This paper presents a detailed land cover mapping study for the state of Victoria in Australia conducted through a machine learning approach (random forest algorithm) using Sentinel-2 imagery. The study uses a hierarchical classification, based on the FAO’s Land Cover Classification Scheme. This paper highlights the importance of spatial random sampling in assessing ground data, and details methods for land cover mapping, remote sensing analysis, calibration as well as validation. The land cover mapping procedure involves the use of a fine-tuned random forest classifier, and an overlaying mask generation technique to improve classification accuracy. The resulting 2021/22 land cover map is accessible through the Victorian Land Use Information System (VLUIS) and has undergone rigorous technical validation with an overall accuracy of 86%. This data set is publicly accessible and regularly released to provide valuable information for a variety of applications, including agricultural policy development, strategic planning, climate change modelling and environmental monitoring.

Soil moisture impacts exchanges of water, energy and carbon fluxes between the land surface and the atmosphere. Passive microwave remote sensing at L-band can capture spatial and temporal patterns of soil moisture in the landscape. Both ESA and NASA have launched L-band radiometers, in the form of the SMOS and SMAP satellites respectively, to monitor soil moisture globally, every 3-day at about 40 km resolution. However, their coarse scale restricts the range of applications. While SMAP included an L-band radar to downscale the radiometer soil moisture to 9 km, the radar failed after 3 months and this initial approach is not applicable to developing a consistent long term soil moisture product across the two missions anymore. Existing optical-, radiometer-, and oversampling-based downscaling methods could be an alternative to the radar-based approach for delivering such data. Nevertheless, retrieval of a consistent high resolution soil moisture product remains a challenge, and there has been no comprehensive intercomparison of the alternate approaches. This research undertakes an assessment of the different downscaling approaches using the SMAPEx-4 field campaign data.

Temperate native grasslands in southeastern Australia have been extensively cleared for agriculture, and the remaining patches are under growing pressure from further land use change, climate variability, and invasive species. Mapping and monitoring their distribution and the cover of native and exotic grasses are critical for their conservation and management. Field-based methods are not always scalable or time-effective, and this study aimed to develop a scalable method to map and monitor the fractional cover-class maps of native C3 and native C4 grass cover as a component of remnant native grasslands on the western outskirts of Melbourne, Victoria, Australia. Field-based reference data for training and validation of random forest machine learning models were collected across multiple sites in 2021. Sentinel-2 optical spectral bands and vegetation indices were used as the primary input data, and Sentinel-1 Synthetic Aperture Radar (SAR)-derived Grey Level Co-occurrence Matrix (GLCM) texture metrics were assessed for their capacity to improve the model. Results show that random forest models trained on Sentinel-2 data without GLCM texture information derived from Sentinel-1 SAR data provided a moderate overall accuracy (C3: 59.1%, C4: 78.1%). Class-specific metrics showed that reliability was highest for better represented lower-cover classes, particularly the 6–25% native C3 class and the 0–5% native C4 class, while higher-cover classes were less reliable because of the limited number of training and validation samples. Grass cover fractions were modelled well for sparse to moderate grass cover, but dense grass cover was not modelled accurately, probably due to limited high-cover samples in the training dataset. Model performance was not improved by the inclusion of Sentinel-1 SAR-derived GLCM texture metrics, indicating that C-band VH-polarised SAR is not sensitive to the fine-scale structural heterogeneity that characterises native grassland ecosystems. Sparse native C3 and C4 grasses could be mapped most reliably in the lower-cover classes as a component of grasslands with optical remote sensing, and the method developed here can now be applied to enable evidence-based management of grasslands, biodiversity conservation and the monitoring of grassland composition in the WGR and elsewhere. Higher-resolution structural datasets and more sophisticated machine learning approaches may be required to accurately predict native C3 and C4 grass cover fractions in denser grasslands.