Traditional Ground Surveys Research Articles

Detecting canopy openings in logged-over forests: a multi-classifier analysis of PlanetScope imagery

This study focused on the detection of forest canopy openings resulting from harvesting activities in hill tropical forests. Canopy openings, whether natural or human-induced, can have detrimental effects on forest ecosystems. Traditional ground surveys to assess the extent of canopy opening can be challenging and time-consuming. Therefore the study aimed to utilise satellite imagery, specifically PlanetScope data, to detect, map and measure canopy openings in logged-over forests. Three different classification algorithms, namely maximum likelihood classifier (MLC), support vector machine (SVM) and object-based image analysis (OBIA) were used and compared to identify canopy opening areas. The assessment was conducted in two stages: a preliminary assessment with three classes (forest, canopy opening and shadow) and a final assessment with two classes (forest and canopy opening). The overall accuracies of the classification algorithms were 82% for MLC, 91% for SVM and 90% for OBIA. Both SVM and OBIA surpassed the accuracy threshold, with SVM being the most effective in detecting and extracting canopy openings in dense forests. Results demonstrated the potential of PlanetSope imagery and advanced classification algorithms to detect canopy openings in logged-over forests. The findings highlighted the importance of regular updates on canopy opening extent, particularly concerning sustainable forest assessment and minimising the negative impacts on forest ecosystems.

The use of orthoimagery and stereoscopic aerial imagery to identify muskrat (Ondatra zibethicus) houses

AbstractThe muskrat (Ondatra zibethicus) is considered a ubiquitous inhabitant of wetlands across Canada and the United States, but recent studies indicate that muskrat populations in many parts of North America have experienced substantial declines over the last 40–60 years. Monitoring of muskrat abundance is therefore an important task for wildlife managers, but traditional methods such as house counts conducted during ground‐based surveys can be labor‐intensive and time‐consuming. Poor conditions or a lack of access may limit how much of a wetland can be surveyed. Aerial imagery has previously been used to census a diverse array of wildlife populations but is not yet a common tool for muskrat surveys. To investigate the accuracy of this alternative survey method, we collected aerial imagery from coastal wetlands along the north shore of Lake Ontario during the winter of 2014 for examination in both 2D orthorectified and 3D stereoscopic formats. We compared muskrat house counts obtained from imagery to counts recorded by ground survey crews in the same wetlands during the same winter. We found no significant difference between mean muskrat house counts obtained by ground survey crews and orthoimagery observers. In contrast, stereoscopic imagery observers overestimated mean house counts compared to ground survey crews, which we interpret was due to an increase in false positives. Our results indicate that orthoimagery is a promising tool for assessing muskrat occupancy, provides comparable broad‐scale results to traditional ground survey methods, and may be preferable to wildlife managers for a variety of reasons.

Is It More Visible from Above? Comparison of the Effectiveness of Methods for Locating the Bobak Marmot Marmota bobak (Rodentia, Sciuridae) Burrows in Ukraine

Burrow systems detection is a reliable method for counting Bobak marmots. We compared the effectiveness and labour costs of three methods of burrow detection: ground survey, unmanned aerial vehicle survey and satellite survey. The research was conducted in 2021 (Kyiv region). Ground surveys mapped 42 (12 main and 24 secondary) burrows, while UAV imagery identified 45 burrows. The proportion of burrows detected by satellite imagery was 48 % of the number of burrows mapped by the ground survey. Secondary burrows were more difficult to detect than main burrows. The largest area of marmot family group territories was calculated from UAV data (0.27 and 0.08 ha), the smallest from satellite imagery data (0.11 and 0.01 ha). To obtain complete information on marmot colonies, it is best to use either UAV surveys or traditional ground surveys. If it is necessary to find potential areas where animals may live, or to estimate their distribution, then the best option for detecting burrows is the analysis of satellite imagery. The most promising approach is to study colonial burrowing mammals using a combination of ground and remote sensing methods. This provides the opportunity to obtain different types of data.

Satellite Remote Sensing and Geographic Information System Technologies for Industrial Tree Plantation Mapping and Monitoring: A Way Forward for the Sustainable Development of the Philippine ITP Industry

KEY POINTS
 ● Industrial Tree Plantations (ITPs) are a vital contributor to the Philippines' log and timber production, but these ITPs requires better and more efficient way of characterization and monitoring.
 ● Current policies require ground validations surveys to monitor ITPs for regulation purposes. However, these approaches are often difficult to conduct, time consuming and expensive.
 ● Satellite remote sensing (SRS) and Geographic Information System (GIS) technologies offer an efficient alternative to traditional ground surveys for mapping and monitoring ITPs.
 ● SRS and GIS enable detailed mapping of ITPs, including species-level identification, and provides valuable data for forest resource management and environmental monitoring.
 ● SRS and GIS technologies should be integrated into existing policies by the Department of Environment and Natural Resources (DENR) to efficiently map, monitor, and manage ITPs in the country, ensuring informed decision-making and sustainable forestry practices.

Study on the relationship between regional soil desertification and salinization and groundwater based on remote sensing inversion: A case study of the windy beach area in Northern Shaanxi

Soil desertification and salinization are important environmental concerns in arid regions, and their relationship with groundwater change must be further clarified. However, the relationships among soil desertification, salinization, and groundwater are difficult to investigate on a large spatiotemporal scale using traditional ground surveys. In the windy beach area in Northern Shaanxi (WBANS), desertification and salinization problems coexist; therefore, this area was selected as the study area. The feasibility of implementing large-scale remote sensing inversions to identify the degree of desertification and salinization was verified based on measured data, and the degree of influence of groundwater burial depth (GBD) on desertification and salinization was quantified using the geodetector and residual trend analysis methods. The results showed that the GBD in the WBANS presented an increasing trend and the degree of salinization showed a decreasing trend. Moreover, the joint influence of the unique natural environment and anthropogenic activities has led to increases in fractional vegetation cover and considerable improvements in the ecological environment. The intensity of desertification explained by GBD in the WBANS increased significantly (p < 0.01) at a rate of change of 0.0190/year, with high q-values above 0.66 for both Yuyang and Shenmu. The contribution rate of potential evapotranspiration and precipitation to salinization in Yuyang and Shenmu was >97 %, and the contribution rate of GBD to salinization in Dingbian, Jingbian, and Hengshan was 34.78 %, 31.15 %, and 29.41 %, respectively. Overall, the suitable GBD in the WBANS is 2–4 m. The study results provide a reference for research on the inversion, monitoring, and prevention of desertification and salinization dynamics on a large spatiotemporal scale and offer a scientific basis for rationally determining GBD.

Real-time measurement of individual tree structure parameters based on augmented reality in an urban environment

The rapid and accurate acquisition of tree structural parameters is a challenging task in urban tree inventories. Here, we developed a mobile application based on augmented reality (AR) called ARTreeWatch to measure individual tree height (TH) and diameter at breast height (DBH) using an Android smartphone. For the experiments, we installed ARTreeWatch on two types of smartphones equipped with different cameras and inertial measurement units (IMUs) to capture the three-dimensional (3D) point clouds and measure the TH and DBH of 51 trees in three plots with different tree species compositions. When the two types of smartphone measurements were compared with traditional ground surveys, the root mean square error (RMSE) of the estimated DBH ranged from 1.04 cm to 1.21 cm, and the RMSE of the TH ranged from 1.29 m to 1.98 m. Meanwhile, half the labor costs and time were saved compared to traditional field surveys. ARTreeWatch is an efficient, accurate, and lower-cost solution for measuring the structural parameters of trees in urban environments, which may be useful to urban landscape managers and researchers.

The use of drones to study the breeding productivity of Whooper Swan Cygnus cygnus

This research was carried out in north-eastern Poland, where Whooper Swan Cygnus cygnus is a sparse breeder. Whooper Swan is timid and exhibits a tendency to hide in emergent vegetation. A drone was used to improve the effectiveness of research into its breeding success and offspring productivity. The breeding density of Whooper Swan in the study area in 2022 was 10 pairs/100 km2. The number of breeding birds detected at the beginning of the breeding season did not differ according to the method used (drone vs ground). The breeding productivity of the sample of swans studied (N = 36) was 2.19 cygnets per breeding pair using the ground method, but 3.71 per pair with the drone, a significant difference (p-value of the Wilcoxon test = 0.0148). The traditional method showed that 50% of the pairs had bred successfully, whereas with the drone this figure was 79%. The birds either did not react to the drone’s presence or else moved slowly away. The drone study of Whooper Swan breeding productivity was considerably faster (9 min per site vs 1–2 h ground survey), more precise, and less invasive for the birds than a traditional ground survey.

Developing a two-decadal time-record of rice field maps using Landsat-derived multi-index image collections with a random forest classifier: A Google Earth Engine based approach

Historic maps showing the temporal distribution of rice fields are important for precision agriculture, irrigation optimisation, forecasting crop yields, land use management and formulating policies. However, mapping rice fields using traditional ground surveys is impractical when high cost, time and labour requirements are considered, and the availability of such detailed records is limited. Although satellite remote sensing appears to be a viable solution, conventional segmentation and classification methods with spectral bands are often unable to contrast the distinct characteristics between rice fields and other vegetation classes. To this end, we explored a novel, Google Earth Engine (GEE) based multi-index random forest (RF) classification approach to map rice fields over two decades. Landsat images from 2000 to 2020 of two Sri Lankan rice cultivation districts were extracted from GEE and a multi-index RF classification algorithm was applied to distinguish the rice fields. The results showed above 80% accuracy for both training and validation, when compared against high spatial resolution Google Earth imagery. In essence, multi-index sampling and RF together synergised the compelling classification accuracy by effectively capturing vegetation, water (ponding) and soil characteristics unique to the rice fields using a single-click approach. The maps developed in this study were further compared against the MODIS land cover type product (MCD12Q1) and the corresponding superior statistics on rice fields demonstrated the robustness of the proposed approach. Future work seeking effective index combinations is recommended, and this approach can potentially be extended to other crop analyses elsewhere.

Classification of desert grassland species based on a local-global feature enhancement network and UAV hyperspectral remote sensing

Grassland ecosystems are an important part of terrestrial ecosystems and are important for building ecological barriers, promoting the pastoral economy, and maintaining social stability. In recent decades, grasslands in northern China have undergone extensive degradation due to the combined effects of global climate change and the anthropogenic overuse of grasslands. An understanding of the spatial distribution of grassland degradation species is helpful for evaluating the process of grassland degradation and formulating appropriate protective measures. This is important for grassland degradation monitoring. To address the limitations of traditional ground surveys and realize intelligent remote sensing grassland degradation monitoring tasks, we use unmanned aerial vehicle (UAV) hyperspectral remote sensing technology to collect data on vegetation species in desert grasslands. In this paper, we propose a local-global feature enhancement network (LGFEN) for the classification of desert grassland species. The method uses the local feature enhancement (LFE) module and global feature enhancement (GFE) module to extract local and spatial information from hyperspectral images (HSIs), respectively. In addition, the introduction of the convolutional block attention module (CBAM) refines the features of HSIs, improving the stability of the classification performance. The results show that the proposed method has superior classification performance compared with existing HSI classification methods. With only 10 training samples per class, the overall accuracy, average accuracy, and kappa coefficient of the proposed method were 98.61%, 97.61%, and 0.9815, respectively. The proposed method provides a new approach for high-precision and high-efficiency dynamic monitoring of grassland ecosystems.

Identification of Landslides in Mountainous Area with the Combination of SBAS-InSAR and Yolo Model.

Landslides have been frequently occurring in the high mountainous areas in China and poses serious threats to peoples’ lives and property, economic development, and national security. Detecting and monitoring quiescent or active landslides is important for predicting risks and mitigating losses. However, traditional ground survey methods, such as field investigation, GNSS, and total stations, are only suitable for field investigation at a specific site rather than identifying landslides over a large area, as they are expensive, time-consuming, and laborious. In this study, the feasibility of using SBAS-InSAR to detect landslides in the high mountainous areas along the Yunnan Myanmar border was tested first, with fifty-four IW mode Sentinel-1A ascending scenes from 12 January 2019 to 8 December 2020. Next, the Yolo deep-learning model with Gaofen-2 images captured on 5 December 2020 was tested. Finally, the two techniques were combined to achieve better performance, given each of them has intrinsic limitations on landslide detection. The experiment indicated that the combination could improve the match rate between detection results and references, which implied that the performance of landslide detection can be improved with the fusion of time series SAR images and optical images.

Enabling Breeding Selection for Biomass in Slash Pine Using UAV-Based Imaging.

Traditional methods used to monitor the aboveground biomass (AGB) and belowground biomass (BGB) of slash pine (Pinus elliottii) rely on on-ground measurements, which are time- and cost-consuming and suited only for small spatial scales. In this paper, we successfully applied unmanned aerial vehicle (UAV) integrated with structure from motion (UAV-SfM) data to estimate the tree height, crown area (CA), AGB, and BGB of slash pine for in slash pine breeding plantations sites. The CA of each tree was segmented by using marker-controlled watershed segmentation with a treetop and a set of minimum three meters heights. Moreover, the genetic variation of these traits has been analyzed and employed to estimate heritability (h2). The results showed a promising correlation between UAV and ground truth data with a range of R2 from 0.58 to 0.85 at 70 m flying heights and a moderate estimate of h2 for all traits ranges from 0.13 to 0.47, where site influenced the h2 value of slash pine trees, where h2 in site 1 ranged from 0.13~0.25 lower than that in site 2 (range: 0.38~0.47). Similar genetic gains were obtained with both UAV and ground truth data; thus, breeding selection is still possible. The method described in this paper provides faster, more high-throughput, and more cost-effective UAV-SfM surveys to monitor a larger area of breeding plantations than traditional ground surveys while maintaining data accuracy.

From Coastal to Montane Forest Ecosystems, Using Drones for Multi-Species Research in the Tropics

Biodiversity monitoring is crucial in tackling defaunation in the Anthropocene, particularly in tropical ecosystems. However, field surveys are often limited by habitat complexity, logistical constraints, financing and detectability. Hence, leveraging drones technology for species monitoring is required to overcome the caveats of conventional surveys. We investigated prospective methods for wildlife monitoring using drones in four ecosystems. We surveyed waterbird populations in Pulau Rambut, a community of ungulates in Baluran and endemic non-human primates in Gunung Halimun-Salak, Indonesia in 2021 using a DJI Matrice 300 RTK and DJI Mavic 2 Enterprise Dual with additional thermal sensors. We then, consecutively, implemented two survey methods at three sites to compare the efficacy of drones against traditional ground survey methods for each species. The results show that drone surveys provide advantages over ground surveys, including precise size estimation, less disturbance and broader area coverage. Moreover, heat signatures helped to detect species which were not easily spotted in the radiometric imagery, while the detailed radiometric imagery allowed for species identification. Our research also demonstrates that machine learning approaches show a relatively high performance in species detection. Our approaches prove promising for wildlife surveys using drones in different ecosystems in tropical forests.

Capturing hedgerow structure and flowering abundance with UAV remote sensing

AbstractHedgerows are an abundant and ecologically important feature of many rural areas. Their biodiversity value depends on composition, structure and availability of food resources, which can be significantly impacted by poor management. However, information about hedgerow condition is very limited due to field surveys being costly and labour‐intensive. Unmanned aerial vehicles (UAVs) equipped with miniaturized cameras could prove a more cost‐effective and time‐efficient hedgerow surveying solution while preserving a high level of detail unattainable with airborne or satellite sensors. This study explored whether UAV remote sensing is a viable alternative for performing hedgerow condition surveys at local scale, focusing on hedgerow structure and flowering abundance. We acquired UAV Red, Green and Blue (RGB) and multispectral nadir and oblique imagery of structurally different hedgerows and used them to generate 3D point clouds and models with SfM workflow. Height thresholding allowed extraction of hedgerow extents, with root‐mean‐square error (RMSE) of height and width ranging from 0.11 to 0.23 m. RGB flower classification showed poor relationship with ground measurements (R2 = 0.31–0.42) due to confusion with woody material of hedgerows. Inclusion of a near‐infrared channel in multispectral imagery significantly improved the relationship (R2 = 0.68–0.75, RMSE = 10%). Our study shows UAV remote sensing has high potential for performing detailed surveys of hedgerows, providing better characterization of structural variations and distribution of flowers than traditional ground surveys due to larger coverage. More comprehensive understanding of hedgerow, or other vegetated buffer strips, conditions offered by UAV surveys can enable better informed decisions on habitat management and biodiversity conservation in rural areas. Acquisitions over larger areas, potentially integrated with satellite remote sensing, can allow assessment of hedgerow connectivity over farm to landscape scales, contributing to better understanding of the hedgerow network and its role as a wildlife corridor.

Preliminary Study on Interpretation of Geological Hazards in Jiuzhaigou Based on Multi-source Remote Sensing Data

A 7.0-magnitude earthquake occurred in Jiuzhaigou County, Sichuan Province at 21:19:46 on August 8, 2017. The epicenter was located in the core scenic spot of Jiuzhaigou. The earthquake triggered thousands of rockfalls and landslides, causing serious damage to scenic landscape such ad Panda Sea and also to infrastructure and roads in Jiuzhaigou. The geological disasters in vegetational cover and steep mountainous areas are difficult to identify. Traditional ground surveys and optical remote sensing are not very applicable. This brought potential risks to scenic reconstruction and natural landscape restoration. The airborne LiDAR data and satellite optical image were used to carry out the interpretation of Jiuzhaigou geological disasters after the earthquake, and the recommended values of the laser point cloud sampling density of the geological disasters in the high vegetation cover mountain area were given. Based on the regional geological data, formation lithology and fault structure, we used optical image and digital elevation model to build disaster 3D interpretation sign. Therefore, the historical geological disasters, co-seismic geological disasters and distribution characteristics of disaster hazards in Jiuzhaigou scenic spot have been obtained, which provides technical support for the restoration and reconstruction of scenic spots.

Using cameras to index waterfowl abundance in winter-flooded rice fields

Extensive wetland habitat loss across the continental United States has caused post-harvested rice fields to become an important surrogate wetland habitat for migratory waterfowl. Flooded rice fields used by waterfowl have the potential to provide agronomic benefits to soil. Increasing interest in the reciprocal relationship between birds and flooded rice fields has given rise to many studies that aim to quantify bird abundance. However, surveying large flocks of birds in open agricultural fields is challenging because traditional ground and aerial surveys can cause birds to flush or re-allocate spatially, thus biasing counts that are reflected in following management practice recommendations. To avoid this, we used camera surveys and an open-access image manipulation program to estimate 24-h bird use of rice fields. Indices of bird abundance from counts were used to estimate fecal matter input to rice fields. Camera surveys have the potential to limit biases seen in other methods because of their ability to capture bird use over a 24-h period over an entire season and the ability for multiple researchers to survey the same site.•Surveying bird flocks by traditional ground or aerial surveys can bias bird abundance estimates.•Camera surveys of waterfowl in rice fields were used to estimate bird abundance and fecal matter input.•Camera surveys reflect static bird use over 24-h which can lower bias seen in traditional methods.

Mapping out bare-nosed wombat (Vombatus ursinus) burrows with the use of a drone

BackgroundWombats are large, nocturnal herbivores that build burrows in a variety of habitats, including grassland communities, and can come into conflict with people. Counting the number of active burrows provides information on the local distribution and abundance of wombats and could prove to be an important management tool to monitor population numbers over time. We compared traditional ground surveys and a new method employing drones, to determine if drones could be used to effectively identify and monitor bare-nosed wombat burrows.ResultsWe surveyed burrows using both methods in eight 5-ha transects in grassland, that was interspersed with patches of tussock grassland. Ground surveys were conducted by systematically walking transects and searching for burrows. Drone surveys involved programming flights over transects to capture multiple images, from which an orthomosaic image of each transect was produced. These were subsequently viewed using ArcMap to detect burrows. A total of 204 individual burrows were recorded by drone and/or ground survey methods. In grassland, the methods were equally effective in terms of the numbers of burrows detected in transects. In the smaller areas of tussock grassland, ground surveys detected significantly more burrows, because burrow openings were obscured in orthomosaic images by overhanging grasses. There was agreement between the methods as to whether burrows were potentially active or inactive for most burrows in both vegetation communities. However, image interpretation tended to classify grassland burrows as potentially active. Overall time taken to conduct surveys was similar for both methods, but ground surveys utilised three observers and more time in the field.ConclusionsDrones provide an effective means to survey bare-nosed wombat burrows that are visible from the air, particularly in areas not accessible to observers and vehicles. Furthermore, drones provide alternative options for monitoring burrows at the landscape level, and for monitoring wombat populations based on observable changes in burrow appearance over time.

Estimation of DBH at Forest Stand Level Based on Multi-Parameters and Generalized Regression Neural Network

The diameter at breast height (DBH) is an important factor used to estimate important forestry indices like forest growing stock, basal area, biomass, and carbon stock. The traditional DBH ground surveys are time-consuming, labor-intensive, and expensive. To reduce the traditional ground surveys, this study focused on the prediction of unknown DBH in forest stands using existing measured data. As a comparison, the tree age was first used as the only independent variable in establishing 13 kinds of empirical models to fit the relationship between the age and DBH of the forest subcompartments and predict DBH growth. Second, the initial independent variables were extended to 19 parameters, including 8 ecological and biological factors and 11 remote sensing factors. By introducing the Spearman correlation analysis, the independent variable parameters were dimension-reduced to satisfy very significant conditions (p ≤ 0.01) and a relatively large correlation coefficient (r ≥ 0.1). Finally, the remaining independent variables were involved in the modeling and prediction of DBH using a multivariate linear regression (MLR) model and generalized regression neural network (GRNN) model. The (root-mean-squared errors) RMSEs of MLR and GRNN were 1.9976 cm and 1.9655 cm, respectively, and the R2 were 0.6459 and 0.6574 respectively, which were much better than the values for the 13 traditional empirical age–DBH models. The use of comprehensive factors is beneficial to improving the prediction accuracy of both the MLR and GRNN models. Regardless of whether remote sensing image factors were included, the experimental results produced by GRNN were better than MLR. By synthetically introducing ecological, biological, and remote sensing factors, GRNN produced the best results with 1.4688 cm in mean absolute error (MAE), 13.78% in MAPE, 1.9655 cm for the RMSE, 0.6574 for the R2, and 0.0810 for the Theil’s inequality coefficient (TIC), respectively. For modeling and prediction based on more complex tree species and a wider range of samples, GRNN is a desirable model with strong generalizability.

Mapping landscape-scale peatland degradation using airborne lidar and multispectral data

ContextAn increased interest in the restoration of peatlands for delivering multiple benefits requires a greater understanding of the extent and location of natural and artificial features that contribute to degradation.ObjectivesWe assessed the utility of multiple, fine-grained remote sensing datasets for mapping peatland features and associated degraded areas at a landscape-scale. Specifically, we developed an integrated approach to identify and quantify multiple types of peatland degradation including: anthropogenic drainage ditches and peat cuttings; erosional gullies and bare peat areas.MethodsAirborne LiDAR, CASI and aerial image datasets of the South West UK, were combined to identify features within Dartmoor National Park peatland area that contribute to degradation. These features were digitised and quantified using ArcGIS before appropriate buffers were applied to estimate the wider ecohydrologically affected area.ResultsUsing fine-scale, large-extent remotely sensed data, combined with aerial imagery enabled key features within the wider expanse of peatland to be successfully identified and mapped at a resolution appropriate to future targeted restoration. Combining multiple datasets increased our understanding of spatial distribution and connectivity within the landscape. An area of 29 km2 or 9.2% of the Dartmoor peatland area was identified as significantly and directly ecohydrologically degraded.ConclusionsUsing a combination of fine-grained remotely sensed datasets has advantages over traditional ground survey methods for identification and mapping of anthropogenic and natural erosion features at a landscape scale. The method is accurate, robust and cost-effective particularly given the remote locations and large extent of these landscapes, facilitating effective and targeted restoration planning, management and monitoring.

Detecting daytime and nighttime land surface temperature anomalies using thermal infrared remote sensing in Dandong geothermal prospect

Traditional ground survey methods have limited the development and use of renewable energy from geothermal resources. In this paper, the geothermal potential area of Dandong, China is studied using the thermal infrared (TIR) remote sensing data from daytime Landsat 8 and nighttime Advanced Spaceborne Thermal Emission and Reflection Radiometer (ASTER). Through pre-processing such as georeferencing, radiometric calibration, and atmospheric correction, the Landsat8 TIRS and ASTER data were used to invert the land surface temperature of the study area during the daytime and nighttime using the single-channel algorithm and temperature and emissivity separation algorithm. Furthermore, the land surface temperatures during daytime and nighttime of three natural land features—water, vegetation, and bare soil—were classified and analyzed. According to the results, vegetation and bare soil show relatively thermal anomalies during the day and relatively cold anomalies during the night. Conversely, water shows relatively cold anomalies during the day and relatively thermal anomalies during the night. Calculating the daytime and nighttime mean of land surface temperature (DNMLST) can eliminate the relatively cold/thermal anomalies exhibited by natural land features during the daytime and nighttime while highlighting the geothermal anomaly zone. Nine geothermal anomaly zones were identified using the threshold method. The auxiliary analysis of geological data excluded non-geothermal effects. Studies have shown that the distribution of the identified geothermal prospects is consistent with the development and distribution of faults. The fractures cut the land surface, causing groundwater to form structural fissure hot water through fracture structures and fissures after magma heating. The use of two higher-resolution TIR remote sensing data products to obtain the DNMLST for the detection of geothermal anomalies has proven cost-effective technical method.

Topographic constrained land cover classification in mountain areas using fully convolutional network

ABSTRACT Mountains are an important kind of landform on the earth’s surface. Due to harsh mountainous environment, such as steep slopes and cliffs, remote sensing has become an indispensable tool for surveying mountain areas instead of traditional ground surveys. However, the accuracy of current land cover products derived from remote sensing in mountain areas is still low. In this paper, we propose a three-level architecture for land cover classification in mountain areas. Topographic partitioning is first performed in order to partition a large area into several smaller zones, and then, multiresolution segmentation is implemented in each individual zone. Thus, we can obtain initial geo-semantic objects with terrain, spectrum and texture homogeneities. A fully convolutional network (FCN)-based classifier (U-Net) is further introduced for supervised classification of land cover. From the perspectives of both visual interpretation and quantitative evaluation, the proposed method achieved robust and high-precision results for all land cover types. We also investigate the contributions of multimodal features for classification accuracy improvement. First, the results showed that additional features resulted in higher classification accuracies than 3-features only; 6-features achieved the best performance on farmland, impervious surfaces and coniferous forests, while 5-features performed well on water and broad-leaved forests. The elevation feature did not have a positive effect on water and broad-leaved forests, which can be explained by their physical distribution in the landscape. Second, the most significant improvement was achieved on water (Kappa coefficient increased from 0.741 to 0.924), followed by coniferous forests (Kappa coefficient increased from 0.629 to 0.805), whereas only a minor improvement was observed for the other three types. Furthermore, the accuracies of farmland and impervious surfaces remained relatively high even without the assistance of additional features, and texture feature plays a key role. The final land cover map was generated by combining the optimal results of each type via a hierarchical integrating strategy. The overall accuracy of classification achieved 90.6%.