Abstract:Abstract Remote sensing is widely used for land cover classification applications. This technology supports the development of land management practices and policies to achieve conservation and economic goals in agricultural landscapes. In animal production, remote sensing aids in reducing grassland degradation and enhancing livestock and crop production. The objective of this work is to discuss recent satellite- and drone-based remote sensing applications for grassland mapping in the Canadian Prairies. Using a satellite-based approach, the agreement between remotely sensed land cover maps and non-spatial government records was assessed for public agricultural Crown lands in southern Manitoba. Two different land cover classification products derived from satellite imagery [i.e., the Manitoba Grassland Inventory (MGI) prepared by the Province of Manitoba and the Annual Crop Inventory (ACI) prepared by Agriculture and Agri-Food Canada] were merged into a single mosaic covering the entire Agro-Manitoba region. Non-spatial official records were georeferenced and summarized through parcel delineation using geographic information system (GIS) and R programming tools. Comparison of the two datasets revealed low agreement between related grassland classes (i.e., forests and shrublands; and native and tame grasslands) due to spectral similarities. However, grouping these related vegetation types into broader categories (i.e., forests and shrublands as woody vegetation; native and tame grasslands as grassy vegetation) significantly improved overall agreement, with a difference of less than 3% observed between remotely sensed datasets and official records. The derived grassland classification was then used to assess the grazing status of the agricultural Crown lands by estimating stocking rates in animal unit months (AUM) and utilizing historical yield data available from three separate field surveys conducted between 2004 and 2020 across the region. Average carrying capacities ranged from 0.71 AUM/ha to 1.99 AUM/ha. Overall, the analysis using remotely sensed data indicated that the forage resources in the Crown lands were underutilized by 45%. To further explore the potential of using high-resolution remote sensing to map grasslands, ongoing research involves the use of a drone-mounted hyperspectral sensor (HySpex Mjolnir VS-620) covering a spectral range from 400 to 2,500 nm across 490 spectral channels. This sensor enables grassland mapping at significantly greater spatial and spectral resolutions. Monthly surveys during the snow-free period will be conducted in 12 study sites across the Agro-Manitoba region in 2024 and 2025, with biomass samples collected to model grassland distribution and productivity. The outcomes of this study will offer critical insights into leveraging specific wavelengths for grassland mapping and strategies for extrapolating these insights to larger geographical areas using satellite imagery. Overall, the efforts described above underscore the significance of remote sensing approaches in managing agricultural landscapes and provide valuable insights for land managers, policymakers, and stakeholders to promote sustainable agricultural production in grasslands.

Assessing spatial distribution and quantification of native trees in Saskatchewan's prairie landscape using remote sensing techniques

Mapping Grasslands in Mixed Grassland Ecoregion of Saskatchewan Using Big Remote Sensing Data and Machine Learning

Improved k-NN Mapping of Forest Attributes in Northern Canada Using Spaceborne L-Band SAR, Multispectral and LiDAR Data

A global method to identify trees outside of closed-canopy forests with medium-resolution satellite imagery

Remote sensing approaches to identify trees to species-level in the urban forest: A review

370 Leveraging remote sensing products to estimate forage productivity in the Canadian Prairies

Mapping with height and spectral remote sensing implies that environment and forest structure jointly constrain tree community composition in temperate coniferous forests of eastern Washington, United States

Classification of tree species in a heterogeneous urban environment using object-based ensemble analysis and World View-2 satellite imagery

Synergistic use of Sentinel-1, Sentinel-2, and Landsat 8 in predicting forest variables

Multi-Horizon Predictive Soil Mapping of Historical Soil Properties Using Remote Sensing Imagery

Applied aerial spectroscopy: A case study on remote sensing of an ancient and semi-natural woodland

Mapping Quaking Aspen Using Seasonal Sentinel-1 and Sentinel-2 Composite Imagery across the Southern Rockies, USA

Modelling, mapping and monitoring of forest cover changes, using support vector machine, kernel logistic regression and naive bayes tree models with optical remote sensing data

Comparison of Deep and Machine Learning Approaches for Quebec Tree Species Classification Using a Combination of Multispectral and LiDAR Data

Spectral-temporal traits in Sentinel-1 C-band SAR and Sentinel-2 multispectral remote sensing time series for 61 tree species in Central Europe

Tree species diversity mapping from spaceborne optical images: The effects of spectral and spatial resolution

Reconstruction and analysis of a deciduous sapling using digital photographs or terrestrial-LiDAR technology

Mapping tree species in natural and planted forests using Sentinel-2 images

A new approach for Spatializing the CAnadian National Forest Inventory (SCANFI) using Landsat dense time series

Multi-Temporal LiDAR and Hyperspectral Data Fusion for Classification of Semi-Arid Woody Cover Species

Local-scale mapping of tree species in a lower mountain area using Sentinel-1 and -2 multitemporal images, vegetation indices, and topographic information