Driven by growing demands for higher resolution and quality in hydrometeorological applications, precipitation products are increasingly being developed at sub-daily and kilometre-scale resolutions by blending information from multiple data sources. However, these developments face substantial modelling complexity, as the integration of multiple heterogeneous sources expands the inter- and intra-source error relationships that need to be represented, particularly when applied across large spatial domains. To address this challenge, this study presents the theoretical formulation and scalable implementation strategies for generalised multi-source precipitation blending within a statistical interpolation framework. We use an error-informed localisation concept, which formalises the insight that, in most blending scenarios, only observations that are both spatially representative and characterised by relatively low errors exert meaningful influence on blended precipitation field. Motivated by this concept, we develop a highly efficient implementation that approximates source influence by selecting only the nearest grid cell from each additional spatial input, enabling a substantial reduction in covariance modelling while preserving the essential blending characteristics. The framework is demonstrated in a three-source blending experiment integrating satellite, radar, and gauge data to produce 2-km and hourly rainfall fields over two radar domains in Australia. Qualitative assessment of precipitation and error spatial structures during extreme events, together with quantitative evaluation using the fractions skill score (FSS), show that the simplified implementation retains blending performance comparable to a comprehensive scheme, while requiring only a fraction of the computational cost. This work provides both conceptual understanding and practical implementation strategies to support large-scale, high-resolution, multi-source precipitation product development.

Verification of atmospheric reanalysis products is crucial for their application in extreme weather and climate-related research. This study evaluates tropical cyclone (TC) characteristics and related variables from 1990 to 2018 over Australia (95-160 degrees E and 0-30 degrees S) in three reanalysis products - the recently developed Australian Bureau of Meteorology's Atmospheric high-resolution Regional Reanalysis for Australia Version 2 (BARRA-R2), its predecessor BARRA-R (both at 12-km spatial resolution) and the widely used European Centre for Medium-Range Weather Forecasts (ECMWF)'s Global Reanalysis Version 5 (ERA5) at 31-km spatial resolution. TCs detected in these reanalyses, using the Okubo-Weiss-Zeta Parameter detection and tracking scheme, are compared with observations from the Bureau's TC database. All three products simulated more than 50% of the observed TC frequency, with ERA5 achieving a higher hit rate of 77% compared with BARRA-R2 (68%) and BARRA-R (53%). Most missed cases involved non-severe TCs. ERA5 showed a clear decline in annual TC frequency consistent with observations, whereas BARRA-R displayed a weak upward trend and BARRA-R2 a statistically insignificant decline. Large differences emerged in surface wind speed and gusts: BARRA-R represented TC surface winds better than ERA5, and BARRA-R2 produced slightly improved gusts compared with ERA-5. Case studies show that temporal evolution is generally well represented in all products, though ERA5 tends to maintain peak intensity for longer, whereas BARRA products sometimes show shifted timing of peak intensity. General discrepancies are attributed to resolution limitations, inherent differences between best-track and gridded data, underestimation of TC peak intensity and wider model forecast constraints.

The Australian offshore wind sector is expanding, yet lacks a comprehensive, high-resolution national assessment of past and projected ocean wind climate. Understanding climate patterns over the next 30-50 years is vital for resource planning, safety, and the resilience of offshore wind infrastructure. This study applies the Bureau of Meteorology Atmospheric Regional Projections for Australia (BARPA) Coordinated Regional Downscaling Experiment (CORDEX) seven-model ensemble to assess surface wind speed projections and their effect on offshore wind energy production. Across declared offshore wind farm regions, the projected signal of change is small and generally lower than seasonal and inter-annual variability. Areas with slight reductions in wind power density and production also show minor increases in variability, which may reduce turbine efficiency. These changes, however, fall within the uncertainty range of climate model projections, leaving future wind resource trends uncertain. Western Australia is an exception, showing a notable wintertime decrease in wind power density under both low-and high-emission scenarios. Overall, the reliability of wind energy production remains largely unaffected, with only slight reductions in the next 30 to 50 years, ranging from 0.1% to 2.6%, depending on the region.

This paper documents AUS2200, a community-driven, high-resolution limited-area modelling project for Australia based on the Met Office Unified Model (UM) coupled to the Joint UK Land Environment Simulator to represent the land surface. Developed through a national partnership involving the Australian Research Council Centre of Excellence for Climate Extremes (CLEX), Bureau of Meteorology (BoM), National Computational Infrastructure (NCI) and Australian Earth System Simulator National Research Infrastructure (ACCESS-NRI), AUS2200 marks a significant advancement in limited-area modelling efforts in the Australian university community. AUS2200 features a convection-permitting configuration with 2.2-km grid spacing, covering the entirety of the Australian continent and portions of surrounding oceans. Its large domain at convection-permitting scales allows simultaneous resolution of both large- and small-scale atmospheric processes. This capability supports scientific investigations into key atmospheric phenomena, including multiscale interactions, across a broad range of spatial and temporal scales, from continent-wide systems to localised events, and across diverse climatic regions spanning the tropics to the mid-latitudes. This paper provides an overview of the AUS2200 project, detailing its overarching aims, modelling framework including model configuration and optimisation efforts, contributions to scientific research and community development, and future directions. Early results from collaborative, cross-organisational efforts to study diverse atmospheric processes and high-impact weather events, including the 2019–20 Black Summer bushfires and record-breaking extreme rainfall events, are also presented. These investigations are contextualised within the broader scope of seasonal and climate variability, highlighting the project’s importance for advancing scientific research and addressing broader societal challenges.

Abstract. High-resolution precipitation information is essential for hydrometeorological applications such as extreme weather monitoring, flood forecasting, and disaster risk management. Despite substantial advances in satellite, radar, and gauge observations, producing kilometre-resolution sub-daily precipitation analyses over continental domains remains challenging due to heterogeneous data availability, scale mismatches, and computational constraints. This study presents the design and trial implementation of BRAIN (blended rainfall), a continental-scale, kilometre-resolution hourly precipitation analysis for Australia. In this initial implementation, BRAIN integrates three key data sources from the Australian Bureau of Meteorology: geostationary satellite rainfall estimates from Himawari (2 km, 10 min), radar rainfall estimates (1 km, 5 min), and sub-daily rain gauge observations. The trialled system incorporates quality control, spatiotemporal aggregation, bias correction, and a simplified statistical interpolation configuration designed to balance performance with scalability at continental scale. Source contributions are weighted according to their spatial and temporal error characteristics, allowing each data type to influence the analysis where it is most informative. The trial implementation produces hourly rainfall fields at 2-km resolution across the Australian continent. Evaluation for the trial period 2022–2023 indicates that the blended analysis improves upon satellite-only, radar-only, and satellite–gauge products, outperforms the gauge-based interpolation approach currently used in flood operations, and provides more spatially coherent and detailed rainfall structures than the current daily operational product. These results demonstrate the feasibility and utility of the proposed design and trial implementation in the Australian context, with potential extension to long-term historical reconstruction and near–real-time applications. The system design is flexible and scalable, enabling future upgrades such as finer spatial and temporal resolutions and the incorporation of additional data sources. Beyond the Australian context, this study provides an additional reference for large-scale multi-source precipitation analysis at kilometre and hourly resolutions.

The Australian Climate Service (ACS) was established in 2021 to uplift Australian climate science and services to support resilience in the face of increasing impacts from weather and climate extremes. Formed as a partnership across the Federal Government of Australia, it consists of four Partner Organisations: The Australian Bureau of Statistics (ABS), the Bureau of Meteorology (BoM), Commonwealth Scientific and Industrial Research Organisation (CSIRO), and Geoscience Australia (GA) forming a virtual agency. Here, we describe the approach taken by the BoM and CSIRO to develop the climate parts of this service, provide examples of successful delivery, and reflect on lessons learnt. The approach to development borrowed from practices established for seasonal climate prediction services, drawing on the Global Framework for Climate Services and World Meteorological Organization good practice guidance documents. In establishing the service, we found that many concepts and frameworks mapped readily from other timescales. For example, the concept of operations, used for BoM weather services and methods benchmarking, was applied to areas such as climate model selection and calibration. Having modellers, scientists, and climate services staff from multiple agencies collaborate within activities as part of a single value chain was critical for delivery, and readily achievable through an end-to-end program approach.

This study applies a benchmarking framework to assess a 34-member ensemble of regional climate models that have dynamically downscaled Coordinated Model Intercomparison Project (CMIP6) models over the Australasian region. Four modelling centres contributed regional climate models to this ensemble using three regional climate models (RCMs) and a total of five model configurations. The RCMs compared are the Conformal Cubic Atmospheric Model (CCAM), the Weather Research and Forecast (WRF) model and the Bureau Atmospheric Regional Projections for Australia (BARPA-R). Assessment is conducted over the Australian continent using a separation into four major climate zones over a 30-year historical climatological period (1985–2014). Rainfall and near-surface temperatures are compared against six benchmarks measuring mean state patterns, spatial and temporal variance, seasonal cycles, long-term trends and selected extreme indices. Benchmark thresholds are derived either from previous studies or comparison with the driving model ensemble. Major model biases vary between ensemble members and include dry biases in northern and southern Australia, winter wet biases and a persistent low bias in the winter diurnal temperature range across all the modelling centres. Daily variability at large length scales is comparable in the driving global climate model and downscaled regional climate model length scales, and long-term trends are largely determined by the driving global climate model. Overall, the ensemble was deemed to be fit for purpose for impact studies. Strengths and weaknesses of the systematic benchmarking framework used here are discussed.

The third version of the Regional Atmosphere and Land (RAL3) science configuration is documented. Developed through international partnerships, RAL configurations define settings for the Unified Model atmosphere and Joint UK Land Environment Simulator (JULES) when applied across timescales with kilometre and sub-kilometre-scale model grids. The RAL3 configuration represents a major advance compared to previous versions by delivering a common science definition suitable for application to tropical and mid-latitude regions. Developments within RAL3 include the introduction of a double-moment microphysics scheme and a bimodal cloud scheme, replacing use of a single-moment scheme and different cloud schemes for mid-latitudes and tropics in previous versions. Updates have been implemented to the boundary layer scheme and a consolidation of land model settings to be more consistent with global atmosphere and land (GAL) science configurations. Physics developments aimed to address priorities for model performance improvement identified by users. This paper documents the RAL3 science configuration, including a series of iterative revisions delivered since its first release, and their characteristics. Evidence is provided from the variety of assessments of RAL3, relative to the previous version (RAL2). Collaborative development and evaluation across organizations have enabled evaluation across a range of domains, grid spacing and timescales. The analysis indicates more realistic precipitation distributions, improved representation of clouds and of visibility, a continued trend to more realistic representation of convection, and reduced near-surface wind speeds but a persistent cold-temperature bias. Overall the convective-scale verification scores and climatological model distributions relative to observations improve for the majority of variables. Ensemble results show improvements to the spread-error relationship. User feedback from subjective assessment activities has also been positive. Differences between RAL3 revisions and RAL2 are further illustrated through a process-based analysis of a convective system over the UK. The latest RAL3 configuration (RAL3.3) is therefore recommended for research, operational numerical weather prediction, and climate production at kilometre and sub-kilometre scales.

Trends in seasonal average dry-bulb temperature T, wet-bulb temperature TWB, and dewpoint temperature TD over the last 42 years (1980-2021) were analyzed over the Southeast (SE) Asian and tropical northern Australian region. All three variables show broad positive trends in most areas, except for relatively weak, nonsignificant trends over northern Australia. Trends in TD and TWB are as high as 0.5 degrees C decade21 in parts of SE Asia, with generally slightly smaller observed trends in TWB than in T. Trends over the ocean tend to be less variable and slightly weaker on average. We observe an increased occurrence of extreme events in recent years (the 2010s) compared to earlier decades. The T, TWB, and TD extremes occur substantially more frequently during El Ni & ntilde;o years, in addition to an underlying increasing trend. SIGNIFICANCE STATEMENT: Heat stress, a combination of high temperature and humidity, is a major issue in the tropics, where population density is high. Increasing temperature and humidity trends and the occurrence of extreme dry-and wet-bulb temperatures across Southeast (SE) Asia over recent decades are of particular concern for human and ecosystem health.

Statistical calibration of raw forecasts from numerical weather prediction (NWP) models aims to produce probabilistic forecasts that are as skilful as possible and reliable in ensemble spread. As lead time increases, the underlying skill of the raw forecasts diminishes. Some calibration models are developed based on a joint probability theory, allowing the calibrated forecasts to approach observed climatology when the underlying skill is very low. The observed climatology serves as a baseline accuracy for probabilistic forecasts and needs to be estimated as accurately as possible. However, accurate estimation of climatology requires quality and lengthy data records. There are many locations where forecast calibration is desired, but observations are not available. This study proposes a method to calibrate NWP forecasts at locations without rain gauge observations by (a) estimating the observed climatology from gridded reanalysis data to generate the background climatology fields; and (b) combining the reanalysis climatology with the remaining calibration model parameters transferred from a donor gauge with similar reanalysis climatology. The method is tested on daily precipitation forecasts using the Seasonally Coherent Calibration (SCC) model and evaluated at 50 masked-out gauged locations across Australia. It is shown to produce skillful forecasts up to 7 days ahead, with 2.16 % skill score reduction compared to a benchmark model which is established ideally using training data exclusively from the test sites. Due to the use of reanalysis data with adequate spatial and temporal coverage, the reliance on rain gauge density does not significantly impact the SCC_U forecast skill.

High-resolution precipitation products (e.g., km-scale and hourly) are essential for capturing extreme rainfall dynamics and improving hydrological models for flood forecasting and related applications. However, single-source estimates—whether from gauges, radar, or satellites—suffer from inherent limitations such as sparse spatial coverage, limited observational extent, and retrieval uncertainties. Multi-source precipitation combinations can potentially provide a more accurate estimate by using the strengths of each source. This review presents globally relevant methodologies for constructing high-resolution precipitation datasets, with a focus on sub-daily temporal and km-scale spatial resolutions. Particular emphasis is placed on their applicability to the development of a new multi-source precipitation dataset for Australia (BRAIN: Blended Rain). We first compile and assess precipitation data sources specific to Australia, particularly those readily available to the Australian Bureau of Meteorology. These include ground-based gauge observations, radar estimates, satellite-derived products, reanalysis datasets, and numerical weather prediction outputs. While the primary emphasis is on Australian datasets, many sources provide global coverage, enhancing the broader relevance of this work. We then review globally applicable blending techniques, highlighting methods such as weighted averaging, multifractal blending, data assimilation, and machine learning. Key challenges, including latency, quality control, spatial heterogeneity, and validation, are discussed alongside opportunities for advancing multi-source precipitation blending. Finally, we recommend specific datasets and blending methods for trial and assessment, considering both current Australian data availability and potential future upgrades. The insights provided aim to support the development of robust high-resolution precipitation products for hydrological applications in Australia and other regions with similar data integration challenges. This article is categorized under:

Bureau of Meteorology high-resolution Atmospheric Regional Reanalysis for Australia version 2 (BARRA2) is a new regional reanalysis, nested in ERA5 (ECMWF (European Centre for Medium-Range Weather Forecasts) Reanalysis ver. 5), extending from 1979 to near present. It is developed for the Australasia domain, including Australia, New Zealand and parts of the Maritime Continent, at a horizontal grid resolution of 12 km, with a finer 4.4-km grid over Australia. Building on its predecessor (BARRA1), BARRA2 introduces significant improvements by incorporating a broader range of observations and previously unavailable pre-processed data sets. BARRA2 employs a four-dimension-variational assimilation method for the atmosphere and an extended Kalman filter for the land surface, integrating conventional observations and satellite-based radiances, atmospheric motion vector winds, satellite and ground-based Global Navigation Satellite System measurements, and satellite soil moisture data. The modelling system is based on the UK Met Office Unified Model (UM) and the Joint UK Land Simulator (JULES), with updated physics configurations that address limitations of BARRA1. This also includes a convection-permitting configuration for the 4.4-km model. This paper describes the BARRA2 system and assesses the quality of its deterministic outputs for key near-surface meteorological parameters, including temperature, wind and precipitation. The added value of BARRA2 over global reanalyses is evident in coastal and high-terrain regions and within the convection-permitting system. BARRA2 shows quality changes from approximately the year 2000, although these do not appear to negatively affect long-term temperature or rainfall trends at studied sites and regions. Remaining challenges, including modelling biases and data assimilation limitations, are discussed to inform future development.

Operational hydrology forecasts provide crucial information to manage the water resources and offer early warnings to prepare for extreme events. Hydrological modelling and forecasting are challenging particularly in Australia due to its high hydro-climatic variability, numerous intermittent or ephemeral rivers as well as flat terrain. This study has two primary objectives: (i) to implement the Catchment-based Macro-scale Floodplain (CaMa-Flood) model to simulate river hydrodynamics across Australia; (ii) to evaluate the performance of land surface models coupled with CaMa-Flood in simulating streamflow for an operational forecasting service. For this purpose, CaMa-Flood is coupled with standalone JULES, the landscape water balance model (AWRA-L) and two reanalysis datasets (BARRA-R2 and ERA5-Land). Analyses against 452 topographically and hydro-climatically diverse catchments indicate very good results of offline JULES and AWRA-L at both daily and monthly timescales. However, offline JULES tends to overestimate runoff in central Australia, while AWRA-L overestimates runoff in Southeast Australia and Eastern Tasmania. BARRA-R2 and ERA5-Land show large underestimations across the country, with all models having their lowest performances in ephemeral catchments. Differences in runoff generation processes and forcings can be attributed to the performance differences in the models. A sensitivity analysis of CaMa-Flood topographic parameters indicates that the default configuration generally produces reliable simulations; however, further improvements maybe achieved in some locations through fine-tuning relevant parameters.

Atmospheric reanalyses are a popular source of wind speed data for energy modelling but are known to exhibit biases. Such biases can have a significant impact on the validity of techno-economic energy assessments that include simulated wind power. This study assesses the Australian BARRA-R2 (Bureau of Meteorology Atmospheric Regional Reanalysis for Australia, version 2) atmospheric reanalysis, and compares it with MERRA-2 (Modern-Era Retrospective analysis for Research and Applications, V2) and ERA5 (European Centre for Medium-Range Weather Forecasts Reanalysis, fifth generation). Simulated wind power is compared with observed power from 54 wind farms across Australia using site-specific wind turbine specifications. We find that all of the reanalyses replicate wind speed patterns associated with the passage of weather systems. However, modelled power can diverge significantly from observed power at times. Assessed by bias, correlation and error, BARRA-R2 gave the best results, followed by MERRA-2, then ERA5. Annual bias can be readily corrected by wind speed scaling; however, linear scaling will not narrow the error distribution, or reduce the associated error in the frequency distribution of wind power. At the level of a wind farm, site-specific factors and microscale wind behaviour are contributing to differences between simulated and observed power. Although the performance of all the reanalyses is good at times, variability is high and site-dependent. We recommend the use of confidence intervals that reflect the degree of uncertainty in wind power simulation, and the degree of confidence required in the energy system model.

Low pressure systems are associated with a number of climate hazards in Australia, including heavy rainfall, strong winds and coastal erosion. Here, we use a new ensemble of 40 CMIP6 (Sixth Coupled Model Intercomparison Project)-based regional model projections to assess future changes in low pressure systems across Australia, with a focus on vertically developed (deep) cyclones that extend between the surface and 500 hPa. Results show robust future declines in extratropical lows in southern Australia throughout the year, with large uncertainty for lows in northern Australia. Projections for strong, rapidly intensifying and slow-moving low pressure systems are also assessed, and are all projected to decline in frequency. The strongest declines in lows are identified for models that also have larger increases in the intensity of 500-hPa zonal winds to the south of Australia (40-50 degrees S), with observed trends in both indices at the high end of the model range. This suggests the potential for constraining future projections of Australian low pressure systems based on monthly mean zonal winds.

By using finer resolution modelling and locally representative model physics, regional climate models (RCMs) have the potential to improve the information provided by global climate models (GCMs). However, RCMs have their own biases and limitations due to remaining unresolved processes. It is therefore necessary to carefully assess RCM outputs through added value analyses. An ensemble of CMIP6-based 12–17-km regional climate projections has been produced for the Australian Climate Service (ACS) based on the Bureau of Meteorology’s regional climate modelling system (BARPA) and CSIRO’s Conformal Cubic Atmospheric Model (CCAM). The historical and potential future added value of this ensemble is assessed, focusing on extremes (cold, hot, wet and dry). Despite variations in added value across different GCM–RCM experiments, quantities, seasons and regions, BARPA and CCAM generally improve on their driving models for the historical period. Added value over ERA5 is generally small, and often negative for wet and dry extremes, especially for CCAM. The most consistent improvements in all GCM–RCM pairs are found for quantities containing daily minimum temperature, whereas hot days above 40°C show the least improvements. CNRM-ESM2-1-CCAM appears to have significant issues in most analysed quantities, especially related to maximum temperature and might not be recommended for downscaling or use by the community. Additionally, RCMs often predict different climate change signals than their driving models, for example the Murray Basin, which combined with the historical added value indicate plausible improvements in future climate projections.

Extreme rainfall driven by tropical cyclones (TCs) has profound effects on Australian coastlines at both local and regional scales. Here, we develop methods for comparing TC-driven widespread and localised rainfall on three broad coastal regions of tropical Australia (west, north and east). Trends, average recurrence intervals (ARIs) and the fractional contribution of TC rainfall are explored in three historical datasets: Australian Gridded Climate Data (AGCD), ECMWF Reanalysis (ver. 5, ERA5) and the Bureau of Meteorology Atmospheric high resolution Regional Reanalysis for Australia (ver. 1, BARRA1). Results for trends and ARIs between the different datasets are generally inconsistent and also differ between regions, partially owing to the short-term temporal records of some of the data as well as inconsistencies in extreme values between datasets. By contrast, there is a general agreement between all datasets on the fractional contribution of TC rainfall, signalling an increase in recent years. This result is considered together with the trend towards fewer TCs occurring in this region over recent decades, indicating a trend towards increased rainfall intensity per TC on average, assuming steady landfall rates. The methods developed here can be applied easily to other data types such as regional climate model experiments, facilitating a multiple lines of evidence approach that incorporates both observational-based and model-based data. This research is intended to help provide new methods and guidance for identifying trends in TC-driven extreme rainfall, relevant for enhanced planning and adaptation to the impacts of these extreme weather systems.

Anthropogenic climate change is changing the Earth system processes that control the characteristics of natural hazards both globally and across Australia. Model projections of hazards under future climate change are necessary for effective adaptation. This paper presents BARPA-R (the Bureau of Meteorology Atmospheric Regional Projections for Australia), a regional climate model designed to downscale climate projections over the Australasian region with the purpose of investigating future hazards. BARPA-R, a limited-area model, has a 17 km horizontal grid spacing and makes use of the Met Office Unified Model (MetUM) atmospheric model and the Joint UK Land Environment Simulator (JULES) land surface model. To establish credibility and in compliance with the Coordinated Regional Climate Downscaling Experiment (CORDEX) experiment design, the BARPA-R framework has been used to downscale ERA5 reanalysis. Here, an assessment of this evaluation experiment is provided. Performance-based evaluation results are benchmarked against ERA5, with comparable performance between the free-running BARPA-R simulations and observationally constrained reanalysis interpreted as a good result. First, an examination of BARPA-R's representation of Australia's surface air temperature, precipitation, and 10 m winds finds good performance overall, with biases including a 1 ∘C cold bias in daily maximum temperatures, reduced diurnal temperature range, and wet biases up to 25 mm per month in inland Australia. Recent trends in daily maximum temperatures are consistent with observational products, while trends in minimum temperatures show overestimated warming and trends in precipitation show underestimated wetting in northern Australia. Precipitation and temperature teleconnections are effectively represented in BARPA-R when present in the driving boundary conditions, while 10 m winds are improved over ERA5 in six out of eight of the Australian regions considered. Secondly, the paper considers the representation of large-scale atmospheric circulation features and weather systems. While generally well represented, convection-related features such as tropical cyclones, the South Pacific Convergence Zone (SPCZ), the Northwest Cloudband, and the monsoon westerlies show more divergence from observations and internal interannual variability than mid-latitude phenomena such as the westerly jets and extratropical cyclones. Having simulated a realistic Australasian climate, the BARPA-R framework will be used to downscale two climate change scenarios from seven CMIP6 global climate models (GCMs).