Explore the vast range of titles available.

Geographic object-based image analysis (GEOBIA) is a remote sensing image analysis paradigm that defines and examines image-objects: groups of neighboring pixels that represent real-world geographic objects. Recent reviews have examined methodological considerations and highlighted how GEOBIA improves upon the 30+ year pixel-based approach, particularly for H-resolution imagery. However, the literature also exposes an opportunity to improve guidance on the application of GEOBIA for novice practitioners. In this paper, we describe the theoretical foundations of GEOBIA and provide a comprehensive overview of the methodological workflow, including: (i) software-specific approaches (open-source and commercial); (ii) best practices informed by research; and (iii) the current status of methodological research. Building on this foundation, we then review recent research on the convergence of GEOBIA with deep convolutional neural networks, which we suggest is a new form of GEOBIA. Specifically, we discuss general integrative approaches and offer recommendations for future research. Overall, this paper describes the past, present, and anticipated future of GEOBIA in a novice-accessible format, while providing innovation and depth to experienced practitioners.

In regular convolutional neural networks (CNN), fully-connected layers act as classifiers to estimate the probabilities for each instance in classification tasks. The accuracy of CNNs can be improved by replacing fully connected layers with gradient boosting algorithms. In this regard, this study investigates three robust classifiers, namely XGBoost, LightGBM, and Catboost, in combination with a CNN for a land cover study in Hanoi, Vietnam. The experiments were implemented using SPOT7 imagery through (1) image segmentation and extraction of features, including spectral information and spatial metrics, (2) normalization of attribute values and generation of graphs, and (3) using graphs as the input dataset to the investigated models for classifying six land cover classes, namely House, Bare land, Vegetation, Water, Impervious Surface, and Shadow. The results show that CNN-based XGBoost (Overall accuracy = 0.8905), LightGBM (0.8956), and CatBoost (0.8956) outperform the other methods used for comparison. It can be seen that the combination of object-based image analysis and CNN-based gradient boosting algorithms significantly improves classification accuracies and can be considered as alternative methods for land cover analysis.

Wildlife management is based on various measurements representative of the health of populations and their habitats. Some agencies are focusing on animal surveys to manage species such as white-tailed deer (Odocoileus virginianus). Current survey methods are faced with the challenge of reduced operating costs as well as estimating and correcting detection biases. Our pilot study (data collected on 6 Nov 2012 at Saint-David-de-Falardeau, QC, Canada) assessed the potential of a new approach detect and count deer based on visible and thermal infrared image processing at very-high spatial resolutions using an unmanned aerial system (UAS). Supervised and unsupervised pixel-based image classification approaches as well as object-based image analysis (OBIA) were assessed for different spatial resolutions and with different combinations of spectral bands. None of the pixel-based approaches were effective for detecting deer. The OBIA approach detected deer with a rate of up to 100% under the best conditions by using a combination of visible and thermal infrared imagery at a spatial resolution of 0.8 cm/pixel. Overall, this approach had an average detection rate of 0.5, which is comparable to conventional aerial surveys. Visual obstructions by coniferous canopy and the spectral confusion associated with certain elements (e.g., bare soil, rocks) are problems that remain unresolved. Using UASs with image processing for surveys of deer and other species of large mammals is promising, but currently limited by the flight range of unmanned aerial vehicles and the associated regulations.

Floodplain wetlands play a crucial role in the global carbon cycle, yet the spatiotemporal variation of carbon flux in floodplain wetlands remains poorly understood due to complex hydrological processes. Here, we present an integrated framework to upscale net ecosystem CO2 exchange (NEE) in floodplain wetlands by combining eddy covariance measurements, object‐based image analysis (OBIA), hydrodynamic models, and integrated machine learning (ML) techniques. The proposed framework was instantiated in the Poyang Lake floodplain wetland, China. Results showed that OBIA could effectively capture the heterogeneous surface of wetlands through multi‐scale segmentation, thereby providing suitable space units for NEE upscaling. Hydrological regimes obtained by the hydrodynamic model, that is, maximum flood level (MFL), mean water level (WL) and water level fluctuation (WLF), were the key drivers to the NEE variability. The integrated ML model could effectively upscale NEE with higher R2 and superior robustness than commonly used individual ML model, benefiting from an optimized integration of outputs from multiple ML model through Powell optimization. The spatiotemporal distribution of the upscaled NEE results indicated that permanently inundated areas in floodplain wetlands mostly functioned as carbon sources or weak carbon sinks, while littoral zones functioned as carbon sinks. Hydrological regime changes lead to a shift between carbon sources and sinks in floodplain wetlands. The proposed framework can provide a feasible way to analyze the spatiotemporal changes of NEE and is of great benefit to achieving carbon sequestration in floodplain wetlands.

Tanjung Kelayang is vulnerable to ecosystem change driven by human activities and environmental factors. This study aimed to map and analyze changes in benthic habitats using Sentinel-2A imagery from 2019 and 2024 by applying an Object-Based Image Analysis (OBIA) approach combined with a Support Vector Machine (SVM) classifier. The analysis included water column correction using the depth-invariant index (DII), multiresolution segmentation, habitat classification, accuracy assessment, and post-classification change detection. Six benthic habitat classes were identified, namely coral, seagrass, macroalgae, sand, rocks, and coral rubble. The results show a decrease in coral cover of 87.99 ha (19.96%) and a substantial increase in coral rubble of 80.33 ha (87.77%), indicating reef degradation. The classification achieved an Overall Accuracy of 70.97% with a kappa coefficient of 0.61, reflecting a good classification performance. These findings demonstrate that Sentinel-2A imagery combined with the OBIA method is effective for monitoring changes in shallow marine habitats, and supports spatial-based coastal zone management.

Questions How to map floristic variation in a patterned fen in an ecologically meaningfully way? Can plant communities be delineated with species data generalized into plant functional types? What are the benefits and drawbacks of the two selected remote‐sensing approaches in mapping vegetation patterns, namely: (a) regression models of floristically defined fuzzy plant community clusters and (b) classification of predefined habitat types that combine vegetation and land cover information? Location Treeless 0.4 km2 mesotrophic string–flark fen in Kaamanen, northern Finland. Methods We delineated plant community clusters with fuzzy c‐means clustering based on two different inventories of plant species and functional type distribution. We used multiple optical remote‐sensing data sets, digital elevation models and vegetation height models derived from drone, aerial and satellite platforms from ultra‐high to very high spatial resolution (0.05–3 m) in an object‐based approach. We mapped spatial patterns for fuzzy and crisp plant community clusters using boosted regression trees, and fuzzy and crisp habitat types using supervised random forest classification. Results Clusters delineated with species‐specific data or plant functional type data produced comparable results. However, species‐specific data for graminoids and mosses improved the accuracy of clustering in the case of flarks and string margins. Mapping accuracy was higher for habitat types (overall accuracy 0.72) than for fuzzy plant community clusters (R2 values between 0.27 and 0.67). Conclusions For ecologically meaningful mapping of a patterned fen vegetation, plant functional types provide enough information. However, if the aim is to capture floristic variation in vegetation as realistically as possible, species‐specific data should be used. Maps of plant community clusters and habitat types complement each other. While fuzzy plant communities appear to be floristically most accurate, crisp habitat types are easiest to interpret and apply to different landscape and biogeochemical cycle analyses and modeling. We tested if plant communities can be delineated using plant functional types instead of species‐specific data. We found that the approaches produce comparable results. We compared two remote‐sensing approaches in mapping vegetation patterns. Regression models of floristically defined plant communities reveal the fuzziness of vegetation. Classification of pre‐defined habitat types is easier to interpret and has higher mapping accuracy.

In Flanders (northern Belgium), wild boar (Sus scrofa) returned in 2006 after 50 years of absence and the population is increasing, both in abundance and geographic extent. In the absence of wild boar, Flanders’ landscape structure changed into a dense, mosaic-like pattern of agricultural, natural, and urban areas. The return of the wild boar increasingly leads to human–wildlife conflicts, mainly linked to damage in agriculture. Hence, there is a growing need for a time-efficient, standardized, and accurate method to assess crop damage. We present an Unmanned Aerial Vehicle-based method, using Geographic Object-Based Image Analysis and Random Forests to estimate the damaged area and associated yield losses, between 2015 and 2017, due to wild boar in individual fields in Flanders. Our approach resulted in an 84.50% overall accuracy in calculating damaged area for maize fields and 94.40% for grasslands. Damage levels ranged between 14.3% and 20.2% in maize fields and 16.5% to 25.4% in grasslands. Our method can provide objective base data for compensation schemes and guide management strategies based on damage assessments.

Remote sensing (RS) platforms such as unmanned aerial vehicles (UAVs) represent an essential source of information in precision agriculture (PA) as they are able to provide images on a daily basis and at a very high resolution. In this framework, this study aims to identify the optimal level of nitrogen (N)-based nutrients for improved productivity in an onion field of “Cipolla Rossa di Tropea” (Tropea red onion). Following an experiment that involved the arrangement of nine plots in the onion field in a randomized complete block design (RCBD), with three replications, three different levels of N fertilization were compared: N150 (150 kg N ha−1), N180 (180 kg N ha−1), and e N210 (210 kg N ha−1). The crop cycle was monitored using multispectral (MS) UAV imagery, producing vigor maps and taking into account the yield of data. The soil-adjusted vegetation index (SAVI) was used to monitor the vigor of the crop. In addition, the coverage’s class onion was spatially identified using geographical object-based image classification (GEOBIA), observing differences in SAVI values obtained in plots subjected to differentiated N fertilizer treatment. The information retrieved from the analysis of soil properties (electrical conductivity, ammonium and nitrate nitrogen), yield performance and mean SAVI index data from each field plot showed significant relationships between the different indicators investigated. A higher onion yield was evident in plot N180, in which SAVI values were higher based on the production data.

The current paper explores the suitability of geomorphons for the automatic extraction of drumlins. To calibrate the geomorphons to the size of drumlins, it is necessary to optimally define the maximum scale of mapping, i.e., the lookup distance parameter (L). Therefore, based on the concept of topographic grain, we introduce a new automated approach for identifying the specific threshold of L (13 cells) and assessing its potential to generate consistent and accurate results in drumlin extraction. Following an object-based image analysis (OBIA) routine, a new method for mapping and detecting drumlins is proposed. The aggregated geomorphons map was employed both as a thematic layer for image segmentation and as a first criterion for the identification of drumlin candidates. The classification results were quantitatively compared with the reference data in order to evaluate the performance of the drumlin classification, by using five additional L values (3, 50, 100, 200, 400 cells). The evaluation revealed that the highest drumlin detection rate of 91.7% was reached at an L value of 13 cells (65 m), while the lowest value of 84.3% was reached at the default value (L-3 cells). We conclude that the use of the automated procedure for the detection of the L value is useful in achieving a rapid computation of geomorphons, which leads to consistent and accurate results in drumlin extraction. A comparison with previous OBIA methods suggests that the proposed approach produced the most accurate extraction of drumlins.

Fatal landslides occur every year during the rainy season (June–November) in the Lower Mekong Region (LMR). There is an urgent need to develop a landslide early warning system in the LMR. In collaboration with the Asian Disasters Preparedness Center and NASA’s SERVIR Programme, we are regionalizing the global Landslide Hazard Assessment System for Situational Awareness model for the LMR (LHASA-Mekong). A robust set of landslide inventories are needed to effectively train the machine learning-based LHASA-Mekong model. In this study, the Semi-Automatic Landslide Detection (SALaD) system was modified by incorporating a change detection module (SALaD-CD) to produce rainfall event-based landslide inventories using pre- and post-imagery from RapidEye and PlanetScope for various locations in the LMR that were identified based on media and government reports. These rainfall-induced landslides are published as initiation points for ease of use. In total, we created 22 inventories: 2 in Laos, 4 in Myanmar, 1 in Thailand and 15 in Vietnam. These inventories are being used to train the LHASA-Mekong model and quantify the effects of Land use/Land cover change on landslide susceptibility. These open data will be a valuable resource for advancing landslide studies in this region.