To monitor forest degradation, deforestation, reforestation, and above-ground biomass, individually delineated tree crowns are key satellite-derived products. Unsupervised individual tree segmentation (ITS) algorithms generate crown delineations from Light Detection And Ranging (LiDAR)-derived point clouds or rasters representing canopies. Supervised deep learning models trained with these delineations can then be used to segment billions of crowns using high-resolution imagery at continental scales. However, diverse vegetation types, variable LiDAR point cloud densities, and algorithm misparameterization introduce errors, undermining the generalizability of supervised models trained on these outputs. The aim of this study is to identify the most suitable CHM generation and unsupervised canopy segmentation approach across diverse canopy height distributions and point cloud densities. To this end, we evaluate detection and delineation performances of (a) three common canopy height models (i.e., (i) point-to-raster; (ii) triangulated irregular network; and (iii) pit-free), (b) four ITS algorithms (i.e., (i) Watershed; (ii) DalPonte; (iii) Li; and (iv) Silva), across (c) 15 vegetation types using (d) three point-cloud density classes, comparing accuracy obtained using default versus heuristically optimised parameters. We found that the DalPonte algorithm achieved the highest delineation accuracy in 84% of cases, while the Watershed algorithm, despite notable optimisation gains and its most frequent use in literature, ranked as the second least accurate method. Vegetation type substantially influenced accuracies, with semi-arid rangelands and woodlands performing best and dense, multi-layered forests the worst. High-density point clouds provided the best results, but heuristically optimised methods performed well with low-density datasets (<3 pt/m(2)). Optimizations generally improved detection and delineation accuracies, with +51% match ratio, -25% oversegmentation, and negligible increases in undersegmented crowns. The largest gains occurred in low-density point clouds, with +92% detection accuracy and four times more accurate delineations. Finally, this research identified the most accurate optimised algorithms for specific vegetation types and point cloud densities, providing a foundation for largescale crown delineations to train highly generalizable deep learning models.

Live Fuel Moisture Content (LFMC) is a key determinant of vegetation flammability and fire behaviour, yet LFMC products have traditionally relied on coarse-resolution sensors such as the Moderate Resolution Imaging Spectroradiometer (MODIS, 500 m), limiting their utility for fine-scale fire management. This study introduces the first continental-scale operational LFMC product for Australia derived from Sentinel-2 imagery at 20 m resolution. We developed a Random Forest regression model trained on approximately 680,000 paired Sentinel-2 reflectance and MODIS-LFMC samples (2015–2022) to emulate outputs from the Australian Flammability Monitoring System (AFMS), a MODIS-based pre-operational LFMC product. Model evaluation against AFMS showed strong agreement for grasslands (R2 = 0.83, RMSE = 32.45%) and moderate performance for forests (R2 = 0.43, RMSE = 20.84%) and shrublands (R2 = 0.21, RMSE = 10.28%). Validation using 2279 in situ LFMC measurements from Globe-LFMC 2.0 indicated improved accuracy at homogeneous sites (NDVI CV ≤ 20th percentile: R2 = 0.42, RMSE = 31.39%). Additionally, when validating with a dedicated field campaign specifically designed for Sentinel-2 LFMC assessment, the model achieved its highest accuracy (R2 = 0.53, RMSE = 32.14%), highlighting the importance of tailored ground protocols for satellite product validation. Predicted LFMC also reproduced observed seasonal dynamics at sites with frequent field monitoring. Despite variability across vegetation types, the Sentinel-2 LFMC product effectively captured spatial patterns and seasonal dynamics, providing a step change in monitoring vegetation moisture at landscape scales. This high-resolution dataset offers actionable intelligence for prescribed burning, fuel treatment planning, and fire behaviour modelling in fire-prone environments.

We discuss the development of high-resolution (1 km) estimates of terrestrial carbon and water fluxes over the Australian continent by empirical upscaling of a regional network of flux tower measurements (“AusEFlux” v1.1 https://zenodo.org/records/7947265). We detail our ensemble learning approach for estimating the per-pixel epistemic uncertainty in flux predictions. Our investigations demonstrate that regional or continental upscaling has several advantages over global upscaling, including: the ability to use regionally derived covariable datasets tailored to the regional environmental context; reduced computational constraints allowing for higher-resolution predictions, thus reducing the impacts of sub-cell landscape heterogeneity; ameliorating spatial biases present in global datasets that often have a strong northern hemisphere bias; simpler interpretation of results due the reduced requirement to generalise across vastly different climates, ecosystem types, and plant functional traits; and increased relevance to local stakeholders. We compare AusEFlux with estimates from nine other products that cover the three broad categories that define current methods for estimating the terrestrial carbon cycle. We argue that consiliences between datasets derived using different methodologies offer alternative value for assessing the quality of an upscaling product than any given cross-validation technique, especially where training datasets have spatial or temporal biases that are difficult to mitigate. Lastly, we discuss the benefits of regularly updating our upscaling product to arrive at a systematic monitoring of terrestrial carbon and water fluxes.

World-wide water resources are threatened by the impacts of natural climate variability and anthropogenic climate change resulting in water stress for many regions. Here, we focus on the Murray-Darling River Basin, Australia, one of the many regions that benefits from a better understanding of water resources availability and their response to climate change and water extraction from surface water and groundwater. This knowledge can help secure a sustainable water management for the future. Particularly, we introduce a novel satellite-based approach to determine the relative contributions of natural climate variability and human-induced impacts on the regional water balance. We found that the contribution ratio of water extraction for irrigation explains 17% of the terrestrial water storage changes that are observed by the GRACE satellite mission and its Follow-On mission since 2003. Water is primarily extracted from surface water (84%) with the remainder (16%) taken from groundwater. Introducing GRACE observations into the W3RA water balance model - which does not simulate the human-induced impact on water resources - via a data assimilation approach improved the representation of water storage variability and intensified trends in drying and wetting periods. We conclude that data assimilation can fundamentally improve our understanding of water resources and how they are impacted by natural and human-induced impacts of climate change. Our results also offer potential for technical improvements of hydrological models and for future policy implementation. The presented study contributes to achieve the Sustainable Development Goals (SDGs), in particular no. 13 (combat climate change and its impact) and no. 6 (availability and sustainable management of water).

Data assimilation (DA) of time-variable satellite gravity observations, such as those from the Gravity Recovery and Climate Experiment (GRACE), GRACE Follow-On (GRACE-FO), and future gravity missions, can be used to constrain simulations of the vertical sum of water storage in Global Hydrological Models (GHMs). However, current DA implementations of these Terrestrial Water Storage (TWS) changes are often performed at regional scales or, if applied globally, at low spatial resolutions. This limitation is primarily due to the high computational demands of DA and numerical challenges, such as instabilities in covariance matrix inversion. To fully exploit the potential of satellite gravity observations and the high spatial resolution of GHMs, we developed PyGLDA, an open-source Python-based system that enables fine-scale and computationally efficient global DA. The key innovations of PyGLDA include (1) a global patch-wise DA approach using domain localization and neighboring-weighted global aggregation and (2) seamless compatibility between basin-scale and grid-scale DA implementations. PyGLDA represents a significant functional improvement over previous DA systems, offering wide-ranging and flexible options for user-specific applications. The modular structure of the system allows users to customize water storage compartments, modify observation representations, and potentially select different GHMs. This paper provides a comprehensive description of PyGLDA and its application in a case study of the Danube River Basin, along with a demonstration of global DA, where experiments involve integrating monthly GRACE TWS fields (2002–2010) with the daily W3RA water balance model at 0.1° spatial resolution.

Land-surface phenology is critical to understanding Earth system responses to environmental change. However, there is a lack of studies that specifically examine Australian phenology trends over time periods long enough to robustly capture the effects of a changing climate. Here we utilise and demonstrate the methodological superiority of circular statistics for quantifying phenology in Australia. Next, we employ circular statistical methods across a long-term harmonised NDVI dataset (1982-2022) to analyse phenological trends across Australia's diverse landscapes. We find that forest ecosystems exhibit inertia to long-term shifts in rainfall regimes and increasing vapour pressure deficits, exhibiting stable growing season length, but increased maximum seasonal productivity (0.012 NDVI/decade). In contrast, shrublands and grasslands show significant phenological shifts, including earlier green-ups (-4.3 and-2.0 days/decade, respectively), earlier senescence (-2.5 and-1.7 days/ decade), and earlier peaks (-2.5 and-3.1 days/decade) linked to altered rainfall regimes and land use changes. Only modest increases in the length of season are observed because the start and end of seasons often advance simultaneously. Importantly, major cropping regions are experiencing shortened growing seasons (-3.5 days/ decade), offset by increased maximum NDVI, stabilising productivity but raising concerns for future agricultural productivity. Increases in maximum NDVI are driving an amplification of Australia's vegetation cycles, with concomitant increases in rates of growth and senescence.

Abstract Global terrestrial water storage anomaly (TWSA) products from the Gravity Recovery and Climate Experiment (GRACE) and its Follow‐On mission (GRACE/FO) have an approximately three‐month latency, significantly limiting their operational use in water management and drought monitoring. To address this challenge, we develop a Bayesian convolutional neural network (BCNN) to predict TWSA fields with uncertainty estimates during the latency period. The results demonstrate that BCNN provides near‐real‐time TWSA estimates that closely match GRACE/FO observations, with median correlation coefficients of 0.92–0.95, Nash‐Sutcliffe efficiencies of 0.81–0.89, and root mean squared errors of 1.79–2.26 cm for one‐ to three‐month ahead predictions. More importantly, the model advances global hydrological drought monitoring by enabling detection up to three months before GRACE/FO data availability, with median characterization mismatches below 16.4%. This breakthrough in early warning capability addresses a fundamental constraint in satellite‐based hydrological monitoring and offers water resource managers critical lead time to implement drought mitigation strategies.

Understanding water availability and its response to climate change and water extraction is crucial for sustainable water management in Australia's Murray-Darling Basin. This study introduces a space-based method that quantifies the natural and human-induced impact on changes in terrestrial water storage. It reveals an impact of 17% due to water extraction for irrigation over the past two decades, with 84% of this extraction coming from surface water and 16% from groundwater. The human-induced impact varies spatially with higher values in the southern Murray (up to 5.6%) and smaller values in the northern Darling (down to 0.2%). Data-model fusion of the satellite-based water storage changes into a hydrological model, which does not simulate water extraction, man-made reservoirs and wetlands, improved the representation of water storage variability and intensified trends in drying and wetting periods. This study adds valuable findings to better understand natural and human-induced impacts on the regional water resources under changing climate and to better represent these impacts (80% and 20% respectively) within hydrological models after data-model fusion.

We introduce Saudi Rainfall (SaRa), a gridded historical and near-real-time precipitation (P) product specifically designed for the Arabian Peninsula, one of the most arid, water-stressed, and data-sparse regions on Earth. The product has an hourly 0.1° resolution spanning 1979 to the present and is continuously updated with a latency of less than 2 h. The algorithm underpinning the product involves 18 machine learning model stacks trained for different combinations of satellite and (re)analysis P products along with several static predictors. As a training target, hourly and daily P observations from gauges in Saudi Arabia (n = 113) and globally (n = 14 256) are used. To evaluate the performance of SaRa, we carried out the most comprehensive evaluation of gridded P products in the region to date, using observations from independent gauges (randomly excluded from training) in Saudi Arabia as a reference (n = 119). Among the 20 evaluated P products, our new product, SaRa, consistently ranked first across all evaluation metrics, including the Kling–Gupta efficiency (KGE), correlation, bias, peak bias, wet-day bias, and critical success index. Notably, SaRa achieved a median KGE – a summary statistic combining correlation, bias, and variability – of 0.36, while widely used non-gauge-based products such as CHIRP, ERA5, GSMaP V8, and IMERG-L V07 achieved values of −0.07, 0.21, −0.13, and −0.39, respectively. SaRa also outperformed four gauge-based products such as CHIRPS V2, CPC Unified, IMERG-F V07, and MSWEP V2.8 which had median KGE values of 0.17, −0.03, 0.29, and 0.20, respectively. Our new P product – available at https://www.gloh2o.org/sara (last access: 24 September 2025) – addresses a crucial need in the Arabian Peninsula, providing a robust and reliable dataset to support hydrological modeling, water resource assessments, flood management, and climate research.

Accurate and comparable annual mapping is critical to understanding changing vegetation distribution and informing land use planning and management. A U-Net convolutional neural network (CNN) model was used to map natural vegetation and forest types based on annual Landsat geomedian reflectance composite images for a 500 km × 500 km study area in southeastern Australia. The CNN was developed using 2018 imagery. Label data were a ten-class natural vegetation and forest classification (i.e., Acacia, Callitris, Casuarina, Eucalyptus, Grassland, Mangrove, Melaleuca, Plantation, Rainforest and Non-Forest) derived by combining current best-available regional-scale maps of Australian forest types, natural vegetation and land use. The best CNN generated using six Landsat geomedian bands as input produced better results than a pixel-based random forest algorithm, with higher overall accuracy (OA) and weighted mean F1 score for all vegetation classes (93 vs. 87% in both cases) and a higher Kappa score (86 vs. 74%). The trained CNN was used to generate annual vegetation maps for 2000–2019 and evaluated for an independent test area of 100 km × 100 km using statistics describing accuracy regarding the label data and temporal stability. Seventy-six percent of pixels did not change over the 20 years (2000–2019), and year-on-year results were highly correlated (94–97% OA). The accuracy of the CNN model was further verified for the study area using 3456 independent vegetation survey plots where the species of interest had ≥ 50% crown cover. The CNN showed an 81% OA compared with the plot data. The model accuracy was also higher than the label data (76%), which suggests that imperfect training data may not be a major obstacle to CNN-based mapping. Applying the CNN to other regions would help to test the spatial transferability of these techniques and whether they can support the automated production of accurate and comparable annual maps of natural vegetation and forest types required for national reporting.

Global Hydrological and Land Surface Models (GHM/LSMs) embody numerous interacting predictors and equations, complicating the diagnosis of primary hydrological relationships. We propose a model diagnostic approach based on Random Forest feature importance to detect the input variables that most influence simulated hydrological processes. We analyzed the JULES, ORCHIDEE, HTESSEL, SURFEX and PCR-GLOBWB models for the relative importance of precipitation, climate, soil, land cover and topographic slope as predictors of simulated average evaporation, runoff, and surface and subsurface runoffs. The machine learning model could reproduce GHM/LSMs outputs with a coefficient of determination over 0.85 in all cases and often considerably better. The GHM/LSMs agreed precipitation, climate and land cover share equal importance for evaporation prediction, and mean precipitation is the most important predictor of runoff. However, the GHM/LSMs disagreed on which features determine surface and subsurface runoff processes, especially with regards to the relative importance of soil texture and topographic slope.

Measuring the spatiotemporal dynamics of lake and reservoir water storage is fundamental for assessing the influence of climate variability and anthropogenic activities on water quantity and quality. Previous studies estimated relative water volume changes for lakes where both satellite-derived extent and radar altimetry data are available. This approach is limited to only a few hundred lakes worldwide and cannot estimate absolute (i.e. total volume) water storage. We increased the number of measured lakes by a factor of 300 by using high-resolution Landsat and Sentinel-2 optical remote sensing and ICESat-2 laser altimetry, in addition to radar altimetry from the Topex/Poseidon; Jason-1, Jason-2 and Jason-3; and Sentinel-3 and Sentinel-6 instruments. Historical time series (1984–2020) of water storage could be derived for more than 170 000 lakes globally with a surface area of at least 1 km2, representing 99 % of the total volume of all water stored in lakes and reservoirs globally. Specifically, absolute lake volumes are estimated based on topographic characteristics and lake properties that can be observed by remote sensing. In addition to that, we also generated relative lake volume changes solely based on satellite-derived heights and extents if both were available. Within this dataset, we investigated how many lakes can be measured in near real time (2020–current) in basins worldwide. We developed an automated workflow for near-real-time global lake monitoring of more than 27 000 lakes. The GloLakes historical and near-real-time lake storage dynamics data from 1984 to current are publicly available through https://doi.org/10.25914/K8ZF-6G46 (Hou et al., 2022c) and a web-based data explorer (http://www.globalwater.online, last access: 12 December 2023).

The 2015–2016 Amazon drought was characterized by below-average regional precipitation for an entire year, which distinguishes it from the dry-season-only droughts in 2005 and 2010. Studies of vegetation indices (VIs) derived from optical remote sensing over the Amazonian forests indicated three stages in canopy response during the 2015–2016 drought, with below-average greenness during the onset and end of the drought, and above-average greenness during the intervening months. To date, a satisfactory explanation for this broad temporal pattern has not been found. A better understanding of rainforest behaviors during this unusually long drought should help predict their response to future droughts. We hypothesized that negative VI anomalies could be caused by water and heat stress exceeding the tolerance ranges of the rainforest. To test our hypothesis, based on monthly observations of terrestrial water storage (TWS), land surface temperature (LST), and vapor pressure deficit (VPD) for January 2003 to December 2016, we proposed an approach to categorize regions into two groups: (1) those exceeding normal hydrological and thermal ranges and (2) those within normal ranges. Accordingly, regions exceeding normal ranges during different stages of the 2015–2016 event were delineated. The results showed a gradual southward shift in these regions: from the northeastern Amazon during August to October 2015 to the north–central part during November 2015 to February 2016 and finally to the southern Amazon in July 2016. Over these regions exceeding normal ranges during droughts, negative VI anomalies were expected, irrespective of radiation anomalies. Over the regions within normal ranges, VI anomalies were assumed to respond positively to radiation anomalies, as is expected under normal conditions. We found that our proposed approach can explain more than 70 % of the observed spatiotemporal patterns in VI anomalies during the 2015–2016 drought. These results suggest that our “exceeding normal ranges”-based approach combining (i) water storage, (ii) temperature, and (iii) atmospheric moisture demand drivers can reasonably identify the most likely drought-affected regions at monthly to seasonal timescales. Using observation-based hydrological and thermal condition thresholds can help with interpreting the response of the Amazon rainforest to future drought events.

Abstract. Long-term, reliable datasets of satellite-based vegetation condition are essential for understanding terrestrial ecosystem responses to global environmental change, particularly in Australia which is characterised by diverse ecosystems and strong interannual climate variability. We comprehensively evaluate several existing global AVHRR NDVI products for their suitability for long-term vegetation monitoring in Australia. Comparisons with MODIS NDVI highlight significant deficiencies, particularly over densely vegetated regions. Moreover, all the assessed products failed to adequately reproduce inter-annual variability in the pre-MODIS era as indicated by Landsat NDVI anomalies. To address these limitations, we propose a new approach to calibrating and harmonising NOAA’s Climate Data Record AVHRR NDVI to MODIS MCD43A4 NDVI for Australia using a gradient-boosting decision tree ensemble method. Two versions of the datasets are developed, one incorporating climate data in the predictors (‘AusENDVI-clim’: Australian Empirical NDVI-climate) and another independent of climate data (‘AusENDVI-noclim’). These datasets, spanning 1982–2013 at a spatial resolution of 0.05°, exhibit strong correlation and low relative errors compared to MODIS NDVI, accurately reproducing seasonal cycles over densely vegetated regions. Furthermore, they closely replicate the interannual variability in vegetation condition in the pre-MODIS era. A reliable method for gap-filling the AusENDVI record is also developed that leverages climate, atmospheric CO2 concentration, and woody cover fraction predictors. The resulting synthetic NDVI dataset shows excellent agreement with observations. Finally, we provide a complete 41-year dataset where gap filled AusENDVI from January 1982 to February 2000 is seamlessly joined with MODIS NDVI from March 2000 to December 2022. Analysing 40-year per-pixel trends in Australia’s annual maximum NDVI revealed increasing values across most of the continent. Moreover, shifts in the timing of annual peak NDVI are identified, underscoring the dataset's potential to address crucial questions regarding changing vegetation phenology and its drivers. The AusENDVI dataset can be used for studying Australia's changing vegetation dynamics and downstream impacts on terrestrial carbon and water cycles, and provides a reliable foundation for further research into the drivers of vegetation change. AusENDVI is open access and available at https://doi.org/10.5281/zenodo.10802704 (Burton, 2024).

Long-term, reliable datasets of satellite-based vegetation condition are essential for understanding terrestrial ecosystem responses to global environmental change, particularly in Australia, which is characterised by diverse ecosystems and strong interannual climate variability. We comprehensively evaluate several existing global Advanced Very High Resolution Radiometer (AVHRR) normalised-difference vegetation index (NDVI) products for their suitability for long-term vegetation monitoring in Australia. Comparisons with the MODIS NDVI highlight significant deficiencies, particularly over densely vegetated regions. Moreover, all the assessed products failed to adequately reproduce the interannual variability in the pre-MODIS era as indicated by Landsat NDVI anomalies. To address these limitations, we propose a new approach to calibrating and harmonising NOAA's Climate Data Record of AVHRR NDVI to the MODIS MCD43A4 NDVI for Australia using a gradient-boosting decision tree ensemble method. Two versions of the datasets are developed, one incorporating climate data in the predictors ("AusENDVI-clim": Australian Empirical NDVI-climate) and another that is independent of climate data ("AusENDVI-noclim"). These datasets, spanning 1982-2013 at a spatial resolution of 0.05 degrees and with a monthly time step, exhibit strong correlations (r(2)=0.89-0.94) and low mean errors compared with MODIS MCD43A4 NDVI (mean absolute error (MAE) = 0.014-0.028, RMSE = 0.021-0.046), accurately reproducing seasonal cycles over densely vegetated regions. Furthermore, they closely replicate the interannual variability in vegetation condition in the pre-MODIS era. A reliable method for gap-filling the AusENDVI record is also developed that leverages climate, atmospheric CO2 concentration, and woody-cover fraction predictors. The resulting synthetic NDVI dataset shows excellent agreement with the MODIS MCD43A4 NDVI and the recalibrated AVHRR NDVI time series (r(2)=0.82-0.95, MAE = 0.016-0.029, RMSE = 0.039-0.041). Finally, we provide a complete 41-year dataset where the gap-filled AusENDVI-clim from January 1982 to February 2000 is joined with the MODIS MCD43A4 NDVI from March 2000 to December 2022. Analysing 40-year per-pixel trends in Australia's annual maximum NDVI revealed increasing values, and shifts in the timing, of the annual peak NDVI across most of the continent, underscoring the dataset's potential to address crucial questions regarding the changing vegetation phenology and its drivers. The AusENDVI dataset can be used for studying Australia's changing vegetation dynamics and downstream impacts on the terrestrial carbon and water cycles, and it provides a reliable foundation for further research into the drivers of vegetation change. AusENDVI is open access and available at 10.5281/zenodo.10802703 (Burton et al., 2024).

Biodiversity conservation is a global imperative, and many national and international actions aim to halt and reverse biodiversity decline. This study tests how ecosystem accounting, as standardised in the System of Environmental-Economic Accounting, could be used to integrate ecological and economic (e.g. environmental protection expenditure) data needed for biodiversity conservation using the example of the critically endangered Box-gum grassy woodlands (dominated by Eucalyptus spp.) in Australia. Despite data limitations, ecosystem extent accounts were produced and indicated that the likely extent of Box-gum grassy woodlands increased by 54,000 ha or 1.5 % over 16 years to 3.5 million ha in 2017. However, the possible extent decreased by more than one million ha, reflecting uncertainty in the estimate and the limitations of remotely-sensed information for estimating ecosystem extent. There was large regional variation in the changes, and the reasons for change, which are allocated to anthropogenic and natural changes in accounting, were not identified owing to a lack of information. As such, we could not determine if illegal activities or the ecosystem's recovery plan had any effect. This study showed that ecosystem accounting has the potential to assist biodiversity conservation by providing information that could support adaptive management by, for example, evaluating the causes of changes in the extent of threatened ecosystems and applying this learning to guide the allocation of resources to more effective conversation interventions. To achieve this potential, better data and regular and standardised reporting are needed, which will require cooperation between all agencies involved in biodiversity conservation.

The McArthur grassland and forest fire danger indices, widely used in Australia, predict six fire danger classes from ‘Low-Moderate’ to ‘Catastrophic.’ These classes were linked to the rate of fire spread and difficulty of suppression. However, the lack of rate of fire spread data, especially for elevated fire danger classes, has hindered improvement of the McArthur methodology or an alternate approach. We explored the relationship between fire danger classes and burned areas (derived from the Moderate Resolution Imaging Spectroradiometer (MODIS) satellite instrument) within six climate zones during the 2000–2016 Australian fire seasons. A negative binomial linear regression model was used to explore this relationship. The fire danger classes demonstrated a corresponding increase in burned area from ‘Low-Moderate’ to ‘Very High’ classes in Australia’s inland regions. The elevated fire danger classes did not contribute to this trend. In coastal regions, the satellite-derived burned area showed no relationship between fire danger classes and satellite-derived burned area. We used accumulated burned area from the daily MODIS product, which could be subjected to lagged detection as observed in the Kilmore East fire. Thus, the satellite-derived total burned area may not be a suitable metric for informing the McArthur fire danger classes across Australia.

Evaluation of fire severity reduction strategies requires the quantification of intervention outcomes and, more broadly, the extent to which fuel characteristics affect fire severity. However, investigations are currently limited by the availability of accurate data on fire severity predictors, particularly relating to fuel. Here, we used airborne LiDAR data collected before the 2019-20 Australian Black Summer fires to investigate the contribution of fuel structure to fire severity under a range of weather conditions. Fire severity was estimated using the Relative Burn Ratio calculated from Sentinel-2 optical remote sensing imagery. We modelled the effects of various fuel structure estimates and other environmental predictors using Random Forest models. In addition to variables estimated at each observation point, we investigated the influence of surrounding landscape characteristics using an innovative method to estimate fireline progression direction. Our models explained 63-76% of fire severity variance using parsimonious predictor sets. Fuel cover in the understorey and canopy, and vertical vegetation heterogeneity, were positively associated with fire severity. Up-fire burnt area and recent planned and unplanned fire reduced fire severity, whereby unplanned fire provided a longer-lasting reduction of fire severity (up to 15 years) than planned fire (up to 10 years). Although fuel structure and land management effects were important predictors, weather and canopy height effects were dominant. By mapping continuous interactions between weather and fuel-related variables, we found strong evidence of diminishing fuel effects below 20-40% relative air humidity. While our findings suggest that land management interventions can provide meaningful fire severity reduction, they also highlight the risk of warmer and drier future climates constraining these advantages.

The System of Environmental Economic Accounting (SEEA) has the potential to support decision making but there are few cases of their application to public policy or planning. In this paper we review five government policies related to land use planning and environmental management in the Australian Capital Territory (ACT), assess the data available to inform these policies and link the polices to the accounts that could assist with their implementation, evaluation and update. We find that a range of SEEA-based accounts are potentially useful for the plans and strategies reviewed. Data available for accounts are of varying quality, scattered and are not integrated nor regularly summarised. The available data were to produce accounts for ecosystem extent and land cover, and these provide a platform for continuing engagement with the ACT government and for developing accounts for ecosystem services, ecosystem condition and environmental protection expenditure that could inform ACT policy for land use planning, management of ecosystems, protected areas, and urban forests. Six steps for developing ecosystem accounts to support policy are recommended to accounts producers and three priority actions for SEEA-based accounting to support environmental management in the ACT are identified. Preparing the accounts and linking them directly to existing plans and strategies is a first step to acceptance and use of SEEA in the ACT and elsewhere.

We develop high-resolution (1 km) estimates of gross primary productivity (GPP), ecosystem respiration (ER), and net ecosystem exchange (NEE) over the Australian continent for the period January 2003 to June 2022 by empirical upscaling of flux tower measurements. We compare our estimates with nine other products that cover the three broad categories that define current methods for estimating the terrestrial carbon cycle and assess if consiliences between datasets can point to the correct dynamics of Australia's carbon cycle. Our results indicate that regional empirical upscaling greatly improves upon the existing global empirical upscaling efforts, outperforms process-based models, and agrees much better with the dynamics of CO2 flux over Australia as estimated by two regional atmospheric inversions. Our nearly 20-year estimates of terrestrial carbon fluxes revealed that Australia is a strong net carbon sink of −0.44 PgC yr−1 (interquartile range, IQR = 0.42 PgC yr−1) on average, with an inter-annual variability of 0.18 PgC yr−1 and an average seasonal amplitude of 0.85 PgC yr−1. Annual mean carbon uptake estimated from other methods ranged considerably, while carbon flux anomalies showed much better agreement between methods. NEE anomalies were predominately driven by cumulative rainfall deficits and surpluses, resulting in larger anomalous responses from GPP than ER. In contrast, we show that the long-term average seasonal cycle is dictated more by the variability in ER than GPP, resulting in peak carbon uptake typically occurring during the cooler, drier austral autumn and winter months. This new estimate of Australia's terrestrial carbon cycle provides a benchmark for assessment against land surface model simulations and a means for monitoring of Australia's terrestrial carbon cycle at an unprecedented high resolution. We call this new estimate of Australia's terrestrial carbon cycle “AusEFlux” (Australian Empirical Fluxes).