Object-based Image Analysis Research Articles

A Drone-Powered Deep Learning Methodology for High Precision Remote Sensing in California’s Coastal Shrubs

Wildland conservation efforts require accurate maps of plant species distribution across large spatial scales. High-resolution species mapping is difficult in diverse, dense plant communities, where extensive ground-based surveys are labor-intensive and risk damaging sensitive flora. High-resolution satellite imagery is available at scales needed for plant community conservation across large areas, but can be cost prohibitive and lack resolution to identify species. Deep learning analysis of drone-based imagery can aid in accurate classification of plant species in these communities across large regions. This study assessed whether drone-based imagery and deep learning modeling approaches could be used to map species in complex chaparral, coastal sage scrub, and oak woodland communities. We tested the effectiveness of random forest, support vector machine, and convolutional neural network (CNN) coupled with object-based image analysis (OBIA) for mapping in diverse shrublands. Our CNN + OBIA approach outperformed random forest and support vector machine methods to accurately identify tree and shrub species, vegetation gaps, and communities, even distinguishing two congeneric shrub species with similar morphological characteristics. Similar accuracies were attained when applied to neighboring sites. This work is key to the accurate species identification and large scale mapping needed for conservation research and monitoring in chaparral and other wildland plant communities. Uncertainty in model application is associated with less common species and intermixed canopies.

Satellite based analysis of mangrove cover and density change in mangroves of Tulang Bawang District, Lampung Province, Indonesia

Abstract. Kaskoyo H, Hartati F, Bakri S, Febryano IG, Dewi BS, Nurcahyani N. 2023. Satellite based analysis of mangrove cover and density change in mangroves of Tulang Bawang District, Lampung Province, Indonesia. Biodiversitas 24: 3019-3028. Being a threatened ecosystem of the world, mangroves require regular monitoring to identify areas that need improvement in conservation measures. Considering this, the present study aimed to assess the mangrove cover and density change in mangrove forest of Tulang Bawang District, Lampung Province, Indonesia. Data was collected by remote sensing techniques using satellite imagery viz., Landsat 5 in 2000 and 2010 and Landsat 8 in 2020. It was then analyzed using remote sensing method, namely Object-Based Image Analysis techniques using the unguided classification method to obtain mangrove cover data; while mangrove density information was obtained using the Normalized Difference Vegetation Index algorithm. The results revealed the decrease in mangrove cover and density between 2000 and 2020. The mangrove cover decreased from 7529 ha in 2000 to 5551 ha in 2020. Furthermore, density of mangrove forest has also witnessed change from dense density class in 2000 to medium density class in 2020. From this result, it is evident that mangroves of East Rawajitu Sub-district are degrading and require focused efforts to prevent the degradation. Communities, entrepreneurs, and governments have different motives for utilizing mangrove forests that result in damage, so reasonable efforts are needed to manage mangrove forests sustainably.

Pruning for image segmentation: Improving computational efficiency for large-scale remote sensing applications

Image segmentation is a fundamental step in object-based image analysis and other workflows. However, high-efficiency remains a challenge, especially for the analysis of large-scale Earth observation images. In recent years, considerable effort has been paid to designing merging criteria, automatic scale selection, and object-specific optimisation. These segmentation methods usually rely on the region-adjacency graph (RAG) model and the nearest neighbour graph (NNG) model, which provide acceptable merging performance. Low efficiency occurs due to many redundant edge weight updates in the RAG model. In this study, we propose a generic dynamic pruning framework to improve the efficiency of existing region-merging-based segmentation algorithms, opening the door for large-scale applications in remote sensing. The proposed pruning framework includes intra-object and inter-object pruning modules for the RAG model. Inter-object pruning divides the RAG model into multiple sub-RAG models to reduce the redundant edge weight updates between adjacent objects. Intra-object pruning iteratively divides the sub-RAG into smaller RAGs. In our experimental analysis, we employ the proposed pruning framework with six region-merging segmentation methods and validate the effectiveness on four 10–20M pixel images and a 100M pixel data set. The pruning framework improves the performance of various segmentation algorithms by reducing computational complexity while maintaining segmentation accuracy. We observed a significant improvement in efficiency, with various achieving super-linear speed-up while maintaining the stability of segmentation accuracy. In single-core mode, the computation time of tested algorithms is enhanced by two to ten times on the four test images. In the multicore mode, speed-up increased up to 40 times with eight CPU cores. The computational cost was reduced by 36.15% to 95.77% in the number of weight updates, which is independent of hardware characteristics. On the large-scale image, two modes achieved speed-ups of 36.07 and 102.74, respectively.

Detection of small-scale landscape elements with remote sensing

Abstract. Landscape elements located on agricultural fields or on their edges play a crucial role in the biodiversity of agricultural land. The landscape elements’ database in Estonia is updated in accordance with the applications of the field owners, and usually it does not represent a real situation of the landscape elements on the field. Hence, the analysis and control over landscape elements are limited. The main aim of this study is to create a methodology to map landscape elements in Estonia with remote sensing data. The first method was created considering the importance of computational efficiency and therefore fast and non-complex map algebra solution was developed. The second, more precise but more computationally expensive way to map landscape elements, was the object-based image analysis method utilizing machine learning classification. Both methods displayed high overall accuracies, but users’ and producers’ accuracies were lower. Taking into account the computational time and accuracy, it was concluded that the map algebra method is better suitable for fast landscape elements’ detection. However, the object-based image analysis method is more suitable for identifying more exact classes of landscape elements.

A Semi-Automated Two-Step Building Stock Monitoring Methodology for Supporting Immediate Solutions in Urban Issues

The Sustainable Development Goals (SDGs) have addressed environmental and social issues in cities, such as insecure land tenure, climate change, and vulnerability to natural disasters. SDGs have motivated authorities to adopt urban land policies that support the quality and safety of urban life. Reliable, accurate, and up-to-date building information should be provided to develop effective land policies to solve the challenges of urbanization. Creating comprehensive and effective systems for land management in urban areas requires a significant long-term effort. However, some procedures should be undertaken immediately to mitigate the potential negative impacts of urban problems on human life. In developing countries, public records may not reflect the current status of buildings. Thus, implementing an automated and rapid building monitoring system using the potential of high-spatial-resolution satellite images and street views may be ideal for urban areas. This study proposed a two-step automated building stock monitoring mechanism. Our proposed method can identify critical building features, such as the building footprint and the number of floors. In the first step, buildings were automatically detected by using the object-based image analysis (OBIA) method on high-resolution spatial satellite images. In the second step, vertical images of the buildings were collected. Then, the number of the building floors was determined automatically using Google Street View Images (GSVI) via the YOLOv5 algorithm and the kernel density estimation method. The first step of the experiment was applied to the high-resolution images of the Pleiades satellite, which covers three different urban areas in Istanbul. The average accuracy metrics of the OBIA experiment for Area 1, Area 2, and Area 3 were 92.74%, 92.23%, and 92.92%, respectively. The second step of the experiment was applied to the image dataset containing the GSVIs of several buildings in different Istanbul streets. The perspective effect, the presence of more than one building in the photograph, some obstacles around the buildings, and different window sizes caused errors in the floor estimations. For this reason, the operator’s manual interpretation when obtaining SVIs increases the floor estimation accuracy. The proposed algorithm estimates the number of floors at a rate of 79.2% accuracy for the SVIs collected by operator interpretation. Consequently, our methodology can easily be used to monitor and document the critical features of the existing buildings. This approach can support an immediate emergency action plan to reduce the possible losses caused by urban problems. In addition, this method can be utilized to analyze the previous conditions after damage or losses occur.

Integrated mixture model and ensemble learning geographic object-based image analysis for road network extraction

ABSTRACT Delineation of road networks is often hindered by shadows, occlusions and spectrally similar objects. A hybrid Geographic object-based Image analysis (GeOBIA) technique that combines Mixture Model segmentation with Ensemble Learning algorithms to extract road networks is proposed. The Multispectral Local Dirichlet Mixture Model (MLDMM) highlights the built-up area. The bagging or subspace integrated ensemble learning-based classification makes the framework immune to overfitting and a novel statistical feature selection phase boosts the performance by 18%. Post-processing by path morphology achieves complete networks. MLDMM-SDEC proffers a precision of 99.86%, a recall of 90.90%, an F1-score of 95.23%, a detection quality of 94.73%, and an accuracy of 96.77%.

Development of an object-based image analysis tool for mass spectrometry imaging ion classification.

Mass spectrometry imaging (MSI) is an analytical technique that can detect and visualize thousands of m/z values resolved in two- and three-dimensional space. These m/z values lead to hundreds of molecular annotations, including on-tissue and background ions. Discrimination of sample-related analytes from ambient ions conventionally involves manual investigation of each ion heatmap, which requires significant researcher time and effort (for a single tissue image, it can take an hour to determine on-tissue and off-tissue species). Moreover, manual investigation lends itself to subjectivity. Herein, we present the utility of an ion classification tool (ICT) developed using object-based image analysis in MATLAB. The ICT functions by segmenting ion heatmap images into on-tissue and off-tissue objects through binary conversion. The binary images are analyzed and within seconds used to classify the ions as on-tissue or background using a binning approach based on the number of detected objects. In a representative dataset with 50 randomly selected annotations, the ICT was able to accurately classify 45/50 ions as on-tissue or background.

Toward the automatic detection of effective gerbil holes in desert grasslands through unmanned aerial vehicle imagery

Rapidly determining the density of effective gerbil holes is ecologically important but technically challenging. However, unmanned aerial vehicles (UAVs) offer new methods to identify effective gerbil holes and understand the spatial relationships between gerbils and grass coverage. In this study, we focused on Meriones unguiculatus, which live in desert grasslands of Ordos, and gathered UAV imagery to explore the performance of rule-based classification in object-based image analysis (OBIA) for identifying effective gerbil holes. To improve the classification accuracy and reduce the time cost of repeated ‘trial and error’, we propose a novel method to build rule sets. We adopted the estimation of scale parameter_2 (ESP2) method during the segmentation stage to enable fast and objective segmentation parameterization. For the classification stage, a correlation-based feature selection (CFS) algorithm was applied for feature selection and threshold prediction. Our analysis determined the following universally recommended rule sets for identifying effective gerbil holes using OBIA: scale parameter of 107, shape factor of 0.2, compactness value of 0.3, excess green value less than −5, brightness feature value in the 143–149 range and roundness feature less than 0.9. With these rule sets, OBIA exhibited higher classification accuracy than maximum likelihood (ML) classification, with an overall classification accuracy of 88.25%. We also found that the relationship between the area of effective gerbil holes (y) and the grass coverage (x) satisfied a quadratic function: This study provides practical guidance for grassland management and rodent infestation control.

Integrated Reconstruction of Late Quaternary Geomorphology and Sediment Dynamics of Prokljan Lake and Krka River Estuary, Croatia

The upper part of the Krka River estuary and Prokljan Lake are a specific example of a well-stratified estuarine environment in a submerged river canyon. Here, we reconstructed the geomorphological evolution of the area and classified the data gathered in the study, integrating multibeam echosounder data, backscatter echosounder data, side-scan sonar morpho-bathymetric surveys, and acoustic sub-bottom profiling, with the addition of ground-truthing and sediment analyses. This led to the successful classification of the bottom sediments using the object-based image analysis method. Additional inputs to the multibeam echosounder data improved the segmentation of the seafloor classification, geology, and morphology of the surveyed area. This study uncovered and precisely defined distinct geomorphological features, specifically submerged tufa barriers and carbonate mounds active during the Holocene warm periods, analogous to recent tufa barriers that still exist and grow in the upstream part of the Krka River. Fine-grained sediments, classified as estuarine sediments, hold more organic carbon than coarse-grained sediments sampled on barriers. A good correlation of organic carbon with silt sediments allowed the construction of a prediction map for marine sedimentary carbon in this estuarine/lake environment using multibeam echosounder data. Our findings highlight the importance of additional inputs to multibeam echosounder data to achieve the most accurate results.

Mapping global distribution of mangrove forests at 10-m resolution

Mangrove forests deliver incredible ecosystem goods and services and are enormously relevant to sustainable living. An accurate assessment of the global status of mangrove forests warrants the necessity of datasets with sufficient information on spatial distributions and patch patterns. However, existing datasets were mostly derived from ∼30 m resolution satellite imagery and used pixel-based image classification methods, which lacked spatial details and reasonable geo-information. Here, based on Sentinel-2 imagery, we created a global mangrove forest dataset at 10-m resolution, namely, High-resolution Global Mangrove Forests (HGMF_2020), using object-based image analysis and random forest classification. We then analyzed the status of global mangrove forests from the perspectives of conservation, threats, and resistance to ocean disasters. We concluded the following: (1) globally, there were 145,068 km2 mangrove forests in 2020, among which Asia contained the largest coverage (39.2%); at the country level, Indonesia had the largest amount of mangrove forests, followed by Brazil and Australia. (2) Mangrove forests in South Asia were estimated to be in the better status due to the higher proportion of conservation and larger individual patch size; in contrast, mangrove forests in East and Southeast Asia were facing intensive threats. (3) Nearly, 99% of mangrove forest areas had a patch width greater than 100 m, suggesting that nearly all mangrove forests were efficient in reducing coastal wave energy and impacts. This study reports an innovative and up-to-date dataset and comprehensive information on mangrove forests status to contribute to related research and policy implementation, especially for supporting sustainable development.

Land Use and Land Cover Mapping with VHR and Multi-Temporal Sentinel-2 Imagery

Land Use/Land Cover (LULC) mapping is the first step in monitoring urban sprawl and its environmental, economic and societal impacts. While satellite imagery and vegetation indices are commonly used for LULC mapping, the limited resolution of these images can hamper object recognition for Geographic Object-Based Image Analysis (GEOBIA). In this study, we utilize very high-resolution (VHR) optical imagery with a resolution of 50 cm to improve object recognition for GEOBIA LULC classification. We focused on the city of Nice, France, and identified ten LULC classes using a Random Forest classifier in Google Earth Engine. We investigate the impact of adding Gray-Level Co-Occurrence Matrix (GLCM) texture information and spectral indices with their temporal components, such as maximum value, standard deviation, phase and amplitude from the multi-spectral and multi-temporal Sentinel-2 imagery. This work focuses on identifying which input features result in the highest increase in accuracy. The results show that adding a single VHR image improves the classification accuracy from 62.62% to 67.05%, especially when the spectral indices and temporal analysis are not included. The impact of the GLCM is similar but smaller than the VHR image. Overall, the inclusion of temporal analysis improves the classification accuracy to 74.30%. The blue band of the VHR image had the largest impact on the classification, followed by the amplitude of the green-red vegetation index and the phase of the normalized multi-band drought index.

Glacier area changes in Novaya Zemlya from 1986–89 to 2019–21 using object-based image analysis in Google Earth Engine

AbstractClimate change has had a significant impact on glacier recession, particularly in the Arctic, where glacier meltwater is an important contributor to global sea-level rise. Therefore, it is important to accurately quantify glacier recession within this sensitive region, using multiple observations of glacier extent. In this study, we mapped 480 glaciers in Novaya Zemlya, Russian Arctic, using object-based image analysis applied to multispectral Landsat satellite imagery in Google Earth Engine and quantify the area changes between 1986–89 and 2019–21. The results show that in 1986–89, the total glacierized area was 22 990 ± 301 km2, in 2000–01 the area was 22 525 ± 308 km2 and by 2019–21 the glacier area reduced to 21 670 ± 292 km2, representing a total of 5.8% reduction in glacier area between 1986–89 and 2019–21. Higher glacier area loss was observed on the Barents Sea coast (7.3%) compared to the Kara (4.2%), reflecting previously observed differences in warming trends. The accuracy of the automatically generated outlines of each layer (1986–89, 2000–01 and 2019–21) was evaluated by comparing with manually corrected outlines (reference data) using random sampling, resulting in an overall accuracy estimate of between 96 and 97% compared to the reference data. This automated approach in Google Earth Engine is a promising tool for rapidly mapping glacier change that reduces the amount of time required to generate accurate glacier outlines.

Combining Drones and Deep Learning to Automate Coral Reef Assessment with RGB Imagery

Coral reefs and their associated marine communities are increasingly threatened by anthropogenic climate change. A key step in the management of climate threats is an efficient and accurate end-to-end system of coral monitoring that can be generally applied to shallow water reefs. Here, we used RGB drone-based imagery and a deep learning algorithm to develop a system of classifying bleached and unbleached corals. Imagery was collected five times across one year, between November 2018 and November 2019, to assess coral bleaching and potential recovery around Lord Howe Island, Australia, using object-based image analysis. This training mask was used to develop a large training dataset, and an mRES-uNet architecture was chosen for automated segmentation. Unbleached coral classifications achieved a precision of 0.96, a recall of 0.92, and a Jaccard index of 0.89, while bleached corals achieved 0.28 precision, 0.58 recall, and a 0.23 Jaccard index score. Subsequently, methods were further refined by creating bleached coral objects (>16 pixels total) using the neural network classifications of bleached coral pixels, to minimize pixel error and count bleached coral colonies. This method achieved a prediction precision of 0.76 in imagery regions with >2000 bleached corals present, and 0.58 when run on an entire orthomosaic image. Bleached corals accounted for the largest percentage of the study area in September 2019 (6.98%), and were also significantly present in March (2.21%). Unbleached corals were the least dominant in March (28.24%), but generally accounted for ~50% of imagery across other months. Overall, we demonstrate that drone-based RGB imagery, combined with artificial intelligence, is an effective method of coral reef monitoring, providing accurate and high-resolution information on shallow reef environments in a cost-effective manner.

LAND USE/LAND COVER CLASSIFICATION IN A HETEROGENEOUS AGRICULTURAL LANDSCAPE USING PLANETSCOPE DATA

Abstract. Land use and land cover (LULC) classification has long been an essential topic in Earth Observation research and plays a key role in the sustainable development of agriculture. This study evaluated the accuracy of LULC classification based on an initial clustering step in a heterogeneous agricultural landscape using PlanetScope imagery while checking for variability among their Normalized Difference Vegetation Index (NDVI) temporal signatures. We adopt an object-based image analysis to generate image-objects and then extract statistical information of PlanetScope spectral bands and vegetation indices as input information for classification. The exploratory analysis focused on the double crop class and calculated the distance between NDVI temporal signatures of paired land parcels. We applied an unsupervised clustering technique along with Random Forest algorithm based on multiple tests to classify and analyse gains and losses in accuracies produced by these approaches. Our results showed that the initial clustering method outperformed the non-clustered classification of LULC in overall accuracy measures. The exploratory analysis demonstrated that double crops might present high intra-class variability and diverse crop calendars for neighbour land parcels. The accuracies achieved represent promising opportunities for the sufficiently accurate classification of such areas, and the knowledge of the intra-class variability allows the analyst to infer the temporal dynamics of crop fields. We reinforce that further work could assess other types of classifiers, especially in areas with a large number of crop types and distinct management practices.

GROUND TRUTHING: STUDYING PERI-URBANIZATION AND LAND-BASED LIVELIHOODS IN MANKWENG AND ITS ENVIRONS (LIMPOPO, SOUTH AFRICA)

Abstract. Peri-urbanization is often referred to as a transition zone following either an urban-rural gradient or an inner and outer peri-urban zone. Contrary to these notions and linked to its history in a former homeland, peri-urbanization in Mankweng appears as a complex and erratic process in reaction to shifting livelihoods, in-migration, and climate change. The process of ground truth for the study was designed by accommodating a combination of remote sensing images of WorldView-3 for the densely, vastly changing populated area (200 km2) and of Sentinel-2 for the overall study area of 400 km2. For the purpose of intensive data gathering, 16 plus three additional sites were selected in the area covered with the very high-resolution imagery as well as transect drives in the rural part. The study sites were selected and well distributed based on specific parameters such as building density, home garden size, road type with the goal of better interpreting what can be distinguished on the imagery and what influences land-based livelihoods. Additional information was gathered from informants during the ground truth through open questions and informal discussions. A solid knowledge of the land uses and land cover, including housing structures, agricultural practices, and challenges of land-based subsistence, was attained. The results revealed influential change drivers, settlement structure types, distinct development patterns, changes in traditional land-based livelihood specifically in terms of agriculture in the region and indicators for a functional zoning of the area. The ground truth provides a thorough basis for the object-based image analysis of the 200 km2 peri-urban area being next.

Smallholder oil palm plantation sustainability assessment using multi-criteria analysis and unmanned aerial vehicles

Oil palm agriculture has caused extensive land cover and land use changes that have adversely affected tropical landscapes and ecosystems. However, monitoring and assessment of oil palm plantation areas to support sustainable management is costly and labour-intensive. This study used an unmanned aerial vehicles (UAV) to map smallholder farms and applied multi-criteria analysis to data generated from orthomosaics, to provide a set of sustainability indicators for the farms. Images were acquired from a UAV, with structure from motion (SfM) photogrammetry then used to produce orthomosaics and digital elevation models of the farm areas. Some of the inherent problems using high spatial resolution imagery for land cover classification were overcome by using texture analysis and geographic object-based image analysis (OBIA). Six spatially explicit environmental metrics were developed using multi-criteria analysis and used to generate sustainability indicator layers from the UAV data. The SfM and OBIA approach provided an accurate, high-resolution (~5 cm) image-based reconstruction of smallholder farm landscapes, with an overall classification accuracy of 89%. The multi-criteria analysis highlighted areas with lower sustainability values, which should be considered targets for adoption of sustainable management practices. The results of this work suggest that UAVs are a cost-effective tool for sustainability assessments of oil palm plantations, but there remains the need to plan surveys and image processing workflows carefully. Future work can build on our proposed approach, including the use of additional and/or alternative indicators developed through consultation with the oil palm industry stakeholders, to support certification schemes such as the Roundtable on Sustainable Palm Oil (RSPO).

A new quantitative evaluation method of urban skyline based on object-based analysis and constitution theory

The skyline is a comprehensive display of the morphological characteristics and cultural features of a city, and its quantitative evaluation is remarkable for perceiving the city and assisting in its planning. However, previous studies focused on the overall profile or tall buildings, lacking a perspective that considers the composition of skyline objects. This paper proposes a new quantitative method to evaluate the skyline based on the object-based analysis method and the constitution theory. Firstly, the skyline objects, that is, buildings, vegetation and mountains are extracted by using the object-based image analysis method. Secondly, the buildings are further classified into four classes according to their relative height. Then, two quantitative indicators, namely, richness of the object category variety and complexity of the object category spatial distribution, are proposed by considering the constitution theory. Finally, this method is applied to typical urban skylines in Shanghai, Hong Kong, New York and Vancouver. Results show that the new indicators can effectively represent the differences of city skylines when their profile indicators are relatively similar. The method can quantitatively evaluate the composition and spatial distribution of skyline objects. This paper is expected to provide a new perspective on the study of skyline aesthetics.

A comparison between Pixel-based deep learning and Object-based image analysis (OBIA) for individual detection of cabbage plants based on UAV Visible-light images

It is challenging to accurately and rapidly extract crops based on the ultra-high spatial resolution images of uncrewed aerial vehicle (UAV). Object-based image analysis (OBIA) was regarded as an effective technique for high-spatial-resolution image classification because of its ability to achieve high accuracy by integrating multi-dimensional features. In recent years, deep learning (DL) techniques, with their ability to automatically learn image features from a large number of images, have shown great potential for crop monitoring. However, a systematic comparison of these two mainstream methods for monitoring the crop phenotype has not been conducted. Therefore, this study compares the performance of two advanced methods, DL and OBIA, in individual cabbage plant detection tasks. The results show that the Mask R-CNN deep learning model outperforms the object-based image analysis-multilevel distance transform watershed segmentation (OBIA-MDTWS) method in crop extraction and counting, with an overall mean F1-Score, accuracy of 2.70, 4.15 percentage points higher, respectively. Moreover, the Mask R-CNN deep learning model has higher computing efficiency, which is 3.74 times higher than the OBIA-MDTWS model. In summary, this study shows that the Mask R-CNN deep learning model performs better in vegetable extraction and quantity estimation, providing technical support for subsequent field nursery management and fine planting.

Can Plot-Level Photographs Accurately Estimate Tundra Vegetation Cover in Northern Alaska?

Plot-level photography is an attractive time-saving alternative to field measurements for vegetation monitoring. However, widespread adoption of this technique relies on efficient workflows for post-processing images and the accuracy of the resulting products. Here, we estimated relative vegetation cover using both traditional field sampling methods (point frame) and semi-automated classification of photographs (plot-level photography) across thirty 1 m2 plots near Utqiaġvik, Alaska, from 2012 to 2021. Geographic object-based image analysis (GEOBIA) was applied to generate objects based on the three spectral bands (red, green, and blue) of the images. Five machine learning algorithms were then applied to classify the objects into vegetation groups, and random forest performed best (60.5% overall accuracy). Objects were reliably classified into the following classes: bryophytes, forbs, graminoids, litter, shadows, and standing dead. Deciduous shrubs and lichens were not reliably classified. Multinomial regression models were used to gauge if the cover estimates from plot-level photography could accurately predict the cover estimates from the point frame across space or time. Plot-level photography yielded useful estimates of vegetation cover for graminoids. However, the predictive performance varied both by vegetation class and whether it was being used to predict cover in new locations or change over time in previously sampled plots. These results suggest that plot-level photography may maximize the efficient use of time, funding, and available technology to monitor vegetation cover in the Arctic, but the accuracy of current semi-automated image analysis is not sufficient to detect small changes in cover.

Ocher deposit prospecting using object-based image analysis of WorldView-3 VNIR data: A case study in Hormuz Island, southern Iran

Mapping lithological units and conducting mineral exploration in a relatively short time and at a reduced cost, requires high-resolution satellite data and state-of-the-art image processing approaches and methods. In this regard, the current paper aims to test an integration of visible-near infrared imagery from WorldView-3 (WV-3) and object-based classification for mapping potential deposits of ocher in Hormuz Island, southern Iran. A combination of field observations, spectroscopy, microscopic mineralogy, and geochemical-based XRD and XRF analyses were conducted on the samples collected throughout the study area. The reflectance and absorption features of the ocher were extracted from spectroscopic measurements of samples from well-known ocher mines at the area, and were then convolved to the WV-3 bands. Results showed that the spectral characteristics of ocher are governed by iron oxides and clay minerals with a presence of distinct strong absorption and high reflectance features in the 510–625 nm and 630–745 nm wavelength regions, respectively. These features are complemented by molecular vibration processes of water O–H intramolecular stretching and H–O–H bending that generate absorption features in the 1440–1940 nm region. Additional absorption features in the 2210–2300 nm are most likely due to the Al-OH and CO3−2 vibrations. The absorptions centered at 480 nm, 540 nm, and 820 nm correspond to bands 2, 3, and 7 of the WV-3, respectively, whereas the high reflectance feature centered near 700 nm corresponds to band 6. These four bands, which were considered as index bands of ocher in this study, were used to assign the segmentation weights and to create the thresholds during image processing. Brightness, density, compactness, and homogeneity features were the primary factors for selecting the training areas in the index bands. The ocher-rich areas were enhanced by using a two-stepapproach of object-based image analysis (OBIA) and image classification, for suitable threshold selection. The information about the feature variables of the ocher within the image and lithological object hierarchy were obtained to evaluate the features of ocher deposits. Information extracted from the index bands provided an important description of the object features, including mean, standard deviation, minimum and maximum pixel values, hue, saturation, and intensity, while the training areas obtained information from the three known ocher mines. Considering the fixed numerical range of ocher mines as a reference, the fixed numerical range of general and specific features of ocher in index bands of WV-3 were achieved. The numerical ranges of pixels were used for creating threshold conditions when applying the “assign class” algorithm in bands 2, 3, and 7 are 815–975 nm, 930–1052 nm, and 1721–1904 nm, respectively. The presented OBIA approach shows a high potential with an overall accuracy of 88 % to discriminate deposits of ocher based on compatibility between ocher mines and identified ocher-bearing pixels.