National and regional yield forecasts can provide important insights into agribusiness beyond the farm gate. The incorporation of dynamical climate models into these forecasting systems strengthen their predictive performance in many cases but may contribute inherent biases to the final yield estimates. Downscaling the native climate model output so that is suitable for crop simulation modelling can also present challenges in representing realistic conditions for plant growth from a climate model. This study evaluated the performance of an operational national wheat yield forecast system for the Australian wheatbelt using climatology and seasonal climate model-based input data, and introduces an alternative approach using a data fusion method. The crop forecasting system uses the APSIM wheat model to estimate water-limited potential yield. The climatology-based forecast tended to over predict national yield (high yield bias; 1.5 to 7% across forecast months), while the model-based method (using ACCESS-S1 dynamical model) tended to under predict yield (low yield bias; -5.9 to -0.5% across forecast months) and had a lower spread than climatology (10 to 50% lower across forecast months). The model-based forecast had skill in terms of accuracy and reliability during the second half of the season. The newly developed data fusion method used a weighting method calibrated for separate forecast locations (stations) to remove bias in the mean forecast yield and reduce ensemble spread. This resulted in improvements in the Australia-wide yield forecasts across all forecast start dates. This study provides a demonstration of how a data-driven approach can be applied to a crop forecast to improve accuracy and resolution of crop yield forecasts without the need for more computationally intensive downscaling approaches.

Agricultural productivity is already being impacted by climate change. There is very high confidence that the climate will continue to change with growing concerns on how this will impact productivity and financial resilience at the farm scale, along the supply chain and for the financial markets. Each region and farming domain has its own particular vulnerabilities to climate perturbations. Broadscale climate projections are useful indicators of change and possible climate futures but come with large uncertainty around many climatic features needed for agricultural assessments. This creates a frustrating and sometimes paralysing gap between the information needed and what can be supplied.Information is needed at the farm scale and for the detailed climate metrics that affect yield. To address this, vulnerability assessments can be conducted to stress test the viability of farming systems to a range of plausible futures. This allows us to identify the most resilient systems and explore no-regrets management options in the face of uncertainty.We explore this approach using a crop model to test the vulnerability of Australia’s wheat belt to a range of plausible climate futures. Here we test a scenario of 2 degree warming in different seasons through the year, varied across maximum and minimum temperatures, and with both increases and decreases to rainfall. This method does not decrease the uncertainty, but it does allow us to explore a more comprehensive risk exposure to the vulnerabilities and sensitivities of the cropping system. It also allows for nationwide vulnerability comparisons which reveal a greater vulnerability in the south of the country than in the north.The power of this approach is that we can invert the question to be ‘what sort of climate are crops most vulnerable to and how likely is this to occur?’ rather than relying on a subset of climate models to dictate a subset of future possible climate changes, subject to biases in downscaling methods. It also means that as climate projections are updated, we do not need to rerun the crop models, but only reassess how likely the most harmful climate changes are likely to be.

Using an extended 120‐year record of El Niño events, we distinguish between central Pacific (CP) and eastern Pacific (EP) types to show that the strength of CP events is a factor in the amplitude and sign of the impact on rainfall over southeastern Australia. Both weak and strong CP events cause widespread rainfall deficits in Australia during the onset phase from April to September. However, this relationship reverses over southeastern Australia including the Murray Darling Basin river catchment region for the strongest CP events after October, leading to positive rainfall anomalies during the mature phase of strong CP El Niños. This reversal can be explained by a change in the circulation over eastern Australia from drier, more westerly orientated flow to moister, more easterly onshore flow. These findings may help with seasonal prediction efforts to predict drought‐breaking rain such as occurred in early 2020.

Regional seasonal forecasting requires accurate simulation of the variability of local climate drivers. The South Pacific Convergence Zone (SPCZ) is a large region of low-level convergence, clouds and precipitation in the South Pacific, whose effects extend as far as northeast Australia (NEA). The location of the SPCZ is modulated by the El Niño-Southern Oscillation (ENSO) which causes rainfall variability in the region. Correctly simulating the ENSO-SPCZ teleconnection and its interplay with local conditions is essential for improving seasonal rainfall forecasts. Here we analyse the ability of the ACCESS-S1 seasonal forecast system to predict the SPCZ’s relationship with ENSO including its latitudinal shifts, zonal slope and rainfall magnitude between 1990 and 2012 for the December–January–February (DJF) season. We found improvements in ACCESS-S1’s SPCZ prediction capability compared to its predecessor (POAMA), although prediction of the slope is still limited. The inability of ACCESS-S1 to replicate seasons with a strong anti-zonal SPCZ slope is attributed to its atmospheric model. This has implications for accurate seasonal rainfall forecasts for NEA and South Pacific Islands. Future challenges in seasonal prediction facing regional communities and developers of coupled ocean–atmosphere forecast models are discussed.

Given the consequences and global significance of El Niño–Southern Oscillation (ENSO) events it is essential to understand the representation of El Niño diversity in climate models for the present day and the future. In recent decades, El Niño events have occurred more frequently in the central Pacific (CP). Eastern Pacific (EP) El Niño events have increased in intensity. However, the processes and future implications of these observed changes in El Niño are not well understood. Here, the frequency and intensity of El Niño events are assessed in models from phases 5 and 6 of the Coupled Model Intercomparison Project (CMIP5 and CMIP6), and results are compared to extended instrumental and multicentury paleoclimate records. Future changes of El Niño are stronger for CP events than for EP events and differ between models. Models with a projected La Niña–like mean-state warming pattern show a tendency toward more EP but fewer CP events compared to models with an El Niño–like warming pattern. Among the models with more El Niño–like warming, differences in future El Niño can be partially explained by Pacific decadal variability (PDV). During positive PDV phases, more El Niño events occur, so future frequency changes are mainly determined by projected changes during positive PDV phases. Similarly, the intensity of El Niño is strongest during positive PDV phases. Future changes to El Niño may thus depend on both mean-state warming and decadal-scale natural variability.

Foresight of grain yields prior to harvest would be empowering for many stakeholders along the supply chain from farmers through to bulk handlers, banks and insurance companies. Estimating Australian grain production ahead of harvest is difficult for many reasons including the highly variable year to year rainfall. The rainfall in the final months prior to harvest, can be crucial to final harvest totals. Here we explore the importance of rainfall from September 1, which broadly corresponds to the close of the top-dressing fertilizer application window, for the remaining cropping season in determining final yield. This is assessed via sensitivity analysis of water-limited wheat potential yield totals from historical climate in the APSIM crop model. At locations where the rainfall influences wheat yield, we compare three methods to forecast wheat yields that differ based on the climate data input: 1) climatology approach, which uses 30 years of observed climate data, 2) analogue climatology, which uses information from climate drivers (El-Nino Southern Oscillation and Indian Ocean Dipole) to create analogue years; and 3) dynamical climate forecasts from a general circulation model (ACCESS-S). We find that potential yields strongly depend on in-season plant available water (PAW) where years with high PAW are unaffected by the late season rainfall. Predicting the potential yield from analogue climatology (climate drivers) had the greatest skill, with smallest Root Mean Squared Error of 0.45 t/ha. This approach ranked first for 42% of the study locations compared to the climatology and ACCESS-S forecasting methods. This knowledge can help inform decision makers about the need to incorporate seasonal climate forecasts and the most appropriate climate forecasting method.

Seasonal climate forecasts from raw climate models at coarse grids are often biased and statistically unreliable for credible crop prediction at the farm scale. We develop a copula-based postprocessing (CPP) method to overcome this mismatch problem. The CPP forecasts are ensemble based and are generated from the predictive distribution conditioned on raw climate forecasts. CPP performs univariate postprocessing procedures at each station, lead time, and variable separately and then applies the Schaake shuffle to reorder ensemble sequence for a more realistic spatial, temporal, and cross-variable dependence structure. The use of copulas makes CPP free of strong distributional assumptions and flexible enough to describe complex dependence structures. In a case study, we apply CPP to postprocess rainfall, minimum temperature, maximum temperature, and radiation forecasts at a monthly level from the Australian Community Climate and Earth-System Simulator Seasonal model (ACCESS-S) to three representative stations in Australia. We evaluate forecast skill at lead times of 0–5 months on a cross-validation theme in the context of both univariate and multivariate forecast verification. When compared with forecasts that use climatological values as the predictor, the CPP forecast has positive skills, although the skills diminish with increasing lead times and finally become comparable at long lead times. When compared with the bias-corrected forecasts and the quantile-mapped forecasts, the CPP forecast is the overall best, with the smallest bias and greatest univariate forecast skill. As a result of the skill gain from univariate forecasts and the effect of the Schaake shuffle, CPP leads to the most skillful multivariate forecast as well. Further results investigate whether using ensemble mean or additional predictors can enhance forecast skill for CPP.

Cropping systems models have become an essential tool to simulate crop growth and yield at different scales to produce actionable information related to climate change, food security, land use and market dynamics (e.g. Waha et al., 2015; Wallach et al., 2016; Porwollik et al., 2017). Despite their increased usage and importance, various sources of uncertainty exist in the modelling process due to the impossibility to model the cropping system with complete determinism (Ramirez-Villegas et al. 2017). In this article, by uncertainty we mean any departure from the unachievable ideal of complete determinism (Walker et al., 2003). Uncertainty is prevalent in every step of crop modelling, starting from the field observations used for model development to value of inputs and parameters to the structure and design of model (Fig. 1).

Northern Australian agricultural industries are vulnerable to a highly variable climate. Year-to-year fluctuations in rainfall strongly affect pasture growth and increase the challenge of matching forage supply with stock numbers. A case study of climate drivers of pasture growth is conducted for the extensively-grazed cattle enterprises in the semi-arid tropics of north-eastern Australia. By assessing past climate, we explore which aspects of climate variability drive variations in pasture growth to inform effective developments in grazing management tools. Insights gained from these investigations could be important to the understanding of how we might integrate dynamical climate models into the decision making process. Currently, dynamical climate models are limited in their ability to provide skilful 6 month forecasts of total rainfall at the paddock scale. In their favour, we show that variables other than total rainfall are more important in forecasting pasture growth. Predicting total rainfall is dominated by the need to effectively predict the extreme events. We demonstrate that these extreme events do not significantly influence pasture growth as much of the rain received during extreme rainfall events is lost as run-off. A forecast would ideally be able to predict the number of rain days that fall when the soil is dry; incorporate the moisture content of soil at the beginning of the season; determine the likelihood of an early break to the wet season resulting in late winter and/or spring rainfall; and establish the interaction of El Nino Southern Oscillation with other climate modes. We stress however, that potential pasture growth is only one aspect of complex farm management decisions. Optimal decision making must also incorporate factors such as grazing management, cattle condition and longer time horizons as part of a whole of farm approach.

Input data aggregation affects crop model estimates at the regional level. Previous studies have focused on the impact of aggregating climate data used to compute crop yields. However, little is known about the combined data aggregation effect of climate (DAEc) and soil (DAEs) on irrigation water requirement (IWR) in cool-temperate and spatially heterogeneous environments. The aims of this study were to quantify DAEc and DAEs of model input data and their combined impacts for simulated irrigated and rainfed yield and IWR. The Agricultural Production Systems sIMulator Next Generation model was applied for the period 1998-2017 across areas suitable for potato (Solanum tuberosum L.) in Tasmania, Australia, using data at 5, 15, 25 and 40 km resolution. Spatial variances of inputs and outputs were evaluated by the relative absolute difference (rAD¯) between the aggregated grids and the 5 km grids. Climate data aggregation resulted in a rAD¯ of 0.7-12.1%, with high values especially for areas with pronounced differences in elevation. The rAD¯ of soil data was higher (5.6-26.3%) than rAD¯ of climate data and was mainly affected by aggregation of organic carbon and maximum plant available water capacity (i.e. the difference between field capacity and wilting point in the effective root zone). For yield estimates, the difference among resolutions (5 km vs. 40 km) was more pronounced for rainfed (rAD¯ = 14.5%) than irrigated conditions (rAD¯ = 3.0%). The rAD¯ of IWR was 15.7% when using input data at 40 km resolution. Therefore, reliable simulations of rainfed yield require a higher spatial resolution than simulation of irrigated yields. This needs to be considered when conducting regional modelling studies across Tasmania. This study also highlights the need to separately quantify the impact of input data aggregation on model outputs to inform about data aggregation errors and identify those variables that explain these errors.

As a cropping season progresses yield forecasts become more reliable. Optimal management strategies however rely on early estimates of climate and yield. These estimates are usually derived from the historic range of climate variations applied to current crop conditions. Early in the season these predictions are wide ranging as they incorporate all past historic climate variability and hence a large range of possible yields. Dynamical seasonal climate models offer the opportunity to narrow this range contingent on the models having adequate predictive skill. This study explores the benefits of using a climate model over historical climate to predict wheat yield in the Australian cropping zone throughout the cropping season. We take an ensemble of daily outputs of temperature, radiation and rainfall from a seasonal climate model (POAMA) and apply a simple downscaling and calibration to align with 57 stations across the Australian cropping zone. These data are then used as an input to a crop model (APSIM) to translate seasonal conditions into a yield prediction. Simulations deploy historic weather data up to a date on which forecast data replace measured data. Here we used a range of dates (April to October) through the cropping season for the period 1981 to 2015, to determine where and when the forecast is skilful compared to using the full weather record up to harvest. The forecasts are categorised in three yield categories low (decile 1-3), average (4-7) and high (8-10) and determined to be 'misleading' if they predict low instead of high or vice versa. In the west and south of Australia less than 3 years in 20 give a misleading forecast in April, and less than 1 in 20 years by August. The forecast for east of Australia has less skill primarily due to a strong rainfall bias with the climate model not being able to simulate the correct daily rainfall patterns. Compared to the predictions gained from using the full range of historical climate, POAMA derived forecasts have a narrower prediction range than the climatology driven ones, however this comes at the expense of a higher number of misleading forecasts. Nevertheless, in June (August) the POAMA driven simulations have a greater than 65% (80%) chance of being in the correct or one category out, which was higher than using climatology in each region at the same lead time. The baseline set by this study demonstrates the potential utility of dynamic climate models to predict yield, which should only improve with on-going advances in climate modelling and techniques in downscaling.

Increasing sea surface temperatures (SSTs) are predicted to adversely impact coral populations worldwide through increasing thermal bleaching events. Future bleaching is unlikely to be spatially uniform. Therefore, understanding what determines regional differences will be critical for adaptation management. Here, using a cumulative heat stress metric, we show that characteristics of regional SST determine the future bleaching risk patterns. Incorporating observed information on SST variability, in assessing future bleaching risk, provides novel options for management strategies. As a consequence, the known biases in climate model variability and the uncertainties in regional warming rate across climate models are less detrimental than previously thought. We also show that the thresholds used to indicate reef viability can strongly influence a decision on what constitutes a potential refugia. Observing and understanding the drivers of regional variability, and the viability limits of coral reefs, is therefore critical for making meaningful projections of coral bleaching risk.

Coupled general circulation models (CGCMs) simulate a diverse range of El Niño–Southern Oscillation behaviors. “Double peaked” El Niño events—where two separate centers of positive sea surface temperature (SST) anomalies evolve concurrently in the eastern and western equatorial Pacific—have been evidenced in Coupled Model Intercomparison Project version 5 CGCMs and are without precedent in observations. The characteristic CGCM double peaked El Niño may be mistaken for a central Pacific warming event in El Niño composites, shifted westwards due to the cold tongue bias. In results from the Australian Community Climate and Earth System Simulator coupled model, we find that the western Pacific warm peak of the double peaked El Niño event emerges due to an excessive westward extension of the climatological cold tongue, displacing the region of strong zonal SST gradients towards the west Pacific. A coincident westward shift in the zonal current anomalies reinforces the western peak in SST anomalies, leading to a zonal separation between the warming effect of zonal advection (in the west Pacific) and that of vertical advection (in the east Pacific). Meridional advection and net surface heat fluxes further drive growth of the western Pacific warm peak. Our results demonstrate that understanding historical CGCM El Niño behaviors is a necessary precursor to interpreting projections of future CGCM El Niño behaviors, such as changes in the frequency of eastern Pacific El Niño events, under global warming scenarios.

Accelerated warming and hiatus periods in the long-term rise of Global Mean Surface Temperature (GMST) have, in recent decades, been associated with the Interdecadal Pacific Oscillation (IPO). Critically, decadal climate prediction relies on the skill of state-of-the-art climate models to reliably represent these low-frequency climate variations. We undertake a systematic evaluation of the simulation of the IPO in the suite of Coupled Model Intercomparison Project 5 (CMIP5) models. We track the IPO in pre-industrial (control) and all-forcings (historical) experiments using the IPO tripole index (TPI). The TPI is explicitly aligned with the observed spatial pattern of the IPO, and circumvents assumptions about the nature of global warming. We find that many models underestimate the ratio of decadal-to-total variance in sea surface temperatures (SSTs). However, the basin-wide spatial pattern of positive and negative phases of the IPO are simulated reasonably well, with spatial pattern correlation coefficients between observations and models spanning the range 0.4–0.8. Deficiencies are mainly in the extratropical Pacific. Models that better capture the spatial pattern of the IPO also tend to more realistically simulate the ratio of decadal to total variance. Of the 13% of model centuries that have a fractional bias in the decadal-to-total TPI variance of 0.2 or less, 84% also have a spatial pattern correlation coefficient with the observed pattern exceeding 0.5. This result is highly consistent across both IPO positive and negative phases. This is evidence that the IPO is related to one or more inherent dynamical mechanisms of the climate system.

Climate models consistently project a substantial decrease in the Indonesian Throughflow (ITF) in response to enhanced greenhouse warming. On interannual timescales ITF changes are largely related to tropical Pacific wind variability. However, on the multidecadal timescales investigated here we demonstrate that regional winds and associated changes in the upper ocean circulation cannot explain the projected ITF decrease. Instead, the decrease is related to a weakening in the northward flow of deep waters entering the Pacific basin at ~40°S and an associated reduction in the net basin‐wide upwelling to the north of the southern tip of Australia. This can be traced back to consistent changes in the Antarctic Circumpolar Current and Southern Ocean overturning, although questions still remain as to the ultimate drivers. In contrast to the ITF decrease, substantial projected changes to the upper ocean circulation of the Pacific basin are well explained by robust changes in the surface winds.

In this article, we analyse the current Australian diet and discuss how climate change might influence the foods we eat and consumer behaviour in coming decades. As climate change proceeds, one likely outcome is that key Australian agricultural regions will experience warmer, drier conditions with more frequent and intense drought and extreme events. These conditions will place pressure on Australia's ability to maintain the quality and quantity of the food it now produces. Amongst other pressures, reduced agricultural supply may contribute to higher prices for grains, red meat, fresh fruit and vegetables; reduced quality of produce due to an increase in pests and disease; and for water, increased costs of treatment. The risks from changes to diet correlate with socioeconomic disadvantage. Wealthier groups lend to spend more on quality produce. For most Australians, but particularly those that are vulnerable and food insecure, increased prices will lead to the consumption of cheaper and lower quality foods, changing diet composition away from healthy options, and exacerbating health issues. 1he interplay between climate, agriculture, economics and human health is complex. To improve, or even maintain, the health of Australians will likely cost more in the future for individuals and for our health system. Now is the time to explore and better understand these relationships in order to prepare for the near future.

Some of the most important development goals for the countries and territories of the Western and Central Pacific Ocean (WCPO) involve the sustainable management of their fisheries in light of environmental, economic and social uncertainties. The responses of fish populations to variability in the marine environment have implications for decision making processes associated with resource management. There is still considerable uncertainty in estimating the responses of tuna populations to short-to-medium-term variability and longer-term change in the oceanic environment. A workshop was organised to examine how advances in oceanography, fisheries science and fisheries economics could be applied to the tuna fisheries of the WCPO and in doing so identify research priorities to improve understanding relevant to progressing management. Research priorities identified included: (i) improved parameterisation of end to end ecosystem model components, processes and feedbacks through expanded biological observations and incorporation of higher resolution climate models; (ii) development of seasonal and inter-annual forecasting tools enabling management responses to short-term variability in tuna distributions and abundances; (iii) improved understanding of the population dynamics of and the energy transfer efficiencies between food web components; (iv) assessment of the optimal value of access rights and overall fishery value under multiple scenarios of tuna distribution and abundance and influences on decision making by fisheries managers and fleets and (v) development of management strategy evaluation frameworks for utilisation in the implementing and testing of fishery management procedures and to help prioritise research directions and investment. Issues discussed and research priorities identified during the workshop have synergies with other internationally managed fisheries and therefore are applicable to many other fisheries.

Multi-disciplinary approaches that combine oceanographic, biogeochemical, ecosystem, fisheries population and socio-economic models are vital tools for modelling whole ecosystems. Interpreting the outputs from such complex models requires an appreciation of the many different types of modelling frameworks being used and their associated limitations and uncertainties. Both users and developers of particular model components will often have little involvement or understanding of other components within such modelling frameworks. Failure to recognise limitations and uncertainties associated with components and how these uncertainties might propagate throughout modelling frameworks can potentially result in poor advice for resource management. Unfortunately, many of the current integrative frameworks do not propagate the uncertainties of their constituent parts. In this review, we outline the major components of a generic whole of ecosystem modelling framework incorporating the external pressures of climate and fishing. We discuss the limitations and uncertainties associated with each component of such a modelling system, along with key research gaps. Major uncertainties in modelling frameworks are broadly categorised into those associated with (i) deficient knowledge in the interactions of climate and ocean dynamics with marine organisms and ecosystems; (ii) lack of observations to assess and advance modelling efforts and (iii) an inability to predict with confidence natural ecosystem variability and longer term changes as a result of external drivers (e.g. greenhouse gases, fishing effort) and the consequences for marine ecosystems. As a result of these uncertainties and intrinsic differences in the structure and parameterisation of models, users are faced with considerable challenges associated with making appropriate choices on which models to use. We suggest research directions required to address these uncertainties, and caution against overconfident predictions. Understanding the full impact of uncertainty makes it clear that full comprehension and robust certainty about the systems themselves are not feasible. A key research direction is the development of management systems that are robust to this unavoidable uncertainty.