Implications of crop model ensemble size and composition for estimates of adaptation effects and agreement of recommendations

Simulation of winter wheat response to variable sowing dates and densities in a high-yielding environment

Crop multi-model ensembles (MME) have proven to be effective in increasing the accuracy of simulations in modelling experiments. However, the ability of MME to capture crop responses to changes in sowing dates and densities has not yet been investigated. These management interventions are some of the main levers for adapting cropping systems to climate change. Here, we explore the performance of a MME of 29 wheat crop models to predict the effect of changing sowing dates and rates on yield and yield components, on two sites located in a high-yielding environment in New Zealand. The experiment was conducted for 6 years and provided 50 combinations of sowing date, sowing density and growing season. We show that the MME simulates seasonal growth of wheat well under standard sowing conditions, but fails under early sowing and high sowing rates. The comparison between observed and simulated in-season fraction of intercepted photosynthetically active radiation (FIPAR) for early sown wheat shows that the MME does not capture the decrease of crop above ground biomass during winter months due to senescence. Models need to better account for tiller competition for light, nutrients, and water during vegetative growth, and early tiller senescence and tiller mortality, which are exacerbated by early sowing, high sowing densities, and warmer winter temperatures.

A public decision support system for the assessment of plant disease infection risk shared by Italian regions

Integrated pest management (IPM) practices proved to be efficient in reducing pesticide use and ensuring economic farming sustainability. Digital decision support systems (DSS) to support the adoption of IPM practices from plant protection services are required by European legislation. Available DSSs used by Italian plant protection services are heterogeneous with regards to disease forecasting models, datasets for their calibration, and level of integration in operational decision-making. This study presents the MISFITS-DSS, which has been jointly developed by a public research institution and nine regional plant protection services with the objective of harmonizing data collection and decision support for Italian farmers. Participatory approach allowed designing a predictive workflow relying on specific domain expertise, in order to explicitly match actual user needs. The DSS calibration entailed the risk of grapevine downy mildew infection (5-point scale from very low to very high), and phenological observations in 2012-2017 as reference data. Process-based models of primary and secondary infections have been implemented and tested via sensitivity analysis (Morris method) under contrasting weather conditions. Hindcast simulations of grapevine phenology, host susceptibility and disease pressure were post-processed by machine-learning classifiers to predict the reference infection risk. Results indicate that IPM principles are implemented by plant protection services since years. The accurate reproduction of grapevine phenology (RMSE = 4-14 days), which drove the dynamic of host susceptibility, and the use of weather forecasts as model inputs contributed to reliably predict the reference infection risk (88% balanced accuracy). We did a pioneering effort to homogenize the methodology to deliver decision support to Italian farmers, by involving plant protection services in the DSS definition, to foster a further adoption of IPM practices.

Evaluating Effects of Medium-Resolution Optical Data Availability on Phenology-Based Rice Mapping in China

The phenology-based approach has proven effective for paddy rice mapping due to the unique flooding and transplanting features of rice during the early growing season. However, the method may be greatly affected if no valid observations are available during the flooding and rice transplanting phase. Here, we compare the effects of data availability of different sensors in the critical phenology phase, thereby supporting paddy rice mapping based on phenology-based approaches. Importantly, our study further analyzed the effects of the spatial pattern of the valid observations related to certain factors (i.e., sideslips, clouds, and temporal window lengths of flooding and rice transplanting), which supply the applicable area of the phenology-based approach indications. We first determined the flooding and rice transplanting phase using in situ observational data from agrometeorological stations and remote sensing data, then evaluated the effects of data availability in this phase of 2020 in China using all Landsat-7 and 8 and Sentinel-2 data. The results show that on the country level, the number of average valid observations during the flooding and rice transplanting phase was more than ten for the integration of Landsat and Sentinel images. On the sub-country level, the number of average valid observations was high in the cold temperate zone (17.4 observations), while it was relatively lower in southern China (6.4 observations), especially in Yunnan–Guizhou Plateau, which only had three valid observations on average. Based on the multicollinearity test, the three factors are significantly correlated with the absence of valid observations: (R2 = 0.481) and Std.Coef. (Std. Err.) are 0.306 (0.094), −0.453 (0.003) and −0.547 (0.019), respectively. Overall, these results highlight the substantial spatial heterogeneity of valid observations in China, confirming the reliability of the integration of Landsat-7 and 8 and Sentinel-2 imagery for paddy rice mapping based on phenology-based approaches. This can pave the way for a national-scale effort of rice mapping in China while further indicating potential omission errors in certain cloud-prone regions without sufficient optical observation data, i.e., the Sichuan Basin.

Exploring the effects of land management change on productivity, carbon and nutrient balance: Application of an Ensemble Modelling Approach to the upper River Taw observatory, UK

Agriculture is challenged to produce healthy food and to contribute to cleaner energy whilst mitigating climate change and protecting ecosystems. To achieve this, policy-driven scenarios need to be evaluated with available data and models to explore trade-offs with robust accounting for the uncertainty in predictions. We developed a novel model ensemble using four complementary state-of-the-art agroecosystems models to explore the impacts of land management change. The ensemble was used to simulate key agricultural and environmental outputs under various scenarios for the upper River Taw observatory, UK. Scenarios assumed (i) reducing livestock production whilst simultaneously increasing the area of arable where it is feasible to cultivate (PG2A), (ii) reducing livestock production whilst simultaneously increasing bioenergy production in areas of the catchment that are amenable to growing bioenergy crops (PG2BE) and (iii) increasing both arable and bioenergy production (PG2A + BE). Our ensemble approach combined model uncertainty using the tower property of expectation and the law of total variance. Results show considerable uncertainty for predicted nutrient losses with different models partitioning the uncertainty into different pathways. Bioenergy crops were predicted to produce greatest yields from Miscanthus in lowland and from SRC-willow (cv. Endurance) in uplands. Each choice of management is associated with trade-offs; e.g. PG2A results in a significant increase of edible calories (6736 Mcal ha^-1) but reduced soil C (-4.32 t C ha^-1). Model ensembles in the agroecosystem context are difficult to implement due to challenges of model availability and input and output alignment. Despite these challenges, we show that ensemble modelling is a powerful approach for applications such as ours, offering benefits such as capturing structural as well as data uncertainty and allowing greater combinations of variables to be explored. Furthermore, the ensemble provides a robust means for combining uncertainty at different scales and enables us to identify weaknesses in system understanding.

Oil Palm Yield Estimation Based on Vegetation and Humidity Indices Generated from Satellite Images and Machine Learning Techniques

Palm oil has become one of the most consumed vegetable oils in the world, and it is a key element in profitable global value chains. In Costa Rica, oil palm cultivation is one of the three crops with the largest occupied agricultural area. The objective of this study was to explain and predict yield in safe time lags for production management by using free-access satellite images. To this end, machine learning methods were performed to a 20-year data set of an oil palm plantation located in the Central Pacific Region of Costa Rica and the corresponding vegetation indices obtained from LANDSAT satellite images. Since the best correlations corresponded to a one-year time lag, the predictive models Random Forest (RF), Least Absolute Shrinkage and Selection Operator (LASSO), Extreme Gradient Boosting (XGBoost), Recursive Partitioning and Regression Trees (RPART), and Neural Network (NN) were built for a Time-lag 1. These models were applied to all genetic material and to the predominant variety (AVROS) separately. While NN showed the best performance for multispecies information (r2 = 0.8139, NSE = 0.8131, RMSE = 0.3437, MAE = 0.2605), RF showed a better fit for AVROS (r2 = 0.8214, NSE = 0.8020, RMSE = 0.3452, MAE = 0.2669). The most relevant vegetation indices (NDMI, MSI) are related to water in the plant. The study also determined that data distribution must be considered for the prediction and evaluation of the oil palm yield in the area under study. The estimation methods of this study provide information on the identification of important variables (NDMI) to characterize palm oil yield. Additionally, it generates a scenario with acceptable uncertainties on the yield forecast one year in advance. This information is of direct interest to the oil palm industry.

Understanding spatial and temporal variability of N leaching reduction by winter cover crops under climate change

Winter cover crops are sown in between main spring crops (e.g. cash and forage crops) to provide a range of benefits, including the reduction of nitrogen (N) leaching losses to groundwater. However, the extent by which winter cover crops will remain effective under future climate change is unclear. We assess variability and uncertainty of climate change effects on the reduction of N leaching by winter oat cover crops. Field data were collected to quantify ranges of cover crop above-ground biomass (7 to 10 t DM/ha) and N uptake (70 to 180 kg N/ha) under contrasting initial soil conditions. The data were also used to evaluate the APSIM-NextGen model (R² from 62 to 96% and RMSEr from 7 to 50%), which was then applied to simulate cover crop and fallow conditions across four key agricultural locations in New Zealand, under baseline and future climate scenarios. Cover crops reduced N leaching risks for all location/scenario combinations but with large variability in space and time (e.g. 21 to 47% of fallow) depending on the climate change scenario. For instance, end-of-century estimates for northern (warmer) locations mostly showed non-significant effects of climate change on cover crop effectiveness and N leaching. In contrast for southern (colder) locations, there was a systematic increase in N leaching risks with climate change intensity despite a concomitant, but less than proportional, increase in cover crop effectiveness (up to ~5% of baseline) due to higher winter yields and N uptake. This implies that climate change may not only modify the geography of N leaching hotspots, but also the extent by which cover crops can locally reduce pollution risks, in some cases requiring complementary adaptive measures. The patchy- and threshold-nature of leaching events indicates that fine spatio-temporal resolutions are better suited to evaluate cover crop effectiveness under climate change.

Ensemble modelling, uncertainty and robust predictions of organic carbon in long-term bare-fallow soils

Simulation models represent soil organic carbon (SOC) dynamics in global carbon (C) cycle scenarios to support climate-change studies. It is imperative to increase confidence in long-term predictions of SOC dynamics by reducing the uncertainty in model estimates. We evaluated SOC simulated from an ensemble of 26 process-based C models by comparing simulations to experimental data from seven long-term bare-fallow (vegetation-free) plots at six sites: Denmark (two sites), France, Russia, Sweden and the United Kingdom. The decay of SOC in these plots has been monitored for decades since the last inputs of plant material, providing the opportunity to test decomposition without the continuous input of new organic material. The models were run independently over multi-year simulation periods (from 28 to 80 years) in a blind test with no calibration (Bln) and with the following three calibration scenarios, each providing different levels of information and/or allowing different levels of model fitting: (a) calibrating decomposition parameters separately at each experimental site (Spe); (b) using a generic, knowledge-based, parameterization applicable in the Central European region (Gen); and (c) using a combination of both (a) and (b) strategies (Mix). We addressed uncertainties from different modelling approaches with or without spin-up initialization of SOC. Changes in the multi-model median (MMM) of SOC were used as descriptors of the ensemble performance. On average across sites, Gen proved adequate in describing changes in SOC, with MMM equal to average SOC (and standard deviation) of 39.2 (±15.5) Mg C/ha compared to the observed mean of 36.0 (±19.7) Mg C/ha (last observed year), indicating sufficiently reliable SOC estimates. Moving to Mix (37.5 ± 16.7 Mg C/ha) and Spe (36.8 ± 19.8 Mg C/ha) provided only marginal gains in accuracy, but modellers would need to apply more knowledge and a greater calibration effort than in Gen, thereby limiting the wider applicability of models.

Towards quantification of the national water footprint in rice production of China: A first assessment from the perspectives of single-double rice

Rice is one of the most important crops in China, contributing to approximately 28% of total cereal yield. Despite substantial production, given that rice is a high water-consuming crop, the water shortage due to the irreversible decline in available water resources on a global scale induced by undergoing climate change will pose grave challenges to rice reproductive growth and related water resources utilization. As a consequence, investigating the responses of rice productivity and water consumption to more pronounced climate changes is of great significance for water resources sustainable utilization in terms of reducing irrigation water requirements and ensuring food security. Present water footprint (WF) methods do not calculate the weighted average of each WF component at the national level when evaluating the effects of prospective climate change upon rice production. The national water footprint (NWF), i.e. taking the share of each province in the total production of crops as weighting factors, has been regarded as an effective approach to determine where each WF component is originally located. In this study, the temporal change characteristics of NWF for single-rice (SR), early-rice (ER) and late-rice (LR) in different agro-ecological zones across China during 2001-2010 were assessed for the first time. The results exhibited that NWF of rice was an estimated 304,848 million cubic meters (MCM) per year. The SR accounted for the greatest portion of NWF, followed by ER and LR. The NWF rank was SR-V > SR-I > ER-VI > SR-IV > LR-III > LR-VI > SR-II > ER-III. The blue water footprint (WF_b) presents decreasing trends in most agro-ecological zones (SR-I, SR-II, SR-IV, ER-III and LR-VI), while green water footprint (WF_g) exhibits increasing trends within these regions. This study provides a beneficial approach for decision-making processes aiming at better agricultural water resources management strategies to alleviate water resources scarcity and reduce food risk in the context of surging demand, which will support agricultural water resources management of China towards a more balanced direction at the national level.

To what extent is climate change adaptation a novel challenge for agricultural modellers?

Modelling is key to adapting agriculture to climate change (CC), facilitating evaluation of the impacts and efficacy of adaptation measures, and the design of optimal strategies. Although there are many challenges to modelling agricultural CC adaptation, it is unclear whether these are novel or, whether adaptation merely adds new motivations to old challenges. Here, qualitative analysis of modellers' views revealed three categories of challenge: Content, Use, and Capacity. Triangulation of findings with reviews of agricultural modelling and Climate Change Risk Assessment was then used to highlight challenges specific to modelling adaptation. These were refined through literature review, focussing attention on how the progressive nature of CC affects the role and impact of modelling. Specific challenges identified were: Scope of adaptations modelled, Information on future adaptation, Collaboration to tackle novel challenges, Optimisation under progressive change with thresholds, and Responsibility given the sensitivity of future outcomes to initial choices under progressive change.