We apply the Support Vector Regression (SVR) machine learning model to estimate surface roughness on a large alluvial fan of the Kosi River in the Himalayan Foreland from satellite images. To train the model, we used input features such as radar backscatter values in Vertical–Vertical (VV) and Vertical–Horizontal (VH) polarisation, incidence angle from Sentinel-1, Normalised Difference Vegetation Index (NDVI) from Sentinel-2, and surface elevation from Shuttle Radar Topographic Mission (SRTM). We generated additional features (VH/VV and VH–VV) through a linear data fusion of the existing features. For the training and validation of our model, we conducted a field campaign during 11–20 December 2019. We measured surface roughness at 78 different locations over the entire fan surface using an in-house-developed mechanical pin-profiler. We used the regression tree ensemble approach to assess the relative importance of individual input feature to predict the surface soil roughness from SVR model. We eliminated the irrelevant input features using an iterative backward elimination approach. We then performed feature sensitivity to evaluate the riskiness of the selected features. Finally, we applied the dimension reduction and scaling to minimise the data redundancy and bring them to a similar level. Based on these, we proposed five SVR methods (PCA-NS-SVR, PCA-CM-SVR, PCA-ZM-SVR, PCA-MM-SVR, and PCA-S-SVR). We trained and evaluated the performance of all variants of SVR with a 60:40 ratio using the input features and the in-situ surface roughness. We compared the performance of SVR models with six different benchmark machine learning models (i.e., Gaussian Process Regression (GPR), Generalised Regression Neural Network (GRNN), Binary Decision Tree (BDT), Bragging Ensemble Learning, Boosting Ensemble Learning, and Automated Machine Learning (AutoML)). We observed that the PCA-MM-SVR perform better with a coefficient of correlation (R = 0.74), Root Mean Square Error (RMSE = 0.16 cm), and Mean Square Error (MSE = 0.025 ${\mathrm{cm}}^2$). To ensure a fair selection of the machine learning model, we evaluated the Akaike’s Information Criterion (AIC), corrected AIC (AICc), and Bayesian Information Criterion (BIC). We observed that SVR exhibits the lowest values of AIC, corrected AIC, and BIC of all the other methods; this indicates the best goodness-of-fit. Eventually, we also compared the result of PCA-MM-SVR with the surface roughness estimated from different empirical and semi-empirical radar backscatter models. The accuracy of the PCA-MM-SVR model is better than the backscatter models. This study provides a robust approach to measure surface roughness at high spatial and temporal resolutions solely from the satellite data.

In early 2025 mainland Australia experienced one of the most extreme inland flooding events in the observational period. Extensive floodwaters gradually flowed southward toward Kati Thanda–Lake Eyre (KT-LE) via the ‘Channel Country’, with the Cooper Creek basin being one of the most severely affected regions. Typically considered as a low-gradient dryland catchment, the Cooper recorded its highest-ever flood with water depths reaching 13.4 meters in the more confined locations and with a flood width up to 55 km; surpassing the previous historical 1974 record flood. The flood caused severe impacts on local communities and infrastructure, damaging roads, agriculture, and resulting in major livestock losses. Despite the flood’s severity, monitoring was limited to just four active gauging stations with none on the lowermost 350 km. We developed a novel approach for assessing the magnitude of this event by combining SWOT Pixel Cloud data product, optical satellite imagery and a Lidar-derived digital elevation model (DEM) to assess both flood depth and volume. SWOT-derived water depth hydrographs were validated against water level data at four gauging stations, showing excellent agreement with errors as small as ± 11 cm (RMSE) and ± 8.1 cm (MAE). By coincidence, the flood’s peak at two gauge locations (Cullyamurra waterhole and Nappa Merrie) coincided with the SWOT data period of acquisition, providing a valuable flood snapshot. Backscattering analysis of the peak flood (mean σ0 = 10–24 dB) reveals that SWOT effectively captures the complex backscatter response of dryland floodwaters. Furthermore, comparison with the JRC Global Flood Model further revealed substantial underestimation of flood volume (23–48%) in existing global datasets across Australia’s Channel Country. Overall, this study demonstrates the capability of SWOT to capture and monitor extreme flooding in low-gradient dryland rivers, providing new opportunities for flood assessment and management in data-sparse regions.

Australia's drylands, covering nearly 70% of the continent exhibit the most variable precipitation and streamflow regimes globally. The endorheic Lake Eyre Basin (LEB) terminates at Kati Thanda-Lake Eyre (KT–LE), Australia’s largest lake and drains 1.14 M km2. This basin experiences remarkable ecological fluctuations with spectacular boom and bust cycles during extreme flooding events. This vast unregulated river basin, despite its ecological significance, has limited stream gauges and no lake monitoring, making the lake's water balance a real challenge due to its vast size, remote location and complex lake geometry. Recent observations reveal significant water loss in endorheic basins worldwide, emphasizing the urgency for improved freshwater monitoring solutions for KT – LE and its basin. Therefore, in this study, we present a space-based monitoring solution to estimate the storage volume of the KT–LE as an alternative to in situ measurements. To do so, we utilized the data from the Surface Water and Ocean Topography (SWOT) satellite, launched in December 2022, to monitor the 2024 KT-LE filling event. The duration of this event was between March and October 2024. The predicted maximum lake storage volume (recorded on 1st May) reached 0.82 Km3 with a predicted average depth of -14.2 metres AHD (Australian Height Datum). We cross-compared the volume estimates from three bathymetry digital elevation models to evaluate the derived estimates in the absence of in situ data. We achieved the accuracy of the derived water surface elevation estimates with a root mean square error (RMSE) of