Object-Based Classification Approaches for Multitemporal Identification and Monitoring of Pastures in Agroforestry Regions using Multispectral Unmanned Aerial Vehicle Products

Abstract

Sown Biodiverse Pastures (SBP) are the basis of a high-yield grazing system tailored for Mediterranean ecosystems and widely implemented in Southern Portugal. The application of precision farming methods in SBP requires cost-effective monitoring using remote sensing (RS). The main hurdle for the remote monitoring of SBP is the fact that the bulk of the pastures are installed in open Montado agroforestry systems. Sparsely distributed trees cast shadows that hinder the identification of the underlaying pasture using Unmanned Aerial Vehicles (UAV) imagery. Image acquisition in the Spring is made difficult by the presence of flowers that mislead the classification algorithms. Here, we tested multiple procedures for the geographical, object-based image classification (GEOBIA) of SBP, aiming to reduce the effects of tree shadows and flowers in open Montado systems. We used remotely sensed data acquired between November 2017 and May 2018 in three Portuguese farms. We used three machine learning supervised classification algorithms: Random Forests (RF), Support Vector Machine (SVM) and Artificial Neural Networks (ANN). We classified SBP based on: (1) a single-period image for the maximum Normalized Difference Vegetation Index (NDVI) epoch in each of the three farms, and (2) multi-temporal image stacking. RF, SVM and ANN were trained using some visible (red, green and blue bands) and near-infrared (NIR) reflectance bands, plus NDVI and a Digital Surface Model (DSM). We obtained high overall accuracy and kappa index (higher than 79% and 0.60, respectively). The RF algorithm had the highest overall accuracy (more than 92%) for all farms. Multitemporal image classification increased the accuracy of the algorithms. as it helped to correctly identify as SBP the areas covered by tree shadows and flower patches, which would be misclassified using single image classification. This study thus established the first workflow for SBP monitoring based on remotely sensed data, suggesting an operational approach for SBP identification. The workflow can be applied to other types of pastures in agroforestry regions to reduce the effects of shadows and flowering in classification problems.

Spatiotemporal Patterns of Pasture Quality Based on NDVI Time-Series in Mediterranean Montado Ecosystem

“…The most extreme case occurred in the experimental field "GRO" between March and April 2019, with about a month and a half without cloud-free information. These results highlight the interest in complementary monitoring systems, for example, by drones [10,24]. Another constraint for RS (by satellite or drone) of permanent grasslands in open woodlands such as Dehesa (in Spain) and Montado (in Portugal) is the presence of scattered trees [4].…”

Sentinel-2 Satellite Imagery for Agronomic and Quality Variability Assessment of Pistachio (Pistacia vera L.)

“…. ) obtained from satellite images, with a spatial resolution ranging from 10 m (Sentinel-2) to 30 m (Landsat) or obtained from UAV images whose resolution is at centimetre level (depending on the characteristics of the multispectral camera and the flight height), have been used in the decision-making process in many crops such as wheat, rapeseed, cotton, corn and woody species [16][17][18][19][20]. Specifically, Sentinel satellites can be very useful and they have been used to predict almond yields [21], to discriminate the quality of the walnut based on the vigour of the trees [22], to assess the bloom dynamics of almond orchards [23] and to classify vineyards according to their vigour [24,25].…”

Comparison and assessment of different object-based classifications using machine learning algorithms and UAVs multispectral imagery: a case study in a citrus orchard and an onion crop

“…Since the VHR DSM, derived from the photogrammetric process, is efficient for plant height detection (De Castro et al, 2018;Zisi et al, 2018), the use of it as an additional input layer increased the accuracy in Orchard-study site, characterized by higher variability in the vegetation height (trees and herbaceous), as already demonstrated in other works (De Luca et al, 2019;Vilar et al, 2020;Zisi et al, 2018). This was fundamental since the three vegetation LC classes' spectral signature was practically identical in the Orchard-study site.…”

“…Moreover, differently from previous research works dealing with the comparison of GEOBIA approaches (De Luca et al, 2019;M. Li et al, 2016b;Qian et al, 2015;Teodoro & Araujo, 2016;Vilar et al, 2020), we implemented the segmentation step using three different algorithms. The Large Scale Mean-Shift (LSMS) Figure 1.…”