models and eight climate projections for the 2050s were applied at two sites.

  • Sensitivity analyses were conducted on the responses of major crop processes to major climatic variables.

  • A new comprehensive analysis was conducted to look at the reasons why crop models diverge substantially.

  • Impacts of increases in temperature and CO2 on leaf area development were the major causes.

  • The study has important implications for models improvement and experimental design.

Abstract

Robust projections of climate impact on crop growth and productivity by crop models are key to designing effective adaptations to cope with future climate risk. However, current crop models diverge strongly in their climate impact projections. Previous studies tried to compare or improve crop models regarding the impact of one single climate variable. However, this approach is insufficient, considering that crop growth and yield are affected by the interactive impacts of multiple climate change factors and multiple interrelated biophysical processes. Here, a new comprehensive analysis was conducted to look holistically at the reasons why crop models diverge substantially in climate impact projections and to investigate which biophysical processes and knowledge gaps are key factors affecting this uncertainty and should be given the highest priorities for improvement. First, eight barley models and eight climate projections for the 2050s were applied to investigate the uncertainty from crop model structure in climate impact projections for barley growth and yield at two sites: Jokioinen, Finland (Boreal) and Lleida, Spain (Mediterranean). Sensitivity analyses were then conducted on the responses of major crop processes to major climatic variables including temperature, precipitation, irradiation, and CO2, as well as their interactions, for each of the eight crop models. The results showed that the temperature and CO2 relationships in the models were the major sources of the large discrepancies among the models in climate impact projections. In particular, the impacts of increases in temperature and CO2 on leaf area development were identified as the major causes for the large uncertainty in simulating changes in evapotranspiration, above-ground biomass, and grain yield. Our findings highlight that advancements in understanding the basic processes and thresholds by which climate warming and CO2 increases will affect leaf area development, crop evapotranspiration, photosynthesis, and grain formation in contrasting environments are needed for modeling their impacts.

Introduction

Impacts of climate change on future crop productivity and food security have been of key concern because agricultural production risks may increase, and food security be threatened, by climate change and increasing extreme climate events (Lobell et al., 2011; Olesen et al., 2011; Porter et al., 2014;Rötter et al., 2018 ). Crop models are popular tools to project climate change impacts on future agricultural production, driven by climate projections from global climate models (GCMs) (White et al., 2011; Porter et al., 2014). However, the projections of climate change impacts are plagued with uncertainties from many sources, such as climate projections, crop model parameters, and crop model structure (Tao et al., 2009, 2018; Rötter et al., 2011;Rötter, 2014; Asseng et al., 2013, 2015; Wallach et al., 2017; Wallach and Thorburn, 2017). These uncertainties have to be quantified and reduced as much as possible in order to better assess climate risk and inform effective adaptation (Tao et al., 2018). Recently, some studies have consistently indicated that the uncertainty from crop model structure is larger than those from climate projections and crop model parameters (Asseng et al., 2013, 2015; Bassu et al., 2014; Li et al., 2015; Zhang et al., 2017; Tao et al., 2018). Therefore, the most effective way to improve climate change impact projections is to reduce the uncertainty from crop model structure through model comparison and improvement, combined with considerations of other sources of uncertainty such as model parameters (Wang et al., 2017; Challinor et al., 2018; Tao et al., 2018; Rötter et al., 2018). Major international efforts, such as the Modelling European Agriculture with Climate Change for Food Security (MACSUR) project (Ewert et al., 2015) and the Agricultural Model Inter-comparison and Improvement Project (AgMIP) (Rosenzweig et al., 2013; Ruane et al., 2017), have tried to compare and improve crop models, and quantify, manage or reduce uncertainty from model structure in projecting climate impact on crop productivity (e.g. Palosuo et al., 2011; Rötter et al., 2012; Asseng et al., 2013, 2015; Bassu et al., 2014; Li et al., 2015; Martre et al., 2015; Pirttioja et al., 2015; Durand et al., 2018; Maiorano et al., 2017; Hasegawa et al., 2017; Müller et al., 2017; Wang et al., 2017; Tao et al., 2018).

Previous studies tried to compare or improve crop models mainly by simulating the impact of one single climate variable, such as temperature (Zhang and Tao, 2013; Asseng et al., 2013, 2015; Maiorano et al., 2017; Wang et al., 2017) or rising atmospheric CO2 concentration (Durand et al., 2018; Hasegawa et al., 2017) on crop development, growth, water use, and grain formation. These studies documented that although individual crop models were able to simulate observed grain yields fairly well, yield projection in response to climate warming or elevated CO2 varied significantly among the crop models because of quite different temperature or CO2 relationships applied in the models (Asseng et al., 2013, 2015; Zhang and Tao, 2013; Durand et al., 2018; Wang et al., 2017; Hasegawa et al., 2017). These studies are important in understanding the uncertainty and for improving models regarding the impact of a specific climate variable. However, final grain yields are subject to the interactive impacts of multiple climate change factors and many interrelated biophysical processes, such as crop phenological development, leaf area development, evapotranspiration (ET), photosynthesis, and grain formation (Porter and Gawith, 1999). The impacts of different climate variables on multiple biophysical processes and consequently on final grain yields can be offset or additive (Swann et al., 2016; Tao et al., 2017). A comprehensive analysis that accounts for the interactive impacts of multiple climate variables and their interactions, as well as for the multiple biophysical processes and their interactions, is necessary in order to understand, in a holistic way, why climate impact projections by different crop models are so different. Furthermore, since most of the biophysical processes in crop development, growth, water use, and grain formation are directly or indirectly affected by temperature and CO2 relationships in crop models, it is necessary to identify, in more detail, which biophysical processes and knowledge gaps are the key sources or bottlenecks for the uncertainty in climate impact projections and therefore should be given the highest priorities for improvement.

As an outgrowth of our previous study (Tao et al., 2018), here we first applied eight barley models and eight climate projections for the 2050s to quantify the uncertainty from crop model structure in climate impact projections for barley growth and yield at two sites with contrasting climate: Jokioinen, Finland in the Boreal climatic zone and Lleida, Spain in the Mediterranean climatic zone. The climate projections, the impact projections from each single model, as well as the uncertainties in impact projections, provide the rationale, target, and basis for the following analyses. We then conducted sensitivity analyses for each of the eight crop models based on their responses to major climatic variables including temperature, precipitation, irradiation, and CO2, as well as their interactions. Finally, we conducted a comprehensive analysis to investigate why crop models diverge substantially in climate impact projections. Differently from previous model inter-comparison studies (Asseng et al., 2013, 2015; Bassu et al., 2014; Li et al., 2015; Castañeda et al., 2015; Hasegawa et al., 2017) and impact response surface studies (Pirttioja et al., 2015), we aimed here, for the first time, to gain insights into the reasons underlying the divergence in climate impact projections in a holistic way.

Section snippets

Study sites

Two study sites with contrasting climates were selected for this study. They represent the North and South of current agro-climatic conditions for barley cultivation areas in Europe. The general information on the geographical location, climate, and barley cultivation is presented in Table 1.

Crop models and data

Eight different barley models with varying complexity were applied: APSIM 7.7 (AP, Holzworth et al., 2014), CropSyst 4.15.04 (CS, Stöckle et al., 2014), HERMES 4.26 (HE, Kersebaum, 2007), MCWLA 2.0 (MC,

Projected climate change impacts on barley development, growth, and yield in the 2050s

In the 2050s, under the eight climate projections, the barley maturity date was projected to advance at both Jokioinen and Lleida, however, by different magnitudes among the eight models and climate projections (Fig. 1a, b). In particular, MO projected the largest advancement in maturity date at Jokioinen, and AP projected the least advancement at Lleida, among the eight crop models. The projected changes in LAImax diverged substantially among the eight crop models and climate projections at

Why do crop models diverge substantially in climate impact projections?

The present study provides insights into how crop models diverge in simulating the major processes of crop development, growth, water use, and yield formation in response to changes in temperature, precipitation, solar radiation, CO2, and their interactions. The divergences can be ascribed to the large differences in model structures among the eight examined crop models (Tables S2-S9). For example, although most of the models accounted for the impacts of temperature, photoperiod, and

Conclusions

In this study, a new comprehensive analysis was conducted to look holistically at the reasons why crop models diverge substantially in climate impact projections and to investigate which biophysical processes and knowledge gaps are key factors affecting this uncertainty and should be given the highest priorities for improvement. Eight barley models and eight climate projections for the 2050s were applied to investigate the uncertainty from crop model structure in climate impact projections for

Declaration of Competing Interest

None.

Acknowledgements

This study was carried out in the context of CropM within the FACCE-MACSUR knowledge hub (031A103B). FT and TP were supported by the Academy of Finland through projects AI-CropPro (decision no. 316172) and DivCSA (decision no. 316215) and Natural Resources Institute Finland through strategic projects ClimSmartAgri and Boost-IA. RPR was supported by the German Federal Ministry of Education and Research via the ‘Limpopo Living Landscapes’ project within the SPACES program (grant number 01LL1304A)

References (77)

  • D.P. Holzworth et al.

    . APSIM - evolution towards a new generation of agricultural systems simulation

    Environ. Modell. Softw.

    (2014)
  • P.D. Jamieson et al.

    Modelling nitrogen uptake and redistribution in wheat

    Field Crops Res.

    (2000)
  • H.Y. Kim et al.

    Effects of free air CO2 enrichment and nitrogen supply on the yield of temperate paddy rice crops

    Field Crops Res.

    (2003)
  • B.A. Kimball

    Crop responses to elevated CO2 and interactions with H2O, N, and temperature

    Curr. Opin. Plant Biol.

    (2016)
  • A. Maiorano et al.

    Crop model improvement reduces the uncertainty of the response to temperature of multi-model ensembles

    Field Crops Res.

    (2017)
  • P. Martre et al.

    Modelling protein content and composition in relation to crop nitrogen dynamics for wheat

    Eur. J. Agron.

    (2006)
  • C. Nendel et al.

    The MONICA model: testing predictability for crop growth, soil moisture and nitrogen dynamics

    Ecol. Modell.

    (2011)
  • J.E. Olesen et al.

    Impacts and adaptation of European crop production systems to climate change

    Eur. J. Agron.

    (2011)
  • T. Palosuo et al.

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

    Eur. J. Agron.

    (2011)
  • J.R. Porter et al.

    Temperatures and the growth and development of wheat: a review

    Eur. J. Agron.

    (1999)
  • A.M. Ratjen et al.

    Key variables for simulating leaf area and N status: biomass based relations versus phenology driven approaches

    Eur. J. Agron.

    (2018)
  • C. Rosenzweig et al.

    The agricultural model intercomparison and improvement project (AgMIP)

    Agric. For. Meteorol.

    (2013)
  • R.P. Rötter et al.

    Linking modelling and experimentation to better capture crop impacts of agroclimatic extremes - a review

    Field Crops Res.

    (2018)
  • R.P. Rötter et al.

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

    Field Crops Res.

    (2012)
  • T.R. Sinclair et al.

    Modeling nitrogen accumulation and use by soybean

    Field Crops Res.

    (2003)
  • C.O. Stöckle et al.

    Cropsyst model evolution: from field to regional to global scales and from research to decision support systems

    Environ. Modell. Softw.

    (2014)
  • F. Tao et al.

    Modeling the impacts of weather and climate variability on crop productivity over a large area: a new process-based model development, optimization, and uncertainties analysis

    Agric. For. Meteorol.

    (2009)
  • F. Tao et al.

    Effects of climate change, CO2 and O3 on wheat productivity in Eastern China, singly and in combination

    Atmos. Environ.

    (2017)
  • D. Wallach et al.

    Accounting for both parameter and model structure uncertainty in crop model predictions of phenology: a case study on rice

    Eur. J. Agron.

    (2017)
  • D. Wallach et al.

    Estimating uncertainty in crop model predictions: current situation and future prospects

    Eur. J. Agron.

    (2017)
  • J. Wang et al.

    Size and variability of crop productivity both impacted by CO2 enrichment and warming—a case study of 4 year field experiment in a Chinese paddy

    Agric. Ecosyst. Environ.

    (2016)
  • J.W. White et al.

    Methodologies for simulating impacts of climate change on crop production

    Field Crops Res.

    (2011)
  • S. Zhang et al.

    Modeling the response of rice phenology to climate change and variability in different climatic zones: comparisons of five models

    Eur. J. Agron.

    (2013)
  • S. Zhang et al.

    Uncertainty from model structure is larger than that from model parameters in simulating rice phenology in China

    Eur. J. Agron.

    (2017)
  • E.A. Ainsworth et al.

    What have we learned from 15 years of free-air CO2 enrichment (FACE)? A meta-analytic review of the responses of photosynthesis, canopy properties and plant production to rising CO2

    New Phytol.

    (2005)
  • S. Asseng et al.

    Rising temperatures reduce global wheat production

    Nat. Clim. Chang.

    (2015)
  • S. Asseng et al.

    Uncertainty in simulating wheat yields under climate change

    Nat. Clim. Chang.

    (2013)
  • M. Bannyayan et al.

    Modelling the interactive effects of atmospheric CO2 and N on rice growth and yield

    Field Crops Res.

    (2005)
  • Cited by (37)

    • Comparison of process-based and statistical approaches for simulation and projections of rainfed crop yields

      2023, Agricultural Water Management
      Citation Excerpt :

      According to the results of three climatic projections, FFBPNN shows an increment in crop yields for warm and dry scenarios (approximately 0.8 tons/ha). Different studies show that using different models and climatic scenarios can lead to different results of crop yield projections, particularly based on machine learning methods, for both an increase and decrease under similar conditions (Çakır et al., 2014; Ewert et al., 2015; Matsumura et al., 2015; Gandhi et al., 2016; Cheng et al., 2017; Chlingaryan et al., 2018; Crane-Droesch, 2018; Tao et al., 2018; Kolberg et al., 2019; Leng and Hall, 2020; Tao et al., 2020; van Klompenburg et al., 2020; Chen et al., 2021b; Eini et al., 2021a). Studies on yield projections in Poland have been scarce to date, while there is a strong evidence on the effect of drought on decreasing yields in historical records (Łabędzki and Bąk, 2017; Oleksiak et al., 2022).

    View all citing articles on Scopus
    View full text