Vassili Kitsios

Climate variability is increasingly recognised as an important driver of infectious disease dynamics; however, climate-informed forecasts of influenza burden remain limited in Australia. Such forecasts could support timely public health preparedness and response. This study aimed to develop a data-driven Bayesian framework to improve short-term influenza forecasting across Australian states and territories by incorporating climate information. We used a spatiotemporal Bayesian hierarchical model to forecast laboratory-confirmed monthly influenza notifications from 2008 to 2022 (excluding the COVID-19 period). Monthly temperature and relative humidity data were included as predictors, with temperature modelled using contemporaneous and one-month lagged effects, and humidity categorised into three levels. The model incorporated region-specific temporally autocorrelated random effects and a time-invariant spatial random effect. Five model variants, ranging from a baseline without meteorological predictors to models with linear, nonlinear, and interaction effects, were evaluated using a scaled root-mean-square error metric. Forecasts were generated for lead times of 1–12 months using a one-month sliding training window. Sensitivity analysis varied training window duration (3–10 years) to identify the minimum data required for stable forecasts. Climate-informed models consistently outperformed the baseline model at short to intermediate lead times. Temperature-based models, particularly those allowing for nonlinear and lagged effects, achieved the highest forecast skill, with improvements observed up to lead times of 4–5 months. Reliable forecasts were obtained using as little as five years of data. Relative humidity provided limited additional predictive value, and forecast skill declined beyond approximately six months. Spatial and temporal random effects indicated persistent regional and temporal variability, while exposure-response analyses showed increased influenza risk at lower temperatures, especially for lagged temperature. Temperature is a key driver of short-term influenza dynamics in Australia. Climate-informed Bayesian models provide skilful and operationally relevant forecasts, supporting the development of early warning systems for public health decision-making.

In global subseasonal forecasting using dynamical models, correcting the systematic biases of Madden–Julian Oscillation (MJO) predictions has proven critical, particularly due to issues of rapid amplitude damping and phase distortion. To address these biases, recent studies have demonstrated that deep learning offers a promising solution by learning mappings that directly translate biased dynamical model outputs to ground truth MJO indices, effectively serving as post‐processing bias correctors. These approaches implicitly assume that a single neural network can extract all relevant features necessary for bias correction across different forecast time steps. However, we observe that the relationships between forecast lead times and their corresponding corrections are more intricate: features relevant to short range forecasting error differ significantly from those that govern long range forecasting errors. Consequently, direct mapping strategies often result in suboptimal performance, excelling in either short or long range forecasts, but rarely both. To address this challenge, we propose a unified attention recurrent neural network framework that processes the full sequence of dynamical forecasts using a masked input tensor, enabling the model to extract temporally contextualized features across all lead times. Additionally, we introduce a phase regulated loss function that explicitly penalizes errors in both amplitude and phase, capturing the cyclical structure of MJO more effectively than traditional mean squared error based objectives. Comprehensive evaluations on operational subseasonal to seasonal reforecast data sets demonstrate that our method significantly improves forecast accuracy and lead‐time consistency.

We describe and evaluate the performance of a spherical Fourier harmonic neural operator climate emulator in training and autoregressive inference of the tropical atmosphere over the ERA5 reanalysis period 1985-2024. Inference and ablation studies performed with an ensemble of independent models at varying resolutions demonstrate stability and physical consistency with reliable emulation of key internal climate modes relevant to tropical variability. The models exhibit low biases over 40 year autoregressive in-sample inference for the modes considered. Examination of out-of-sample inference and ablation studies indicate that long term forecast skill of the key tropical climate processes and teleconnections is dependent on accurate prediction of large scale variations in the incident short wave radiation and to a lesser degree outward longwave radiation. Dependencies on resolution and the impact of inclusion of surface forcing as an additional boundary constraint are described.

This paper introduces a Bayesian hierarchical modeling framework within a fully probabilistic setting for crop yield estimation, model selection, and uncertainty forecasting under multiple future greenhouse gas emission scenarios. By informing on regional agricultural impacts, this approach addresses broader risks to global food security. Extending an established multivariate econometric crop-yield model to incorporate country-specific error variances, the framework systematically relaxes restrictive homogeneity assumptions and enables transparent decomposition of predictive uncertainty into contributions from climate models, emission scenarios, and crop model parameters. In both in-sample and out-of-sample analyses focused on global wheat production, the results demonstrate significant improvements in calibration and probabilistic accuracy of yield projections. These advances provide policymakers and stakeholders with detailed, risk-sensitive information to support the development of more resilient and adaptive agricultural and climate strategies in response to escalating climate-related risks.

Climate change poses a serious threat to every sector of the economy. Agricultural sector is particularly vulnerable, due to its exposure to extreme conditions, temperature increases and systematic precipitation redistribution. How this susceptibility impacts crop production will have substantial repercussions on policies related to food security. Projecting the prospective impacts of climate change on crop production necessitates a comprehensive modelling system outlining crop responses to future conditions. Here we introduce a multivariate autoregressive econometrics model that includes a time-varying non-linear variable to account for the decreasing impact of technology on crop yields. Our model is designed to capture the relationships between technology, climate variables and the annual growth rate in crop yield across the world’s producing regions. Utilizing historical national crop production data and climate variables from 1961 to 2018, the developed model outperforms traditional panel regression methods. Additionally, a novel machine learning climate model emulator allows efficient estimation of crop production growth under a multitude of carbon equivalent emissions scenarios. Our key finding is that technological effects are prone to have a diminishing impact on wheat and rice production over time and may not adequately offset the negative effects of climate change under certain future emission scenarios. Naive assumptions surrounding technology result in overestimates of production exceeding 200% for wheat and 150% for rice. Our study implies that integrated assessment and other economic models that use oversimplified climate damage functions can compound inaccuracies in production estimates with adverse repercussions on policy decisions.

A reduced-order model (ROM) of the global oceans is developed by projecting the hydrostatic Boussinesq equations of motion onto a proper orthogonal decomposition (POD) basis. Three-dimensional POD modes are calculated from the ocean fields of an ensemble climate reanalysis dataset. The coefficients in the POD ROM are calculated using a regression approach. The performance of various POD ROM configurations are assessed. Each configuration is derived from an alternate sea-water equation of state, linking the density and temperature fields. POD ROM variants incorporating an equation of state in which density is a quadratic function of temperature, are able to reproduce the statistics of the large-scale structures at a fraction of the computational cost required to numerically simulate this flow. Due to the speed and efficiency of calculation, such reduced-order models of the global geophysical system will enable researchers and policy makers to assess the physical risk for a broader range of potential future climate scenarios.

Turbulence is the most important, ubiquitous, and difficult problem of classical physics. Feynman viewed it as essentially unsolved, without a rigorous mathematical basis to describe the statistical dynamics of this most complex of fluid motion. However, the paradigm shift came in 1959, with the formulation of the Eulerian direct interaction approximation (DIA) closure by Kraichnan. It was based on renormalized perturbation theory, like quantum electrodynamics, and is a bare vertex theory that is manifestly realizable. Here, we review some of the subsequent exciting achievements in closure theory and subgrid modelling. We also document in some detail the progress that has been made in extending statistical dynamical turbulence theory to the real world of interactions with mean flows, waves and inhomogeneities such as topography. This includes numerically efficient inhomogeneous closures, like the realizable quasi-diagonal direct interaction approximation (QDIA), and even more efficient Markovian Inhomogeneous Closures (MICs). Recent developments include the formulation and testing of an eddy-damped Markovian anisotropic closure (EDMAC) that is realizable in interactions with transient waves but is as efficient as the eddy-damped quasi-normal Markovian (EDQNM). As well, a similarly efficient closure, the realizable eddy-damped Markovian inhomogeneous closure (EDMIC) has been developed. Moreover, we present subgrid models that cater for the complex interactions that occur in geophysical flows. Recent progress includes the determination of complete sets of subgrid terms for skilful large-eddy simulations of baroclinic inhomogeneous turbulent atmospheric and oceanic flows interacting with Rossby waves and topography. The success of these inhomogeneous closures has also led to further applications in data assimilation and ensemble prediction and generalization to quantum fields.