Land cover classification in a mixed forest-grassland ecosystem using LResU-net and UAV imagery

Using an unmanned aerial vehicle (UAV) paired with image semantic segmentation to classify land cover within natural vegetation can promote the development of forest and grassland field. Semantic segmentation normally excels in medical and building classification, but its usefulness in mixed forest-grassland ecosystems in semi-arid to semi-humid climates is unknown. This study proposes a new semantic segmentation network of LResU-net in which residual convolution unit (RCU) and loop convolution unit (LCU) are added to the U-net framework to classify images of different land covers generated by UAV high resolution. The selected model enhanced classification accuracy by increasing gradient mapping via RCU and modifying the size of convolution layers via LCU as well as reducing convolution kernels. To achieve this objective, a group of orthophotos were taken at an altitude of 260 m for testing in a natural forest-grassland ecosystem of Keyouqianqi, Inner Mongolia, China, and compared the results with those of three other network models (U-net, ResU-net and LU-net). The results show that both the highest kappa coefficient (0.86) and the highest overall accuracy (93.7%) resulted from LResU-net, and the value of most land covers provided by the producer’s and user’s accuracy generated in LResU-net exceeded 0.85. The pixel-area ratio approach was used to calculate the real areas of 10 different land covers where grasslands were 67.3%. The analysis of the effect of RCU and LCU on the model training performance indicates that the time of each epoch was shortened from U-net (358 s) to LResU-net (282 s). In addition, in order to classify areas that are not distinguishable, unclassified areas were defined and their impact on classification. LResU-net generated significantly more accurate results than the other three models and was regarded as the most appropriate approach to classify land cover in mixed forest-grassland ecosystems.

Multiscale Fusion Transformer Network for Hyperspectral Image Classification

Convolutional neural network(CNN)has excellent ability to model locally contextual information.However,CNNs face challenges for descripting long-range semantic features,which will lead to relatively low classification accuracy of hyperspectral images.To address this problem,this article proposes an algorithm based on multiscale fusion and transformer network for hyperspectral image classification.Firstly,the low-level spatial-spectral features are extracted by multi-scale residual structure.Secondly,an attention module is introduced to focus on the more important spatialspectral information.Finally,high-level semantic features are represented and learned by a token learner and an improved transformer encoder.The proposed algorithm is compared with six classical hyperspectral classification algorithms on real hyperspectral images.The experimental results show that the proposed algorithm effectively improves the land cover classification accuracy of hyperspectral images.

Land use and cover change and influencing factor analysis in the Shiyang River Basin,China

Land use and cover change(LUCC)is the most direct manifestation of the interaction between anthropological activities and the natural environment on Earth's surface,with significant impacts on the environment and social economy.Rapid economic development and climate change have resulted in significant changes in land use and cover.The Shiyang River Basin,located in the eastern part of the Hexi Corridor in China,has undergone significant climate change and LUCC over the past few decades.In this study,we used the random forest classification to obtain the land use and cover datasets of the Shiyang River Basin in 1991,1995,2000,2005,2010,2015,and 2020 based on Landsat images.We validated the land use and cover data in 2015 from the random forest classification results(this study),the high-resolution dataset of annual global land cover from 2000 to 2015(AGLC-2000-2015),the global 30 m land cover classification with a fine classification system(GLC_FCS30),and the first Landsat-derived annual China Land Cover Dataset(CLCD)against ground-truth classification results to evaluate the accuracy of the classification results in this study.Furthermore,we explored and compared the spatiotemporal patterns of LUCC in the upper,middle,and lower reaches of the Shiyang River Basin over the past 30 years,and employed the random forest importance ranking method to analyze the influencing factors of LUCC based on natural(evapotranspiration,precipitation,temperature,and surface soil moisture)and anthropogenic(nighttime light,gross domestic product(GDP),and population)factors.The results indicated that the random forest classification results for land use and cover in the Shiyang River Basin in 2015 outperformed the AGLC-2000-2015,GLC_FCS30,and CLCD datasets in both overall and partial validations.Moreover,the classification results in this study exhibited a high level of agreement with the ground truth features.From 1991 to 2020,the area of bare land exhibited a decreasing trend,with changes primarily occurring in the middle and lower reaches of the basin.The area of grassland initially decreased and then increased,with changes occurring mainly in the upper and middle reaches of the basin.In contrast,the area of cropland initially increased and then decreased,with changes occurring in the middle and lower reaches.The LUCC was influenced by both natural and anthropogenic factors.Climatic factors and population contributed significantly to LUCC,and the importance values of evapotranspiration,precipitation,temperature,and population were 22.12%,32.41%,21.89%,and 19.65%,respectively.Moreover,policy interventions also played an important role.Land use and cover in the Shiyang River Basin exhibited fluctuating changes over the past 30 years,with the ecological environment improving in the last 10 years.This suggests that governance efforts in the study area have had some effects,and the government can continue to move in this direction in the future.The findings can provide crucial insights for related research and regional sustainable development in the Shiyang River Basin and other similar arid and semi-arid areas.

Diversity-accuracy assessment of multiple classifier systems for the land cover classification of the Khumbu region in the Himalayas

Land cover classification of mountainous environments continues to be a challenging remote sensing problem,owing to landscape complexities exhibited by the region.This study explored a multiple classifier system(MCS)approach to the classification of mountain land cover for the Khumbu region in the Himalayas using Sentinel-2 images and a cloud-based model framework.The relationship between classification accuracy and MCS diversity was investigated,and the effects of different diversification and combination methods on MCS classification performance were comparatively assessed for this environment.We present ten MCS models that implement a homogeneous ensemble approach,using the high performing Random Forest(RF)algorithm as the selected classifier.The base classifiers of each MCS model were developed using different combinations of three diversity techniques:(1)distinct training sets,(2)Mean Decrease Accuracy feature selection,and(3)‘One-vs-All’problem reduction.The base classifier predictions of each RFMCS model were combined using:(1)majority vote,(2)weighted argmax,and(3)a meta RF classifier.All MCS models reported higher classification accuracies than the benchmark classifier(overall accuracy with 95% confidence interval:87.33%±0.97%),with the highest performing model reporting an overall accuracy(±95% confidence interval)of 90.95%±0.84%.Our key findings include:(1)MCS is effective in mountainous environments prone to noise from landscape complexities,(2)problem reduction is indicated as a stronger method over feature selection in improving the diversity of the MCS,(3)although the MCS diversity and accuracy have a positive correlation,our results suggest this is a weak relationship for mountainous classifications,and(4)the selected diversity methods improve the discriminability of MCS against vegetation and forest classes in mountainous land cover classifications and exhibit a cumulative effect on MCS diversity for this context.

Effects of RapidEye Imagery's Red-edge Band and Vegetation Indices on Land Cover Classification in an Arid Region

Land cover classification(LCC) in arid regions is of great significance to the assessment, prediction, and management of land desertification. Some studies have shown that the red-edge band of RapidE ye images was effective for vegetation identification and could improve LCC accuracy. However, there has been no investigation of the effects of RapidE ye images' red-edge band and vegetation indices on LCC in arid regions where there are spectrally similar land covers mixed with very high or low vegetation coverage information and bare land. This study focused on a typical inland arid desert region located in Dunhuang Basin of northwestern China. First, five feature sets including or excluding the red-edge band and vegetation indices were constructed. Then, a land cover classification system involving plant communities was developed. Finally, random forest algorithm-based models with different feature sets were utilized for LCC. The conclusions drawn were as follows: 1) the red-edge band showed slight contribution to LCC accuracy; 2) vegetation indices had a significant positive effect on LCC; 3) simultaneous addition of the red-edge band and vegetation indices achieved a significant overall accuracy improvement(3.46% from 86.67%). In general, vegetation indices had larger effect than the red-edge band, and simultaneous addition of them significantly increased the accuracy of LCC in arid regions.

Integrating TM and Ancillary Geographical Data with Classification Trees for Land Cover Classification of Marsh Area

The main objective of this research is to determine the capacity of land cover classification combining spec- tral and textural features of Landsat TM imagery with ancillary geographical data in wetlands of the Sanjiang Plain, Heilongjiang Province, China. Semi-variograms and Z-test value were calculated to assess the separability of grey-level co-occurrence texture measures to maximize the difference between land cover types. The degree of spatial autocorrelation showed that window sizes of 3×3 pixels and 11×11 pixels were most appropriate for Landsat TM im- age texture calculations. The texture analysis showed that co-occurrence entropy, dissimilarity, and variance texture measures, derived from the Landsat TM spectrum bands and vegetation indices provided the most significant statistical differentiation between land cover types. Subsequently, a Classification and Regression Tree (CART) algorithm was applied to three different combinations of predictors: 1) TM imagery alone (TM-only); 2) TM imagery plus image texture (TM+TXT model); and 3) all predictors including TM imagery, image texture and additional ancillary GIS in- formation (TM+TXT+GIS model). Compared with traditional Maximum Likelihood Classification (MLC) supervised classification, three classification trees predictive models reduced the overall error rate significantly. Image texture measures and ancillary geographical variables depressed the speckle noise effectively and reduced classification error rate of marsh obviously. For classification trees model making use of all available predictors, omission error rate was 12.90% and commission error rate was 10.99% for marsh. The developed method is portable, relatively easy to im- plement and should be applicable in other settings and over larger extents.

Classification and Gradation of Cultivated Land Quality in Bishan County of Chongqing, China

The conflicts among food security, economic development and ecological protection are the “sticking point” of undeveloped southwestern mountainous areas of China. The objectives of this study are to identify appropriate inte- grated indicators influencing the classification and gradation of cultivated land quality in the southwestern mountainous area of China based on semi-structure interview, and to promote the monitoring of cultivated land quality in this region. Taking Bishan County of Chongqing as a study case, the integrated indicators involve the productivity, protection, ac- ceptability, and stability of cultivated land. The integrated indicators accord with the characteristics of land resources and human preference in southwestern mountainous area of China. In different agricultural zones, we emphasize different indicators, such as emphasizing productivity, stabilization and acceptability in low hilly and plain agricultural integrative zone (LHP-AIZ), protection, productivity and stability in low mountain and hill agro-forestry ecological zone (LMH-AEZ), and acceptability in plain outskirts integrative agricultural zone (PO-IAZ), respectively. The pronounced difference of classification and gradation of cultivated land, regardless of inter-region or intra-region, is observed, with the reducible rank from PO-IAZ, LHP-AIZ to LMH-AEZ. Research results accord with the characteristics of assets management and intensive utilization of cultivated land resources in the southwestern mountainous area of China. Semi-structure interview adequately presents the principal agent of farmers in agricultural land use and rural land market. This method is very effective and feasible to obtain data of the quality of cultivated land in the southwestern mountainous area of China.

Land Cover Classification with Multi-source Data Using Evidential Reasoning Approach

Land cover classification is the core of converting satellite imagery to available geographic data.However,spectral signatures do not always provide enough information in classification decisions.Thus,the application of multi-source data becomes necessary.This paper presents an evidential reasoning (ER) approach to incorporate Landsat TM imagery,altitude and slope data.Results show that multi-source data contribute to the classification accuracy achieved by the ER method,whereas play a negative role to that derived by maximum likelihood classifier (MLC).In comparison to the results derived based on TM imagery alone,the overall accuracy rate of the ER method increases by 7.66% and that of the MLC method decreases by 8.35% when all data sources (TM plus altitude and slope) are accessible.The ER method is regarded as a better approach for multi-source image classification.In addition,the method produces not only an accurate classification result,but also the uncertainty which presents the inherent difficulty in classification decisions.The uncertainty associated to the ER classification image is evaluated and proved to be useful for improved classification accuracy.

Evaluation of a deep-learning model for multispectral remote sensing of land use and crop classification

High-resolution deep-learning-based remote-sensing imagery analysis has been widely used in land-use and crop-classification mapping. However, the influence of composite feature bands, including complex feature indices arising from different sensors on the backbone, patch size, and predictions in transferable deep models require further testing. The experiments were conducted in six sites in Henan province from2019 to 2021. This study sought to enable the transfer of classification models across regions and years for Sentinel-2 A(10-m resolution) and Gaofen PMS(2-m resolution) imagery. With feature selection and up-sampling of small samples, the performance of UNet++ architecture on five backbones and four patch sizes was examined. Joint loss, mean Intersection over Union(m Io U), and epoch time were analyzed, and the optimal backbone and patch size for both sensors were Timm-Reg Net Y-320 and 256 × 256, respectively. The overall accuracy and Fscores of the Sentinel-2 A predictions ranged from 96.86% to 97.72%and 71.29% to 80.75%, respectively, compared to 75.34%–97.72% and 54.89%–73.25% for the Gaofen predictions. The accuracies of each site indicated that patch size exerted a greater influence on model performance than the backbone. The feature-selection-based predictions with UNet++ architecture and upsampling of minor classes demonstrated the capabilities of deep-learning generalization for classifying complex ground objects, offering improved performance compared to the UNet, Deeplab V3+, Random Forest, and Object-Oriented Classification models. In addition to the overall accuracy, confusion matrices,precision, recall, and F1 scores should be evaluated for minor land-cover types. This study contributes to large-scale, dynamic, and near-real-time land-use and crop mapping by integrating deep learning and multi-source remote-sensing imagery.

Fully Polarimetric Land Cover Classification Based on Markov Chains

A novel land cover classification procedure is presented utilizing the information content of fully polarimetric SAR images. The Cameron coherent target decomposition (CTD) is employed to characterize land cover pixel by pixel. Cameron’s CTD is employed since it provides a complete set of elementary scattering mechanisms to describe the physical properties of the scatterer. The novelty of the proposed land classification approach lies on the fact that the features used for classification are not the types of the elementary scatterers themselves, but the way these types of scatterers alternate from pixel to pixel on the SAR image. Thus, transition matrices that represent local Markov models are used as classification features for land cover classification. The classification rule employs only the most important transitions for decision making. The Frobenius inner product is employed as similarity measure. Ten different types of land cover are used for testing the proposed method. In this aspect, the classification performance is significantly high.

Integration of SAR Polarimetric Features and Multi-spectral Data for Object-Based Land Cover Classification

An object-based approach is proposed for land cover classification using optimal polarimetric parameters.The ability to identify targets is effectively enhanced by the integration of SAR and optical images.The innovation of the presented method can be summarized in the following two main points:①estimating polarimetric parameters(H-A-Alpha decomposition)through the optical image as a driver;②a multi-resolution segmentation based on the optical image only is deployed to refine classification results.The proposed method is verified by using Sentinel-1/2 datasets over the Bakersfield area,California.The results are compared against those from pixel-based SVM classification using the ground truth from the National Land Cover Database(NLCD).A detailed accuracy assessment complied with seven classes shows that the proposed method outperforms the conventional approach by around 10%,with an overall accuracy of 92.6%over regions with rich texture.

Improving the Interpretability and Reliability of Regional Land Cover Classification by U-Net Using Remote Sensing Data

The accurate and reliable interpretation of regional land cover data is very important for natural resource monitoring and environmental assessment.At present,refined land cover data are mainly obtained by manual visual interpretation,which has the problems of heavy workload and inconsistent interpretation scales.Deep learning has greatly improved the automatic processing and analysis of remote sensing data.However,the accurate interpretation of feature information from massive datasets remains a difficult problem in wide regional land cover classification.To improve the efficiency of deep learning-based remote sensing image interpretation,we selected multisource remote sensing data,assessed the interpretability of the U-Net model based on surface spatial scenes with different levels of complexity,and proposed a new method of stereoscopic accuracy verification(SAV)to evaluate the reliability of the classification result.The results show that classification accuracy is more highly correlated with terrain and landscape than with other factors related to image data,such as platform and spatial resolution.As the complexity of surface spatial scenes increases,the accuracy of the classification results mainly shows a fluctuating declining trend.We also find the distribution characteristics from the SAV evaluation results of different land cover types in each surface spatial scene.Based on the results observed in this study,we consider the distinction of interpretability and reliability in diverse ground object types and design targeted classification strategies for different surface scenes,which can greatly improve the classification efficiency.The key achievement of this study is to provide the theoretical basis for remote sensing information analysis and an accuracy evaluation method for regional land cover classification,and the proposed method can help improve the likelihood that intelligent interpretation can replace manual acquisition.

Land classification for choice of tree species on farm lands in the Attock District of Punjab, Pakistan

In Pakistan, particularly in Punjab Province, it is difficult for agrofarmers to combine their indigenous knowledge and modern scientific methods to evaluate existing traditional farming systems and forestry practices. This requires an evaluation of indigenous soil classification in simple terms along with knowledge of the local flora, especially trees. This study focuses on land suitability classification for trees in the Attock District of Punjab, Pakistan. A survey was conducted which included interviews of local agrofarmers as well as standard soil analyses including both chemical and physical determinations of local soil types. An evaluation of soil types for cultivation of various crops was carried out given its total extent, component soil series and their proportions, spotting characteristics of each soil series and their major limitations/hazards for trees/crops. These would lead to the identification of various tree species according to soil characteristics. Then, according to the soil types and species, a land suitability map was obtained for the choice of tree species by using geographic information system （GIS） software. Land suitability classification will help local agroforesters/agrofarmers in matching suitable agricultural trees/crops properly for different soils in the area.

The Comparison of the Forms of Land Capability Classification of Atalay and USA in Eskişehir Province (Turkey)

Turkey is an area where climate changes immediately, vegetation, or land gets different in a short distance. Geological and lithological features show diversity. Also, our country’s territorial existence and diversity also bring about different land use conditions. Therefore, land capability also differs from each other. Nevertheless, the classification of land capability used in Turkey is the classification of land capability for agricultural lands prepared by the United States (USA) in 1961. Due to this, [1] have made suggestion on a new classification of land capability considering our country’s geographical conditions. In this study, comparing the land capability with the classification carrying out in our country, the classification which Atalay and Gündüzo&#287lu suggested, has been aimed. Working method has been established according to regional approach and field observations have been done. In preparing the cartographical material, ArcGIS 10.3 has been used. The map of this study as a material topography, physical map, slope, aspect, the usage of the land, ground, geology, land capability, geomorphology, temperature, and precipitation has been examined, meteorological data have been appreciated. According to the findings attained, Eski&#351ehir’s map of land capability has been done through the criteria of the suggestions of Atalay and Gündüzo&#287lu. As a result, it has been understood that there is a difference between the USA land capability that applied in Eski&#351ehir and Atalay and Gündüzo&#287lu’s criteria. In the study, it is suggested to determine the land capabilities again considering the ecological conditions of Turkey.

Simulated Annealing for Land Cover Classification in PolSAR Images

Simulated Annealing (SA) is used in this work as a global optimization technique applied in discrete search spaces in order to change the characterization of pixels in a Polarimetric Synthetic Aperture Radar (PolSAR) image which have been classified with different label than the surrounding land cover type. Accordingly, Land Cover type classification is achieved with high reliability. For this purpose, an energy function is employed which is minimized by means of SA when the false classified pixels are correctly labeled. All PolSAR pixels are initially classified using 9 specifically selected types of land cover by means of Google Earth maps. Each Land Cover Type is represented by a histogram of the 8 Cameron’s elemental scatterers by means of coherent target decomposition (CTD). Each PolSAR pixel is categorized according to the local histogram of the elemental scatterers. SA is applied in the discreet space of nine land cover types. Classification results prove that the Simulated Annealing approach used is very successful for correctly separating regions with different Land Cover Types.

Analysis on the Application of Quality Classification of Cultivated Land Resources in Municipal Land Space Planning:A Case Study of Chongzuo City

Cultivated land is the most important strategic resource to ensure food security.The newly constructed quality classification system of cultivated land resources considers the cultivated land health index for the first time.How the new classification and grading index system and the quality classification results of cultivated land resources to effectively guide the preparation of municipal land space planning has become a key research direction.This paper expounds the overall design idea for quality classification of cultivated land resources and classification index system.Taking Chongzuo City as an example,through the analysis of the quality classification results of cultivated land resources in the study area,using GIS spatial analysis and factor pairwise comparison method,this paper explores the application ideas and methods of quality classification research results of cultivated land resources in the formulation of cultivated land retention target,the delineation of dominant areas of cultivated land protection,the delineation of three control lines,the comprehensive improvement of land,and ecological restoration zoning in the municipal land space planning.

Fully Polarimetric Land Cover Classification Based on Hidden Markov Models Trained with Multiple Observations

A land cover classification procedure is presented utilizing the information content of fully polarimetric SAR images. The Cameron coherent target decomposition (CTD) is employed to characterize each pixel, using a set of canonical scattering mechanisms in order to describe the physical properties of the scatterer. The novelty of the proposed classification approach lies on the use of Hidden Markov Models (HMM) to uniquely characterize each type of land cover. The motivation to this approach is the investigation of the alternation between scattering mechanisms from SAR pixel to pixel. Depending on the observations-scattering mechanisms and exploiting the transitions between the scattering mechanisms we decide upon the HMM-land cover type. The classification process is based on the likelihood of observation sequences been evaluated by each model. The performance of the classification approach is assessed my means of fully polarimetric SLC SAR data from the broader area of Vancouver, Canada and was found satisfactory, reaching a success from 87% to over 99%.

Assessing the Impact of Using Different Land Cover Classification in Regional Modeling Studies for the Manaus Area,Brazil

Land cover classification is one of the main components of the modern weather research and forecasting models, which can influence the meteorological variable, and in turn the concentration of air pollutants. In this study the impact of using two traditional land use classifications, the United States Geological Survey (USGS) and the Moderate-resolution Imaging Spectroradiometer (MODIS), were evaluated. The Weather Research and Forecasting model (WRF, version 3.2.1) was run for the period 18 - 22 August, 2014 (dry season) at a grid spacing of 3 km centered on the city of Manaus. The comparison between simulated and ground-based observed data revealed significant differences in the meteorological fields, for instance, the temperature. Compared to USGS, MODIS classification showed better skill in representing observed temperature for urban areas of Manaus, while the two files showed similar results for nearby areas. The analysis of the files suggests that the better quality of the simulations favorable to the MODIS file is straightly related to its better representation of urban class of land use, which is observed to be not adequately represented by USGS.

Application of GIS in Ecological Land Type(ELT)mapping--A case in Changbai Mountain area

This paper depicted the physiographic landscape features and natural vegetation situation of study area (the eastern Jilin Province), and expatiates the definition, basic characters and its development of Ecological Land Classification (ELC). Based on the combination of relief map, satellite photography for study area and vegetation inventory data of 480 sample sites, a 5-class and a 15-class ecological land type map was concluded according to 4 important factors including slope, aspect, vegetation and elevation. Ecological Classification System (ECS) is a method to identify, characterize, and map ecosystems. The Ecological Land Type (ELT) was examined and applied initially in eastern Jilin Province.

Crop Classification Using MODIS NDVI Data Denoised by Wavelet: A Case Study in Hebei Plain, China

Time-series Moderate Resolution Imaging Spectroradiometer (MODIS) Normalized Difference Vegetation Index (NDVI) data have been widely used for large area crop mapping.However,the temporal crop signatures generated from these data were always accompanied by noise.In this study,a denoising method combined with Time series Inverse Distance Weighted (T-IDW) interpolating and Discrete Wavelet Transform (DWT) was presented.The detail crop planting patterns in Hebei Plain,China were classified using denoised time-series MODIS NDVI data at 250 m resolution.The denoising approach improved original MODIS NDVI product significantly in several periods,which may affect the accuracy of classification.The MODIS NDVI-derived crop map of the Hebei Plain achieved satisfactory classification accuracies through validation with field observation,statistical data and high resolution image.The field investigation accuracy was 85% at pixel level.At county-level,for winter wheat,there is relatively more significant correlation between the estimated area derived from satellite data with noise reduction and the statistical area (R2 = 0.814,p < 0.01).Moreover,the MODIS-derived crop patterns were highly consistent with the map generated by high resolution Landsat image in the same period.The overall accuracy achieved 91.01%.The results indicate that the method combining T-IDW and DWT can provide a gain in time-series MODIS NDVI data noise reduction and crop classification.