Improvement of pasture biomass modelling using high-resolution satellite imagery and machine learning.

Robust quantification of vegetative biomass using satellite imagery using one or more forms of machine learning (ML) has hitherto been hindered by the extent and quality of training data. Here, we showcase how ML predictive demonstrably improves when additional training data is used. We collated field datasets of pasture biomass obtained via destructive sampling, 'C-Dax' reflective measurements and rising plate meters (RPM) from ten livestock farms across four States in Australia. Remotely sensed data from the Sentinel-2 constellation was used to retrieve aboveground biomass using a novel machine learning paradigm hereafter termed "SPECTRA-FOR" (Spectral Pasture Estimation using Combined Techniques of Random-forest Algorithm for Features Optimisation and Retrieval). Using this framework, we show that the low temporal resolution of Sentinel-2 in high latitude regions with persistent cloud cover leads to extensive gaps between cloud-free images, hindering model performance and, thus, contemporaneous ability to forecast real-time pasture biomass. By leveraging the spectral consistency between Sentinel-2 and Planet Lab SuperDove to overcome this limitation, we used ten spectral bands of Sentinel-2, four bands of Sentinel-2 as a proxy for pre-2022 SuperDove (referred to as synthetic SuperDove or SSD), and the actual SuperDove (ASD), given that SuperDove imagery has a higher resolution and more frequent passage compared with Sentinel-2. Using their respective bands as input features to SPECRA-FOR, model performance for the ten bands of Sentinel-2 were R2 = 0.87, root mean squared error (RMSE) of 439 kg DM/ha and mean absolute error (MAE) of 255 kg DM/ha, while that for SSD increased to an R2 of 0.92, RMSE of 346 kg DM/ha and MAE = 208 kg DM/ha. The study revealed the importance of robust data mining, imagery harmonisation and model validation for accurate real-time modelling of pasture biomass with ML.

Green space in health research: an overview of common indicators of greenness.

Human environments influence human health in both positive and negative ways. Green space is considered an environmental exposure that confers benefits to human health and has attracted a high level of interest from researchers, policy makers, and increasingly clinicians. Green space has been associated with a range of health benefits, such as improvements in physical, mental, and social wellbeing. There are different sources, metrics and indicators of green space used in research, all of which measure different aspects of the environment. It is important that readers of green space research understand the terminology used in this field, and what the green space indicators used in the studies represent in the real world. This paper provides an overview of the major definitions of green space and the indicators used to assess exposure for health practitioners, public health researchers, and health policy experts who may be interested in understanding this field more clearly, either in the provision of public health-promoting services or to undertake research.

The association of fractional cover, foliage projective cover and biodiversity with birthweight.

INTRODUCTION: Environmental exposures can contribute both benefits and risks to human health. Maternal exposure to green space has been associated with improvements in birthweight, among other birth outcomes. Newer measures of green space have been developed, which allows for an exploration of the effect of different ground covers (green, dry and bare earth), as well as measures of biodiversity. This study explores the association of these novel green space measures with birthweight in a large birth cohort in Queensland, Australia. METHODS: Birthweight was acquired from the routine health records. Records were allocated green space values for fractional cover, biodiversity and foliage projective cover. Directed acyclic graphs were developed to guide variable selection. Mixed-effects linear regression and generalised linear mixed-effects models were developed, with random intercepts for maternal residential locality and year of birth. Results are presented as standardised beta coefficients or odds ratios, with 95% confidence intervals. RESULTS: An IQR increase of green cover (29.6 g, 95% CI 13.8-45.5) and foliage projective cover (26.0 g, 95% CI 10.8-41.3) are associated with birthweight in urban areas. An IQR increase in dry cover -34.4 g, 95% CI -60.4 to -8.4) and bare earth (-17.7 g, 95% CI -32.8 to -2.6) are associated with lower birthweight. Mothers living in rural areas had similar results, with an IQR increase in green cover (17.8 g, 95% CI 2.9-32.7) associated with higher birthweight, and bare earth (-27.7 g, 95% CI -45.7 to -9.7) was associated with lower birthweight. The biodiversity measure used in this study was not associated with any birthweight outcomes. CONCLUSION: This study finds that the types of ground cover within the maternal residential locality are associated with small, but significant, changes in estimated birthweight, and these effects are not limited to urban areas.