WorldView-2 Research Articles

Tree Species Classification in Mixed Deciduous Forests Using Very High Spatial Resolution Satellite Imagery and Machine Learning Methods

Spatially explicit information on tree species composition is important for both the forest management and conservation sectors. In combination with machine learning algorithms, very high-resolution satellite imagery may provide an effective solution to reduce the need for labor-intensive and time-consuming field-based surveys. In this study, we evaluated the possibility of using multispectral WorldView-3 (WV-3) satellite imagery for the classification of three main tree species (Quercus robur L., Carpinus betulus L., and Alnus glutinosa (L.) Geartn.) in a lowland, mixed deciduous forest in central Croatia. The pixel-based supervised classification was performed using two machine learning algorithms: random forest (RF) and support vector machine (SVM). Additionally, the contribution of gray level cooccurrence matrix (GLCM) texture features from WV-3 imagery in tree species classification was evaluated. Principal component analysis confirmed GLCM variance to be the most significant texture feature. Of the 373 visually interpreted reference polygons, 237 were used as training polygons and 136 were used as validation polygons. The validation results show relatively high overall accuracy (85%) for tree species classification based solely on WV-3 spectral characteristics and the RF classification approach. As expected, an improvement in classification accuracy was achieved by a combination of spectral and textural features. With the additional use of GLCM variance, the overall accuracy improved by 10% and 7% for RF and SVM classification approaches, respectively.

The Use of Colorimeters to Support Remote Sensing Techniques on Asphalt Pavements

Characterization of asphalt pavements, based on ground spectroradiometers, has been studied in the past to determine their spectral response concerning the physical, chemical, and condition properties of the pavement. This paper suggests an alternative technique for characterizing ageing of asphalt pavements using a colorimeter. Colorimeters are considered as affordable equipment in laboratories in contrast to other scientific instruments and turn remote sensing ground techniques more accessible to industry. Therefore, the study proposes a new methodology indicating how colorimeters can be used in combination with satellite data for the age characterization of asphalt pavements. Spectroradiometer data are compared in a two-way methodology to colorimeter data. The final steps of the methodology used in the study show very similar results for both equipment after a comparison of separability indices (Euclidean and Mahalanobis distances). It is a fact that colorimeter data can be used as ground truth data. The application was performed using an in-band analysis of WorldView 3 (WV3) spectral bands situated in the visible electromagnetic spectrum. Based on the findings of this study, we proposed the Normalized Difference Equation/filter for asphalt Pavement Age characterization Index (NDPAI).

Geographic Accuracy Assessment of Geometric Corrections of world view-3 Satellite Images Using Polynomial Model (Dept.C)

Updating large-scale maps is one of the main problems in Egypt. These maps can be either many years out of date or even non-existent. On the other hand, the recent enhancements that have been made to the spatial resolution of remote sensing imagery have more accurately enabled the modeling of the Earth's surface features at short intervals. So, if the current High-Resolution Satellite Images (HRSI) are corrected from available geo-referenced topographic maps, digital layers for different applications will be available, as well as rationalization of national budget that are spent on their updating will be offered. In the current research, high resolution satellite images of world view-3 dated 2018, have been corrected geometrically using polynomial model. This model depend on the selection of several ground control points (GCPs) on the distorted image and map them to their true locations in ground coordinates measured from geo-referenced topographic maps. The geometric corrections accuracy of resulted image had been assessed quantitatively and qualitatively. The suitable scale for producing GIS layers from HRSI was also determined. The final result of the experimental works of geometric correction of HRSI using eight ground control points was calculated (Average of RMS error of image = 0.27m), and also the result of the positioning errors using eleven check points available for accuracy quantitative assessment was calculated as (2.59m). The output suitable scale maps for layers production is 1:5000, calculated depending on planimetric allowable accuracy (RMS), taking into consideration the standards of 0.5 mm of map scale.

A semi-empirical approach to quantify and handle the effect of moisture on spectral unmixing

Spectral unmixing techniques aim to retrieve the abundances of the materials contributing to a spectral measurement by inverting the underlying spectral mixture occurring at the surface level. In environment and Earth sciences, the darkening effect of moisture on soils and rocks is a common challenge that may lead to major errors when neglected. In this study, a linear function was calibrated to relate a single moisture index to its spectral effect. The function was then combined with the Neighbor-Band Ratio Unmixing (NBRU) and the Linear Spectral Unmixing (LSU) approaches and applied on spectra of a crafted mixture and a rock sample acquired in laboratory conditions, and on a WorldView-3 (WV3) superspectral image of the Canadian Malartic open pit mine. In laboratory conditions, the addition of the moisture function led to a higher stability and a lower deviation of the abundances retrieved with varying water contents (WC). The combined method also obtained a moisture map of the Mine Canadian Malartic consistent with the known systematic watering performed by the mining company.

A New Benchmark Based on Recent Advances in Multispectral Pansharpening: Revisiting Pansharpening With Classical and Emerging Pansharpening Methods

Pansharpening refers to the fusion of a multispectral (MS) image and panchromatic (PAN) data aimed at generating an outcome with the same spatial resolution of the PAN data and the spectral resolution of the MS image. In the last 30 years, several approaches to deal with this issue have been proposed. However, the reproducibility of these methods is often limited, making the comparison with the state of the art hard to achieve. Thus, to fill this gap, we propose a new benchmark consisting of recent advances in MS pansharpening. In particular, optimized classical approaches [multiresolution analysis (MRA) and component substitution (CS)] are compared with methods belonging to the third generation of pansharpening, represented by variational optimization-based (VO) and machine learning (ML) techniques. The benchmark is tested on different scenarios (from urban to rural areas) acquired by different commercial sensors [i.e., IKONOS (IK), GeoEye-1 (GE-1), and WorldView-3 (WV-3)]. Both quantitative and qualitative assessments and the computational burden are analyzed in this article, and all of the implementations have been collected in a MATLAB toolbox that is made available to the community.

Optimization of Optical Image Geometric Modeling, Application to Topography Extraction and Topographic Change Measurements Using PlanetScope and SkySat Imagery

The volume of data generated by earth observation satellites has increased tremendously over the last few decades and will increase further in the coming decade thanks in particular to the launch of nanosatellites constellations. These data should open new avenues for Earth surface monitoring due to highly improved spectral, spatial and temporal resolution. Many applications depend, however, on the accuracy of the image geometric model. The geometry of optical images, whether acquired from pushbroom or frame systems, is now commonly represented using a Rational Function Model (RFM). While the formalism has become standard, the procedures used to generate these models and their accuracies are diverse. As a result, the RFM models delivered with commercial data are commonly not accurate enough for 3-D extraction, subpixel registration or ground deformation measurements. In this study, we present a methodology for RFM optimization and demonstrate its potential for 3D reconstruction using tri-stereo and multi-date Cubesat images provided by SkySat and PlanetScope, respectively. We use SkySat data over the Morenci Mine, Arizona, which is the largest copper mine in the United States. The re-projection error after the RFM refinement is 0.42 pix without using ground control points (GCPs). Comparison of our Digital Elevation Model (DEM with ~3 m GSD) with a reference DEM obtained from an airborne LiDAR survey (with ~1 m GSD) over stable areas yields a standard deviation of the elevation differences of ~3.9 m. The comparison of the two DEMs allows detecting and measuring the topographic changes due to the mine activity (excavation and stockpiles). We assess the potential of PlanetScope data, using multi-date DOVE-C images from the Shisper glacier, located in the Karakoram (Pakistan), which is known for its recent surge. We extracted DEMs in 2017 and 2019 before and after the surge. The re-projection error after the RFM refinement is 0.38 pix without using GCPs. The accuracy of our DEMs (with ~9 m GSD) is evaluated through comparison with the SRTM DEM (GSD ~30 m) and with a DEM (GSD ~2 m) calculated from Geoeye-1 (GE-1) and World-View-2 (WV-2) stereo images. The standard deviation of the elevation differences in stable areas between the PlanetScope DEM and SRTM is ~12 m, and ~7 m with the GE-1&WV-2 DEM. The mass transfer due to the surge is clearly revealed from a comparison of the 2017 and 2019 DEMs. The study demonstrates that, with the proposed scheme for RFM optimization, times series of DEM extracted from SkySat and PlanetScope images can be used to measure topographic changes due to mining activities or ice flow, and could also be used to monitor geomorphic processes such as landslides, or coastal erosion for example.

Assessing the Fractional Abundance of Highly Mixed Salt-Marsh Vegetation Using Random Forest Soft Classification

Coastal salt marshes are valuable and critical components of tidal landscapes, currently threatened by increasing rates of sea level rise, wave-induced lateral erosion, decreasing sediment supply, and human pressure. Halophytic vegetation plays an important role in salt-marsh erosional and depositional patterns and marsh survival. Mapping salt-marsh halophytic vegetation species and their fractional abundance within plant associations can provide important information on marsh vulnerability and coastal management. Remote sensing has often provided valuable methods for salt-marsh vegetation mapping; however, it has seldom been used to assess the fractional abundance of halophytes. In this study, we developed and tested a novel approach to estimate fractional abundance of halophytic species and bare soil that is based on Random Forest (RF) soft classification. This approach can fully use the information contained in the frequency of decision tree “votes” to estimate fractional abundance of each species. Such a method was applied to WorldView-2 (WV-2) data acquired for the Venice lagoon (Italy), where marshes are characterized by a high diversity of vegetation species. The proposed method was successfully tested against field observations derived from ancillary field surveys. Our results show that the new approach allows one to obtain high accuracy (6.7% < root-mean-square error (RMSE) < 18.7% and 0.65 < R2 < 0.96) in estimating the sub-pixel fractional abundance of marsh-vegetation species. Comparing results obtained with the new RF soft-classification approach with those obtained using the traditional RF regression method for fractional abundance estimation, we find a superior performance of the novel RF soft-classification approach with respect to the existing RF regression methods. The distribution of the dominant species obtained from the RF soft classification was compared to the one obtained from an RF hard classification, showing that numerous mixed areas are wrongly labeled as populated by specific species by the hard classifier. As for the effectiveness of using WV-2 for salt-marsh vegetation mapping, feature importance analyses suggest that Yellow (584–632 nm), NIR 1 (near-infrared 1, 765–901 nm) and NIR 2 (near-infrared 2, 856–1043 nm) bands are critical in RF soft classification. Our results bear important consequences for mapping and monitoring vegetation-species fractional abundance within plant associations and their dynamics, which are key aspects in biogeomorphic analyses of salt-marsh landscapes.

Mapping eucalypts trees using high resolution multispectral images: A study comparing WorldView 2 vs. SPOT 7

Invasive alien plants are considered as a major threat to many ecological and socio-economic systems. Nevertheless, the management of some of these plants is often controversial due to the positive socio-economic and ecosystem roles they play. This necessitates proper mapping and monitoring of the extent and spatial distribution of such plants to prioritize resource allocation and management. However, mapping plant species using remote sensing in a heterogeneous environment such as an urban area is often challenged by high levels of spectral muddle. This study investigated the utility of the high and medium spectral and spatial resolution imageries from the WorldView-2 (WV-2) and SPOT-7 satellites, respectively, to map eucalypts trees in urban areas. Furthermore, the classification performances of Random Forest (RF) and Support Vector Machines (SVM) were compared. Both WV-2 and SPOT-7 imageries attained overall accuracies of 81.67% (0.78 kappa) and 72.78% (0.67 kappa), respectively, when the RF algorithm was used and 80% (0.76 Kappa) and 71.11% (0.65 Kappa), respectively when SVM algorithm was used. The user’s accuracies for the eucalypts class in both WV-2 and SPOT 7 imageries were 73.33% and 60%, respectively, for the RF and 70% and 56.67% for the SVM algorithm, respectively. Thus, WV-2 imagery is more suitable for mapping eucalypts trees in a heterogeneous urban environment. Therefore, the classification of WV-2 imageries using RF produced a relatively more accurate map of the eucalypts trees for the study areas, the southern part of Johannesburg city.

Prospecting for Gold Mineralization Using Geochemical, Mineralogical, and WorldView-2 Data: Siyah Jangal Area Case Study, Northern Taftan Volcano, SE Iran

This study applies a combination of geochemical, mineralogical, and WorldView-2 (WV-2) data for gold prospecting in the Siyah Jangal area, southeastern Iran. Geochemical and mineralogical studies, image processing, and field studies were used to identify areas of high-potential gold mineralization. Elemental concentrations in samples collected from exploration trenches in a well-known gold mineralization area were applied to identify Au pathfinders using bivariate and multivariate statistics. Visible-near infrared-shortwave infrared spectroscopy was applied to identify mineral phases. Correlation analysis, principal component analysis, and cluster analysis of measured element concentrations showed that Au has strong associations with As, Pb, Fe, and Cu, among which, Fe is the only element that forms major minerals in the area. The spectroscopy analyses showed that iron oxide-hydroxide bearing minerals including jarosite, hematite, and goethite are the major Fe-bearing phases in the surficial environment. Therefore, mineral fraction images of these minerals were produced by performing matched filtering method on WV-2 data, and the fractions of more than 0.25 of goethite, 0.15 of hematite, and 0.10 of jarosite were combined and converted as map layers to generate a false color composite (RGB) image for targeting potential areas of gold mineralization. The accuracy of iron oxide-hydroxides mapping was proven by field observations and XRD analyses of surficial samples. Moreover, concentrations of Au in samples collected from exploration boreholes in previously mapped areas showed good correlations between enhanced areas by remote sensing studies and gold occurrences.

AUTOMATIC INDICES BASED CLASSIFICATION METHOD FOR MAP UPDATING USING VHR SATELLITE IMAGES

Urban land cover classification using Very High Resolution (VHR) satellite images is a very important source of information for map updating. Egyptian environment has more challenges in feature extraction. The main problem lies in the spectral similarity between different land cover classes. Also, great diversity in sizes, shapes, and materials of each class. The main aim of this work is to represent a new automatic indices-based classification method for map updating using VHR satellite images. The method uses a set of spectral indices with their thresholds in consecutive order, chosen based on WorldView-2 (WV-2) bands, to classify land cover in the Egyptian environment. For this study, WV-2 satellite images with eight spectral bands were used. The proposed method is compared with five traditional classification methods; Minimum distance, Spectral angle mapper, Mahalanobis distance, Spectral correlation mapper, and Maximum likelihood method, which included in ERDAS 2015 software, for validation purpose and checking its stability. The results show that the extracted features with the proposed method can contribute significantly to update Egyptian medium scale maps. The average overall accuracy achieved with the proposed approach (75.31%) is higher than those obtained using Minimum distance (54.0%), Spectral angle mapper (69.50%), and Mahalanobis distance (73.63%). Also, it is near to those obtained by the Spectral correlation mapper (76.50%), and Maximum likelihood method (78.25%).

Toward Mapping Dietary Fibers in Northern Ecosystems Using Hyperspectral and Multispectral Data

Shrub proliferation across the Arctic from climate warming is expanding herbivore habitat but may also alter forage quality. Dietary fibers—an important component of forage quality—influence shrub palatability, and changes in dietary fiber concentrations may have broad ecological implications. While airborne hyperspectral instruments may effectively estimate dietary fibers, such data captures a limited portion of landscapes. Satellite data such as the multispectral WorldView-3 (WV-3) instrument may enable dietary fiber estimation to be extrapolated across larger areas. We assessed how variation in dietary fibers of Salix alaxensis (Andersson), a palatable northern shrub, could be estimated using hyperspectral and multispectral WV-3 spectral vegetation indices (SVIs) in a greenhouse setting, and whether including structural information (i.e., leaf area) would improve predictions. We collected canopy-level hyperspectral reflectance readings, which we convolved to the band equivalent reflectance of WV-3. We calculated every possible SVI combination using hyperspectral and convolved WV-3 bands. We identified the best performing SVIs for both sensors using the coefficient of determination (adjusted R2) and the root mean square error (RMSE) using simple linear regression. Next, we assessed the importance of plant structure by adding shade leaf area, sun leaf area, and total leaf area to models individually. We evaluated model fits using Akaike’s information criterion for small sample sizes and conducted leave-one-out cross validation. We compared cross validation slopes and predictive power (Spearman rank coefficients ρ) between models. Hyperspectral SVIs (R2 = 0.48–0.68; RMSE = 0.04–0.91%) outperformed WV-3 SVIs (R2 = 0.13–0.35; RMSE = 0.05–1.18%) for estimating dietary fibers, suggesting hyperspectral remote sensing is best suited for estimating dietary fibers in a palatable northern shrub. Three dietary fibers showed improved predictive power when leaf area metrics were included (cross validation ρ = +2–8%), suggesting plant structure and the light environment may augment our ability to estimate some dietary fibers in northern landscapes. Monitoring dietary fibers in northern ecosystems may benefit from upcoming hyperspectral satellites such as the environmental mapping and analysis program (EnMAP).

ANALYSIS ON THE EFFECT OF SPATIAL AND SPECTRAL RESOLUTION OF DIFFERENT REMOTE SENSING DATA IN SUGARCANE CROP YIELD STUDY

Abstract. Sugarcane is a perennial crop that contributes to nearly 80% of the global sugar-based products. Therefore, sugarcane growers and food companies are seeking ways to address the concerns related to sugarcane crop yield and health. In this study, a spatial and spectral analysis on the peak growth stage of the sugarcane fields in Bundaberg, Queensland, Australia is performed using the Normalized Difference Vegetation Index (NDVI) and Normalized Difference Red Edge Index (NDRE) derived from high-resolution WorldView-2 (WV2) images and multispectral Unmanned Aerial Vehicle (UAV) images. Two topics are chosen for this study: 1) the difference and correlation between NDVI and NDRE that are commonly used to estimate Leaf Area Index, a common crop parameter for the assessment of crop yield and health stages; 2) the impact of spatial resolution on the systematic difference in the abovementioned two Vegetation Indices (VIs). The statistical correlation analysis between the WV2 and UAV images produced correlation coefficients of 0.68 and 0.71 for NDVI and NDRE, respectively. In addition, an overall comparison of the WV2 and UAV-derived VIs indicated that the UAV images produced a better accuracy than the WV2 images because UAV can effectively distinguish various status of vegetation owing to its high spatial resolution. The results illustrated a strong positive correlation between NDVI and NDRE, each derived from the WV2 and UAV images, and the correlation coefficients were 0.81 and 0.90, respectively, i.e. the correlation between NDVI and NDRE is higher in the UAV images than the WV2 images.

TREE SPECIES CLASSIFICATION BASED ON NEUTROSOPHIC LOGIC AND DEMPSTER-SHAFER THEORY

Abstract. The objective of this study was to explore the use of multi-source remotely sensed data for individual tree species. To achieve this, a neutrosophic logic-based method was developed for tree species classification using the combined spectral, textural and structural information derived from WorldView-2 (WV-2) multispectral bands, WV-2 panchromatic band, and LiDAR (Light Detection And Ranging)-derived canopy height model (CHM), respectively. The developed method was tested on the data obtained over the Keele campus, York University, Toronto Canada and the KNN (K Nearest Neighbour) classification method. Twenty-one spectral, three textural and three structural features were used to classify five species (Norway maple, honey locust, Austrian pine, blue spruce, and white spruce). For this study, 522 trees were used for training and 223 for testing. The overall classification accuracy obtained by the proposed method was 0.82. It was significantly improved compared with the KNN (0.73), weighted KNN (0.76), and fuzzy KNN (0.75) methods. In addition, Dempster-Shafer (DS) theory was explored to perform information fusion at the decision level in comparison to that at the feature level. The accuracies obtained by the fusion at the decision level were generally lower than those at the feature level. Even though promising results based on the neutrosophic logic were obtained during this proof-concept stage, studies are underway to perform more tests with a large number of tree crowns and more species and exploit other classification methods, such as support vector machine.

Evaluation of Sentinel-2 and Landsat 8 Images for Estimating Chlorophyll-a Concentrations in Lake Chad, Africa

Much effort has been applied in estimating the concentrations of chlorophyll-a (Chl a) in lakes. The optical complexity and lack of in situ data complicate estimating Chl a in such water bodies. We compared four established satellite reflectance algorithms—the two-band and three-band algorithms (2BDA, 3BDA), fluorescence line height (FLH), and normalized difference chlorophyll index (NDCI)—to estimate Chl a concentration in Lake Chad. We evaluated the performance and applicability of Landsat-8 (L8) and Sentinel-2 (S2) images with the four Chl a estimation algorithms. For accuracy, we compared the concentration levels from the four algorithms to those from Worldview-3 (WV3) images. We identified two promising algorithms that could be used alongside L8 and S2 satellite images to monitor Chl a concentrations in Lake Chad. With an averaged R2 of 0.8, the 3BDA and NDCI Chl a algorithms performed accurately with S2 and L8 images. For the S2 and L8 images, 3BDA had the highest performance when compared to the WV3 estimates. We demonstrate the usefulness of sensor images in improving water quality information for areas that are difficult to access or when conventional data are limited.

The potential of in-situ hyperspectral remote sensing for differentiating 12 banana genotypes grown in Uganda

Bananas and plantains provide food and income for more than 50 million smallholder farmers in East and Central African (ECA) countries. However, banana productivity generally achieves less than optimal yield potential (<30%) in most regions, including Uganda. Numerous studies have been undertaken to identify the key challenges that smallholder banana growers face at different stages of the banana value chain, with one of the main constraints being a lack of policy-relevant agricultural data. The World Bank (WB) initiated a methodological survey design aimed at identifying the distribution of banana varieties across a number of key Ugandan growing regions, at the individual household scale. To achieve this outcome a number of approaches including ground-based surveys, DNA tissue collection of selected banana plants and remote sensing were evaluated. For the remote sensing component, the set objectives were to develop statistical models from the hyperspectral reflectance properties of individual leaves that could differentiate typical ECA banana varieties, as well as their parentage (usage). The study also explored the potential of extrapolating the ground-based hyperspectral measures to high-resolution WorldView-3 (WV3) satellite imagery, therefore creating the potential of mapping the distribution of banana varieties at a regional scale. The DNA testing of 43 banana varieties propagated at the National Banana Research Program site at National Agricultural Research Organization (NARO) research station in Kampala, Uganda, identified 12 genetically different varieties. A canonical powered partial least square (CPPLS) model developed from hyperspectral reflectance properties of the sampled banana leaves successfully differentiated BLU, BOG, GON, GRO and KAY genotypes. The Random Forest (RF) algorithm was also evaluated to determine if spectral bands coinciding with those provided by WV3 data could segregate banana varieties. The results suggested that this was achievable and as such presents an opportunity to extrapolate the hyperspectral classifications to broader areas of land. The ability to spectrally differentiate these five genotypes has merit as they are not typical east African varieties. As such, identifying the distribution and density of these varieties across Uganda provides vital information to the banana breeders of NARO of where their new varieties are being disseminated too, data that has been previously difficult to obtain. Although the results from this pilot study indicated that not all banana varieties could be spectrally differentiated, the methodology developed and the positive results that were achieved do present remote sensing as a complimentary technology to the ongoing surveying of banana and other crop types grown within Ugandan household farming systems.

ASTER and WorldView-3 satellite data for mapping lithology and alteration minerals associated with Pb-Zn mineralization

Advanced Spaceborne Thermal Emission and Reflection Radiometer (ASTER) and WorldView-3 (WV-3) satellite remote sensing data were used for mapping lithological units and hydrothermal alteration zones associated with Pb-Zn mineralization in the Kerman–Kashmar Tectonic Zone (KKTZ), Iran. The visible near infrared (VNIR), shortwave infrared (SWIR) and thermal infrared (TIR) bands of ASTER were used to map iron oxide/hydroxides, Al-OH minerals, Fe,Mg-OH minerals, quartz and carbonate minerals. The VNIR bands of WV-3 were used to discriminate Fe3+ and Fe2+ absorption intensities. Selective Principal Component analysis (SPCA), Spectral Angle Mapper (SAM), Linear Spectral Unmixing (LSU) and Automatic Lineament Extraction techniques were implemented. Lithological units were discriminated based on Al/Fe-OH, Fe2+/Fe3+ and Mg-Fe-OH/CO3 absorption properties. The spatial distribution of hematite, goethite, jarosite, gypsum, calcite, dolomite, kaolinite and muscovite were comprehensively detected. Some prospective zones were identified in the intersection of N-S, NW-SE and NE-SW trending fault systems, gossan, argillic/phyllic and dolomitic units.

Parcel-level mapping of crops in a smallholder agricultural area: A case of central China using single-temporal VHSR imagery

Timely and accurate monitoring of crop patterns in smallholder agricultural areas is essential for guiding local crop yield estimates, agricultural subsidy allocations, and food security policy formulation. The household-operated land management model leads to a highly fragmented and heterogeneous agricultural landscape; therefore, fine crop mapping in smallholder agricultural areas remains challenging. By using very high spatial resolution (VHSR) images, this study aimed to explore parcel-level crop mapping methods using the case of a typical smallholder agricultural area in Wuhan, China. Object-based image analysis techniques as well as random forest (RF) and support vector machine (SVM) classifiers were used to classify WorldView-2 (WV2) images into eight crop-level agricultural land use categories. Several classification models were built using the combination of two classifiers and different image features, including spectral, geometrical, and textural features. The results showed that the classification model using the RF classifier and all 27 selected features had the highest accuracy, with an overall accuracy of 80.04% and a kappa value of 0.78; specifically, the user’s and producer’s accuracies of rice, cotton, lotus, bare paddy field and bare upland field exceeded 80%. We found that the performance of the RF and SVM classifiers was generally comparable, although as the input features increased, the accuracy of the RF was slightly higher than that of the SVM. The use of spatial features, such as the gray level cooccurrence matrix (GLCM) standard deviation, GLCM correlation, and area of image objects, could help improve the accuracy of parcel-level crop mapping. Our research confirmed the practical value of single-temporal VHRS images and RF classifiers in mapping parcel-level crops in complex agricultural areas. This framework provides a methodological reference for accurately monitoring crop distribution in smallholder agriculture areas to support the development of local precision agriculture.

Fusion of multispectral imagery and LiDAR data for roofing materials and roofing surface conditions assessment

ABSTRACT Assessment of rooftop rainwater harvesting (RRWH) quality and suitability requires detail and reliable information on roofs. Characterization of roof surface conditions affects the quality of harvested rainwater. Nevertheless, the implementation of the system requires improvement in terms of the roof detection techniques to ensure the roof of the building is selected appropriately. Thus, the classification techniques need to be optimized to detect roof materials and roof surface conditions (new or old) with high accuracy. This study aimed to produce high precision detailed roof materials and roof surface conditions map with using high-resolution remote sensing imagery, WorldView-3 (WV3) and light detection and ranging (LiDAR) data. Three different fusion methods; layer stacking (LS), Gram-Schmidt (GS) and principal components spectral sharpening (PCSS) were explored and their performances were compared to improve the spatial and spectral richness of the image. Subsequently, the roof materials and roof surface conditions classes which include old concrete, new concrete, old metal, new metal, old asbestos and new asbestos had been discriminated by employing support vector machine (SVM) and the rule-based technique known as a decision tree (DT). Generally, generated rule-sets present a higher overall accuracy with 87%, 72% and 66% for LS, GS and PCSS, respectively. For SVM classifier, the maximum accuracy recorded for LS, PCSS and GS were 70%, 63% and 43% respectively. Therefore, rule-based classification via LS fusion technique was utilized to identify suitable rooftops for the development of harvested rainwater system in the urban area. Findings indicate that the degradation status of a roof in heterogenous urban environments could be determined from satellite observation and the quality of roof-based harvested rainwater affected by roofing materials and roofing surface conditions can be analysed effectively.

Comparative landslide susceptibility assessment using statistical information value and index of entropy model in Bhanupali-Beri region, Himachal Pradesh, India

Landslide is a complex natural hazard that sometimes causes disaster resulting in loss of life, assets and infrastructure, especially in the Himalayas. Recent studies suggest that for effective mitigation and resilience through proper planning and policymaking, it is equally important to justify and select a suitable scientific technique that most appropriately addresses the salient causes of a landslide in any area. The principal objective of this study is to carry out a comparative assessment between two contemporary statistical techniques, i.e., the statistical information value (SIV) and index of entropy (IOE), to find out the effectiveness of the two said methods in landslide susceptibility mapping in Bhanupali-Beri region. During the analysis, the higher-resolution satellite images, i.e., World view-2 image of 2017 and Landsat-8 OLI image of 2018, have been used for delineation of various triggering parameters used for landslide susceptibility. The contemporary GIS technique integrated with the remote sensing applications was distinct in preparing the prominent landslide conditioning factor layers such as slope, slope aspect, thrust and fault proximity, geomorphology, landuse–landcover, stream power index, topographic wetness index, geology, roads proximity, lineament density and past landslide inventory. The final assessment was performed using GIS software through raster re-sampling, and the values derived for each conditioning factors were combined using defined SIV and IOE equations. The study area was categorized into five distinct landslide susceptible zones (very low, low, moderate, high and very high) using the Jenk’s Natural Breaks algorithm. Index of entropy model has given better results compared to SIV. The utmost vital factors triggering landslide (estimated for entropy values) in the area are landuse–landcover with barren land and sparse vegetation followed by TWI, lineament density, geomorphology, and slope.

Stress Detection in New Zealand Kauri Canopies with WorldView-2 Satellite and LiDAR Data

New Zealand kauri trees are threatened by the kauri dieback disease (Phytophthora agathidicida (PA)). In this study, we investigate the use of pan-sharpened WorldView-2 (WV2) satellite and Light Detection and Ranging (LiDAR) data for detecting stress symptoms in the canopy of kauri trees. A total of 1089 reference crowns were located in the Waitakere Ranges west of Auckland and assessed by fieldwork and the interpretation of aerial images. Canopy stress symptoms were graded based on five basic stress levels and further refined for the first symptom stages. The crown polygons were manually edited on a LiDAR crown height model. Crowns with a mean diameter smaller than 4 m caused most outliers with the 1.8 m pixel size of the WV2 multispectral bands, especially at the more advanced stress levels of dying and dead trees. The exclusion of crowns with a diameter smaller than 4 m increased the correlation in an object-based random forest regression from 0.85 to 0.89 with only WV2 attributes (root mean squared error (RMSE) of 0.48, mean absolute error (MAE) of 0.34). Additional LiDAR attributes increased the correlation to 0.92 (RMSE of 0.43, MAE of 0.31). A red/near-infrared (NIR) normalised difference vegetation index (NDVI) and a ratio of the red and green bands were the most important indices for an assessment of the full range of stress symptoms. For detection of the first stress symptoms, an NDVI on the red-edge and green bands increased the performance. This study is the first to analyse the use of spaceborne images for monitoring canopy stress symptoms in native New Zealand kauri forest. The method presented shows promising results for a cost-efficient stress monitoring of kauri crowns over large areas. It will be tested in a full processing chain with automatic kauri identification and crown segmentation.