Abstract:Abstract Grasslands cover approximately 40.5% of the surface of the earth and 80% of agriculturally productive land. After forests, grasslands are the primary carbon sink source and the most used feed source for livestock production. Regularly monitoring grasslands assures efficient management and sustainability of pasture-based production systems. Conventional ground-based methods to monitor grassland production and management rely on field measurements, which are time-consuming and usually restricted to small-scale assessment. Using satellite information allows for large-scale monitoring of grasslands and capturing the spatial variability of the land surface with high temporal resolution. Various methods for grassland monitoring based on satellite data can be applied, such as classifications, correlations/regression analyses, and time series analyses. Depending on the purpose of the application, these methods are sometimes combined to derive grassland management and production information. The ability of satellite-based data to quantify vegetation characteristics depends on the type of sensor and instrumentation features, such as spectral, radiometric, spatial, and temporal resolution, polarization, and angularity. The models to estimate grassland biomass based on remote sensing have been chiefly focused on optical systems. The spectral reflectance of raw bands and vegetation indices were used as proxies to investigate spatial and temporal patterns of grassland production. Optical (multispectral or hyperspectral) sensors are passive and require sunlight, so they depend highly on the weather (cloud) and light conditions. Thus, there has been increasing interest in active sensors, such as Synthetic Aperture Radar (SAR) and Light Detection and Ranging (LIDAR) sensors, which are not constrained due to clouds but are more complex. Using satellite data in combination with field measurements has commonly yielded regression models (e.g., linear, power, logarithmic, multiple linear) for estimating grassland biomass or biophysical characteristics (e.g., chlorophyll, leaf area index) of different types of grasslands. The exponential evolution of digital computers has pushed forward machine learning-based regression methods to estimate biomass. Random forest, support vector machines, and artificial neural networks are the most used algorithms. The possibility of accurate mapping and monitoring of biomass and nutritional attributes of grasslands based on satellite provides essential insights into the decision support system for pasture management. A better understanding of the nitrogen status of pastures, forage biomass, and its nutritive value is instrumental in livestock and forage management. Timely prediction of these variables can help improve decision-making by grazing land managers on, for instance, the adjustment of stocking rate or adequate supplementation to match the needs of animals toward more sustainable production. Future use of satellite-based grazing models in tandem with ruminant nutrition models will enable to development of decision-support tools to assist with many aspects of livestock production in diverse environmental conditions and accounting for temporal variability. (FAPESP #2020/14367-7)

Top-down Green-ups: Satellite Sensing and Deep Models to Predict Buffelgrass Phenology

Phenology forecasting models for detection and management of invasive annual grasses

Predicting fine-scale taxonomic variation in landscape vegetation using large satellite imagery data sets

Multi-modal learning for geospatial vegetation forecasting

Deep Learning Approaches for Wildland Fires Using Satellite Remote Sensing Data: Detection, Mapping, and Prediction

Analyzing Multispectral Satellite Imagery of South American Wildfires Using Deep Learning

Meta-Prediction of Bromus Tectorum Invasion in Central Utah, United States

Deep learning detects invasive plant species across complex landscapes using Worldview‐2 and Planetscope satellite imagery

Enhanced prediction of vegetation responses to extreme drought using deep learning and Earth observation data

Remote Sensing-Based Time-Series Analysis of Cheatgrass (bromus Tectorum L.) Phenology.

Modelling biomass and biodiversity in temperate grasslands with Sentinel-1, Sentinel-2 and Unmanned Aerial Vehicles.

WeedScout: Real-Time Autonomous blackgrass Classification and Mapping using dedicated hardware

Predicting Ground Cover with Deep Learning Models—An Application of Spatio-Temporal Prediction Methods to Satellite-Derived Ground Cover Maps in the Great Barrier Reef Catchments

Utilizing unsupervised learning to improve sward content prediction and herbage mass estimation

A Deep Learning Framework: Predicting Fire Radiative Power From the Combination of Polar-Orbiting and Geostationary Satellite Data During Wildfire Spread

Nationwide operational mapping of grassland first mowing dates combining machine learning and Sentinel-2 time series

2 Satellite-Based Decision Support Tools to Assist Grazing Cattle Production

A Machine Learning Approach to Forecasting Remotely Sensed Vegetation Health

Deep learning surrogate models of JULES-INFERNO for wildfire prediction on a global scale

High-Throughput Plot-Level Quantitative Phenotyping Using Convolutional Neural Networks on Very High-Resolution Satellite Images

Upscaling dryland carbon and water fluxes with artificial neural networks of optical, thermal, and microwave satellite remote sensing