No abstract

Seasonal pasture monitoring can increase the efficiency of pasture utilization in livestock grazing enterprises. However, manual monitoring of pasture over large areas is often infeasible due to time and financial constraints. Here, we monitor changes in botanical composition in Tasmania, Australia, through application of supervised learning using satellite imagery (Sentinel-2). In the field, we measured ground cover and botanical composition over a 12-month period to develop a supervised classification approach used to identify pasture classes. Across seasons and paddocks, the approach predicted pasture classes with 75–81 % accuracy. Botanical composition varied seasonally in response to biophysical factors (primarily climate) and grazing behaviour, with seasonal highs in spring and troughs in autumn. Overall, we demonstrated that 10-m multispectral imagery can be reliably used to distinguish between pasture species as well as seasonal changes in botanical composition. Our results suggest that farmers and land managers should aim to quantify within-paddock variability rather than paddock average cover, because the extent and duration of very low ground cover puts the paddock/field at risk of adverse grazing outcomes, such as soil erosion and loss of pasture biomass, soil carbon and biodiversity. Our results indicate that satellite imagery can be used to support grazing management decisions for the benefit of pasture production and the improvement of environmental sustainability.

Crop field boundaries aid in mapping crop types, predicting yields, and delivering field-scale analytics to farmers. Recent years have seen the successful application of deep learning to delineating field boundaries in industrial agricultural systems, but field boundary datasets remain missing in smallholder systems due to (1) small fields that require high resolution satellite imagery to delineate and (2) a lack of ground labels for model training and validation. In this work, we use newly-accessible high-resolution satellite imagery and combine transfer learning with weak supervision to address these challenges in India. Our best model uses 1.5 m resolution Airbus SPOT imagery as input, pre-trains a state-of-the-art neural network on France field boundaries, and fine-tunes on India labels to achieve a median Intersection over Union (mIoU) of 0.85 in India. When we decouple field delineation from cropland classification, a model trained in France and applied as-is to India Airbus SPOT imagery delineates fields with a mIoU of 0.74. If using 4.8 m resolution PlanetScope imagery instead, high average performance (mIoU>0.8) is only achievable for fields larger than 1 hectare. Experiments also show that pre-training in France reduces the number of India field labels needed to achieve a given performance level by as much as 10× when datasets are small. These findings suggest our method is a scalable approach for delineating crop fields in regions of the world that currently lack field boundary datasets. We publicly release 10,000 Indian field boundary labels and our delineation model to facilitate the creation of field boundary maps and new methods by the community.

Many private and public actors are incentivized by the promises of big data technologies: digital tools underpinned by capabilities like artificial intelligence and machine learning. While many shared value propositions exist regarding what these technologies afford, public-facing concerns related to individual privacy, algorithm fairness, and the access to insights requires attention if the widespread use and subsequent value of these technologies are to be fully realized. Drawing from perspectives of data science, social science and technology acceptance, we present an interdisciplinary analysis that links these concerns with traditional research and development (R&D) activities. We suggest a reframing of the public R&D ‘brand’ that responds to legitimate concerns related to data collection, development, and the implementation of big data technologies. We offer as a case study Australian agriculture, which is currently undergoing such digitalization, and where concerns have been raised by landholders and the research community. With seemingly limitless possibilities, an updated account of responsible R&D in an increasingly digitalized world may accelerate the ways in which we might realize the benefits of big data and mitigate harmful social and environmental costs.

No abstract