It is significant to adopt deep learning algorithms and higher-resolution remote sensing images in mapping large-scale and high-precision of aeolian landform. In this study, the western part of Horqin Sandy Land was taken as the study area. Based on the data collected from 2,786 verification points located in sandy land and remote sensing images of high-spectral and spatial resolution Sentinel-1, Sentinel-2, and GDEM (V3), this article made a research on data of large-scale and high-precision mapping classification of this area between 2015 and 2020 by using convolutional neural network deep learning algorithm. The results showed that the types of aeolian sandy landform in the west of Horqin Sandy Land mainly include longitudinal dune, flat sandy land, mild undulating sand land, nest-shaped land, parabolic dune, barchan dune, and dune chain, with an area of 1735.62, 51.32, 251.38, 902.07, 49.57, and 101.63 km2. Among them, longitudinal dune, barchan dune, and dune chain have the largest area, while parabolic dunes and flat sand land are smaller. Between 2015 and 2020, the area of aeolian landforms was reduced by 89.27 km2 and transformed into an oasis from a desert. This study adopted remote sensing data by high-resolution Sentinel and GDEM (V3) and convolutional neural network deep learning algorithm to map the aeolian landforms effectively. The precision of aeolian landform classification and Kappa coefficient in the western part of Horqin Sandy Land is as high as 95.51% and 0.8961. Combined with Sentinel-1, Sentinel-2, and GDEM (V3), the deep learning algorithm based on the convolution neural network can timely and effectively monitor the changes of sand dunes, which can be used for large-scale aeolian landforms.
Sandy land refers to land covered with sand and less vegetation. It is mainly located in the semi-humid and semi-arid transition zone and is also the transition one between grassland and desert [1,2]. The sandy ecosystem is an essential part of the earth’s ecosystem, and the landscape change of the sandy ecosystem is an important indicator that characterizes the global climate change and the intensity of human activities . Therefore, the precise identification of dune types and areas is of great significance . Dong et al.  visually distinguished various dunes on the surface of Mars based on the dune classification method invented by the French scholar, Mainguet. However, due to current science and technology limitations, visual identification is still the most effective method for identifying types of sandy landforms on other planets. Besides, there are extensive and complex types of aeolian landforms on the surface of the earth, so it takes much energy and time to observe visually. Compared with traditional manual survey methods requiring a long cycle and high labor cost, remote sensing technology can detect a wide range and rapidly acquire data. Therefore, most scholars use computer interpretation methods and data collected by satellite remote sensing data, such as supervised classification, unsupervised classification, object-oriented classification, and other methods [6,7]. In the recent years, convolutional neural network (CNN) deep learning is considered an efficient and accurate deep learning method widely used in image recognition and classification research [8,10]. Timm and Mcgarigal  used the random forest classification algorithm to construct a remote sensing coverage map with a resolution of 1 m pixels based on the fine dunes of the Cape Cod National Coast in the United States. The accuracy of coastal dunes drawing was 75.1%, and the classification was mainly concentrated on similar overlapping dunes. By using decision tree methods to extract land types such as cultivated land, shrubland, and woodland, Wu et al. analyzed the changes in land utilization in Horqin Sandy Land based on Landsat TM/OLI remote sensing images over the past 30 years, and the accuracy of the extraction and classification results have reached more than 88% . Zhao et al. made a study on crop classification using deep learning models such as N-dimensional convolutional neural networks, recurrent neural networks, long- and short-term memory RNNs, and gated recurrent unit RNNs . In this study, Zhao et al. demonstrated the background and significance of using deep learning methods and related advantages, challenges, and limitation of existing solutions. However, previous studies have mainly focused on remote sensing classification of sandy landscape types and vegetation community identification, and there are few studies on remote sensing identification of large-scale sandy landform types [14,15,16,17].
Therefore, in this article, the western part of Horqin Sandy Land was taken as the study area. Based on the data collected from 2,786 verification points located in sandy land and remote sensing images of high-spectral and spatial resolution Sentinel-1, Sentinel-2, and GDEM (V3), this article made a research on data of large-scale and high-precision mapping classification of this area between 2015 and 2020 by using convolutional neural network deep learning algorithm. Based on the application of deep learning in remote sensing classification of sandy aeolian landforms, the research results can provide data support for remote sensing monitoring of desertification and provide a case reference for large-scale and high-precision remote sensing classification and mapping of sandy aeolian landforms in other regions.
2 Data and methods
2.1 Overview of the study area
The western part of Horqin Sandy Land (118°31′40″E–120°45′10″E, 42°39′29″N–43°24′44″N) is located in Wengniute Banner, Inner Mongolia Autonomous Region, with a total area of 5788.63 km2 (Figure 1). The terrain is high in the west and low in the east, and the elevation is between 284 and 909 m. The study area is located in a temperate semi-arid continental climate zone with an annual average temperature is 5.2–6.4°C with an annual average precipitation of 300–400 mm. The precipitation is mainly in summer. The average annual wind speed is 3.5 m/s, and the maximum wind speed is 21.7 m/s. The frost-free period is from 140 to 150 days, and the water heat and strong wind conditions have become the main factors for the formation and development of sandy land. The area lies between the Xar Moron River and the Shaolang River. The soil type is mainly sandy soil. The main vegetation is the large scale of grassland and man-made forest including Ostryopsis davidiana decaisne and Pinus tabuliformis carriere. Nowadays, thanks to the national policies on the protection of grass, sand, tourism, and other related industries, the desertification condition has been effectively improved .
2.2 Data source and processing
Since the successful launch of the Sentinel binary star system in 2014, its higher spectral and spatial resolution image data have been widely used in ground object recognition research in lake and wetland and mountain vegetation [19,20]. Compared with Landsat series data, its resolution has been significantly improved. In this study, it was merged with DEM data to significantly improve the accuracy of remote sensing classification of sandy landforms [21,22].
Sentinel-1 radar images were downloaded from the official website of Google Earth Engine (https://earthengine.google.com), and the VV and VH backscattering coefficient raster images of the western Horqin Sandy Land in 2015 and 2020 were obtained, respectively . Sentinel-1 is an all-weather, all-weather radar imaging system, including two satellites A and B. Sentinel-1 A was successfully launched on April 3, 2014, and then, Sentinel-1B was successfully launched on April 25, 2016. The Sentinel-1C-band-based imaging system uses four imaging modes (with a resolution of up to 5 m and a width of 400 km) for observation. Sentinel-1 operates in a near-polar sun-synchronous orbit. The revisit period of one satellite is 12 days, and observations of two satellites complement each other, and the revisit period is 6 days. The Sentinel-2 multispectral high-resolution remote sensing image (Table 1) was downloaded from the ESA data website (https://glovis.usgs.gov). The Sentinel-2 high-resolution multispectral imaging satellite is divided into two satellites, 2A and 2B. 2A was launched by the “Vega” carrier rocket on June 23, 2015, and 2B was launched by the “Vega” carrier rocket on March 7, 2017. The Sentinel-2 satellite carries a multispectral imager (MSI) with 13 spectral bands. The spatial resolution of bands 2, 3, 4, and 8 is 10 m, and the spatial resolution of bands 5, 6, 7, 8A, 11, and 12 is 20 m. The band 1, 9, and 10 is 60 m, and the revisit period of one satellite is 10 days. The observations of two satellites complement each other, which can form a revisit period of less than 5 days for terrestrial monitoring. The Sentinel image was calibrated based on atmospheric, terrain, and super-resolution fusion by using official processing software SNAP (V8.0) of ESA, and the data cropping, mosaic, and convolutional neural network deep learning classification were completed with ENVI (V5.6) . GDEM (V3) data were downloaded from Geospatial Data Cloud (http://www.gscloud.cn/search) . Based on Sentinel and GDEM (V3) remote sensing images, Sentinel-1VV, VH polarization backscattering coefficients, DEM and Sentinel-2 spectra, vegetation, water bodies, and textures were extracted (Table 2). The study acquired 2,786 survey samples through Google Earth observations and field surveys in October 2020, including information about types of sand dune landform, latitude and longitude (error < 5 m), vegetation and corresponding photos, 464 longitudinal dunes, 342 flat sand lands, 440 nest-shaped dunes, 466 mild undulating sand lands, 230 parabolic dunes, and 724 barchan dune and dune chains, which are randomly allocated as training samples and verification points in a 6:4 ratio.
|Date||Path||UTM||Center longitude||Center latitude||Cloudiness (%)||Size (MB)|
|Feature name||Feature description|
|VV polarization backscattering coefficient||VV|
|VH polarization backscattering coefficient||VH|
|Digital elevation model||DEM|
|Sentinel-2 spectral band||B2, B3, B4, B5, B6, B7, B8, B8a, B11, B12|
|NDVI||Normalized difference vegetation index|
|DVI||Difference vegetation index|
|RVI||Ratio vegetation index|
|SAVI||Soil-adjusted vegetation index|
|NDWI||Normalized difference water index|
|MNDWI||Modified normalized difference water index|
|Maximum probability||Maximum probability|
|Angle second pitch||Angular second moment|
3 Research methods
3.1 Technical process for classification of sand dune landform types
Based on high-resolution Sentinel-1 and Sentinel-2, GDEM (V3) remote sensing images and verification point data of 2,768 aeolian landform fields collected from 2015 to 2020 (Figure 2), this article calibrated the atmosphere and terrain images and obtained waveband characteristics including radar, DEM, visible light, red edge, vegetation index, water index, and texture ; by using convolution neural network deep learning algorithm, this article mapped remote sensing classification of aeolian sand landform in the west of Horqin Sandy Land; the classification results are calibrated by wind direction, vegetation, terrain, and other factors, and the verification points were used for accuracy verification, and the evolution map of the aeolian landforms in the west of the Horqin Sandy Land from 2015 to 2020 is obtained, and the deep learning and high-resolution remote sensing of the aeolian sandy landforms in the west of the Horqin Sandy Land are discussed. The advantages of images combined with information recognition and the evolution process of sandy land.
3.2 Convolutional neural network algorithm
CNN is a deep learning neural network designed for processing high-dimensional grid data . It can start directly from the original pixels and recognize the laws of vision with little preprocessing. The most significant difference between CNN and other deep learning models is that it has the characteristics of shared weights and sparse connectivity. Shared weight can significantly reduce the number of parameters. Sparse connectivity structure can simplify the network calculation and effectively extract primary visual features and spatial information such as directional line segments, inflection points, and endpoints. Compared with classic neural networks, CNN has multiple convolutional layers and pooling layers for calculation .
The convolutional layer is composed of multiple convolution kernels, which are used to calculate different feature maps. Each neuron of the feature map is connected to the area of the adjacent neuron in the previous layer. Such adjacent neurons are called receptive fields of neurons in the previous layer. A new feature map will be obtained by convolving and learning the feature map of the first layer and by applying the nonlinear activation function to the convolution result. In convolution processing, to generate each feature map, the convolution kernel is shared by all spatial positions of the input. The complete feature map is calculated by all the convolution kernels. The formula is expressed as follows:
where is the output image pixel matrix, (…×…) is the nonlinear function, a is the weight matrix, is the input image pixel matrix, n is the number of bands, and is the offset matrix.
The pooling layer, also known as the downsampling layer or subsampling layer, is another core structural layer of CNN. By sampling the space of each dimension of the input data, the data scale can be further reduced, the local linear transformation is invariable with the input data, and the generalization processing of the network is enhanced. Each feature map of the pooling layer is connected to the corresponding feature map at the previous convolutional layer. After multiple convolutions and pooling, the dimensionality of the input features has been reduced to the point where the feed-forward network can be used directly for processing. At the fully connected layer, the feature maps of all two-dimensional images are spliced into one dimension as fully connected input. Each neuron in the fully connected layer is connected to the neuron at the previous layer, but the neurons at the same layer are not connected.
The U-net deep learning model has the advantages of the fast training speed, small training data demand, and maximum fidelity and restoration of original image information. The U-net model under the tensor flow framework is used for deep learning of image features of multiscale object-oriented segmentation. The model follows a typical convolution neural network structure, which forms a symmetrical structure of “shrinking” and “expanding” through multiple convolutions, downsampling, and upsampling can construct high-level feature learningand feature training library. The image is convolved with a 3 × 3 convolution kernel and a modified linear unit activation function, and the downsampling and upsampling are completed, respectively, by the 2 × 2 maximum pooling and matrix cascade methods. At the same time, the convergence speed of the model is controlled by setting the learning rate and dropout to prevent overfitting.
3.3 Accuracy verification
Considering the complexity of the land type in the study area, combined with the verification sample data to make a 64 × 64 pixel size image to form a labeled data set, including three parts: training data set, verification data set and test data set. Among them, the training data set is used to train the U-net deep learning network. The verification data set is used to verify and debug the U-net deep learning network. After the debugging of the U-net deep learning network is completed, the U-net deep learning model is generated, and then, the test data set is used to test the model and evaluate its running results.
By means of confusion matrix, the overall classification accuracy, Kappa coefficient, mapping accuracy, and user accuracy of the classification results are obtained. By comparing extracted category attributes of the corresponding position in the classification with each test pixel’s position and category attributes, the accuracy of classification results is obtained and evaluated . The formula for overall accuracy (OA) and Kappa coefficient is given as follows:
where is a positive sample that the model correctly classifies, is a negative sample correctly classified by the model, is a negative sample that the model incorrectly classifies, and is a positive sample incorrectly classified by the model.
where is the overall classification accuracy, is the number of actual samples for each category, is the number of samples predicted by each category, is the total number of categories, and is the total number of samples.
4 Results and analysis
4.1 Spatial distribution of aeolian sand landform types
The aeolian sandy landform in the west of Horqin Sandy Land mainly includes longitudinal dune, flat sand land, mild undulating sand land, nest-shaped dune, parabolic dune, crescent dune, barchan dune, and dune chain. There are six types of sand dune landform distributed alternately and relative to each other (Figures 3 and 4). The longitudinal dunes are mainly distributed in the middle of the study area, with the largest area of 1735.62 km2 among these six types of sandy dune landform (Table 3); the nest-shaped dunes and barchans dune and dune chains are mainly distributed around the longitudinal dunes, with an area of 902.07 and 110.63 km2, respectively; flat sand land and parabolic dunes are mainly distributed in the west and south of the study area, with an area of 51.32 and 49.57 km2, respectively; mild undulating sand land is mainly distributed in the northeast of the study area, with an area of 251.38 km2.
|Mild undulating sand land||260.35||251.38|
|Flat sand land||34.57||51.32|
|Barchan dune and dune chain||1250.14||1101.63|
4.2 Dynamic evolution of aeolian sand landform types
The aeolian sandy landforms in Horqin Sandy Land are still in the process of continuous evolution. From 2015 to 2020, the areas of nest-shaped dunes, parabolic dunes, and flat sand land increased by 233.34, 18.26, and 16.75 km2, respectively (Figure 5). The increased areas are mainly concentrated in the central, southern, and southwestern parts of the study area. In the same period, the area of mild undulating sand land, longitudinal dune, and barchans dune decreased by 8.97 200.14, and 148.51 km2, respectively. The reduced areas are mainly distributed in the northeast, middle, and west of the study area.
4.3 Accuracy evaluation of classification of aeolian sand landform
With the help of the confusion matrix, the overall classification accuracy, Kappa coefficient, mapping accuracy, and user accuracy of the classification results are obtained, and the classification accuracy is calculated by checking the position of each test pixel and comparing the classification with the classification of the corresponding position in the classification image. The accuracy of the classification results is evaluated .
Based on the sample data of 1,114 ground survey verification points, a confusion matrix analysis was performed on the deep learning remote sensing classification results of the aeolian sand landform types in the western Horqin Sandy Land (Table 4). The results show that the effect of CNN deep learning on aeolian landforms has a significant effect on remote sensing classification. The accuracy and Kappa coefficient are as high as 95.51% and 0.8961, respectively. The mapping accuracy and user accuracy of each dune landform type are above 90%, including barchan dune and dune chain, parabolic dune, and longitudinal dune. The mapping accuracy of sand dunes and sand ridges are 97.58, 96.29, and 95.33%. The user accuracy of barchan dune and dune chain, mild undulating sand land, and longitudinal dune are 96.79, 96.24, and 95.49%, respectively. Therefore, the CNN deep learning algorithm combined with Sentinel high-resolution remote sensing images can significantly improve the remote sensing classification accuracy of sandy landforms.
|Type||Mild undulating sand land||Nest-shaped dune||Parabolic dune||Flat sand land||Longitudinal dune||Barchan dune and dune chain|
|Mild undulating sand land||179||1||0||1||1||2|
|Flat sand land||0||1||1||91||2||2|
|Barchan dune and dune chain of dunes||1||2||1||3||4||241|
|Overall accuracy: 95.51%||Kappa: 0.8961|
5 Discussion and conclusion
The evolution process of sandy land is manifested as the change of land surface landscape pattern, which can be characterized by landscape index . Li et al.  believed that landscape fragmentation and spatial heterogeneity have a positive effect on the desertification process. Li et al.  also proposed that the area of mobile dune patches is positively correlated with the degree of desertification. The Horqin Sandy Land is characterized by a decrease in the total area of aeolian landforms and an increase in fixed and semi-fixed dunes such as shrub-grass dunes, showing the development of deserts in the direction of oasis, which is consistent with the report by Yue et al. . The Horqin sandy landscape is in a dynamic balance under the interactive erosion of wind and water and maintains its development stability. Moisture is the first dominant factor restricting the growth and survival of plants, and it has an important impact on the structure of the regional sand flow field and sand transport . In addition, the construction of reservoirs in the upper reaches of the river, such as the Hongshan Reservoir in the upper reaches of the Laoha River, reaches in the Shanwanzi Reservoir, Gangouzi Reservoir, and Ulan Besu Reservoir built in the middle and lower reaches of the river will inevitably affect the downstream runoff and change the water and sand conditions of the river. If the flow is reduced, the river bed is dry, and the broad floodplain is exposed, the river bank sand dunes can develop. Surface water decreases, groundwater level drops, and the groundwater level of yards and sand marshes will also drop. Wet-loving plants will decrease and drought-tolerant plants will increase. In particular, trees will die, vegetation coverage will be less, sand dunes will be activated, and desertification will be repeated and will continue to expand [34,35]. As a unique geographical unit, the Horqin Sandy Land has an obvious trend of warming and drying. Zhao et al.  also confirmed this point in the study of climate change in the Horqin Sandy Land. Increasing temperature and increased evaporation will lead to a decrease in water volume, degradation of vegetation, and increased desert activities , which plays an important role in inhibiting the evolution of desertification. At the same time, it should be pointed out that short-term and high-intensity human activities can accelerate or delay the evolution of sand dunes on the surface. Wang et al.  surveyed the Horqin area and showed that due to human adjustment of land use structure, there was a need for the implementation of measures such as returning farmland to forests and grasses, restraining overgrazing, and so on. The desertification of the land in this area has the reversed trend.
Based on Sentinel high-resolution remote sensing images, this research applies the CNN deep learning algorithm to the remote sensing classification of aeolian sand landforms and also conducts verification studies in the western part of Horqin Sandy Land. The mapping is accurate, indicating that it is feasible to be extended to the research of remote sensing mapping of aeolian landform types in other regions. Deep learning is currently the most advanced method of remote sensing classification and is widely used. Compared with traditional supervised classification and object-oriented classification, the accuracy is increased by 46% [28,39].
The construction of a deep learning model involves multiple hyperparameters (such as the division ratio of data used for training, verification, and testing of the model, the size of the labeled data set image, the size of the convolution kernel, etc.), and the settings of the model are different, and the performance of the built model is also different [40,41]. In the experiment, the hyperparameter setting of the deep learning model is mainly selected based on the researcher’s experience, which is highly subjective. Experts and scholars have conducted effective research on how to determine the hyperparameters of deep learning models, but their study objects are mostly on cultivated land and forests [42–44], and the studies on the types of sandy landforms are insufficient. Therefore, an in-depth research is urgently needed. Related research work to further improve the accuracy of drawing. The deep learning method in this article can be extended to the high-precision classification of other sandy landforms in arid areas.
From the perspective of methodology, the remote sensing classification and mapping of sandy landforms based on the CNN deep learning algorithm have a significantly positive effect. In the remote sensing mapping of the aeolian sandy landform in western Horqin, the overall accuracy and Kappa coefficient are 95.51% and 0.8961, respectively. This method can be extended to the research of remote sensing mapping of aeolian landforms in other deserts.
The western part of Horqin Sandy Land has various types of sandy landform, including longitudinal dune, flat sandy land, mild undulating sand land, nest-shape land, parabolic dune, barchan dune, and dune chain, with an area of 1,735.62, 51.32, 251.38, 902.07, 49.57, and 1,101.63 km2, respectively. The longitudinal dune, barchan dune, and dune chain area are relatively large, but the parabolic dune and flat sand land areas are relatively small. From 2015 to 2020, the area of sand dunes decreased by 89.27 km2, and the effectiveness of sand protection was remarkable, which shows a transformation of this area from a desert to an oasis.
The experimental scheme of CNN deep learning algorithm combined with Sentinel high-resolution remote sensing images used in remote sensing mapping of wind-sand landforms can be applied to the remote sensing recognition of sand dune types in other areas. The focus of the research is to explore the feasibility of the CNN deep learning algorithm in the classification of sandy landforms and to test it in the western part of Horqin Sandy Land. The experiment focuses on the optimization of image data resolution and computer algorithms. There is still a large space for exploration and application. The next step is to optimize the deep learning parameters of convolution neural networks, introduce long-term images, meteorology (wind direction, temperature, and precipitation), and human activity factors, to carry out remote sensing classification, dynamic evolution, and driving factors of long-time sandy landform types, and to extend it to other decertified areas to conduct remote sensing classification and mapping experimental research to provide data support for remote sensing monitoring of desertification.
This research was funded by National Natural Science Foundation of China (No. 41871022).
Author contributions: Du HS formulated the review and identified the these to be covered. Wang JF and Han C are responsible for data processing.
Conflict of interest: Authors state no conflict of interest.
 Du HS, Hasi ED, Li S, Zhao YY, Sun XX. Landscape evolution and influence of Aeolian sand and lake in Horqin sandy land. Sci Geographica Sin. 2018;38(12):2109–17.Search in Google Scholar
 Xu ZW, Lu HY. Theories and new understandings of the study on the changes of wind-sand environment in the MuUs Sandy Land. Acta Geographica Sin. 2021;76(9):2203–23.Search in Google Scholar
 Liu X, Huang Z, Havrilla CA, Liu Y, Wu GL. Plant litter crust role in nutrients cycling potentials by bacterial communities in a sandy land ecosystem. Land Degrad Dev. 2021;32(11):3194–203.10.1002/ldr.3973Search in Google Scholar
 Dong ZB, Li C, Lu P, Hu GY. Eroded dunes: inspiration from Mars. Adv Earth Sci. 2021;36(2):125–38.Search in Google Scholar
 Thorne K, MacDonald G, Guntenspergen G, Ambrose R, Buffington K, Dugger B, et al. US Pacif-ic coastal wetland resilience and vulnerability to sea-level rise. Sci Adv. 2018;4(2):3270.10.1126/sciadv.aao3270Search in Google Scholar PubMed PubMed Central
 Li J, Liu W, Zhang XX. Evaluation algorithm of alhagi sparsifolia desertification control under different irrigation amounts. Earth Sci Res J. 2020;24(4):449–57.10.15446/esrj.v24n4.91626Search in Google Scholar
 Sun Z., Di L., Fang H., Burgess A. Deep learning classification for crop types in North Dakota. IEEE J Sel Top Appl Earth Observ Remote Sens. 2020;13:2200–13.10.1109/JSTARS.2020.2990104Search in Google Scholar
 Xue L, Liu J, Wen G, Wang H. Efficient, high-resolution topology optimization method based on convolutional neural networks. Front Mech Eng. 2021;16(1):80–96.10.1007/s11465-020-0614-2Search in Google Scholar
 Khoshboresh-Masouleh M, Shah-Hosseini R. Building panoptic change segmentation with the use of uncertainty estimation in squeeze-and-attention CNN and remote sensing observations. Int J Remote Sens. 2021;42(20):7798–280.10.1080/01431161.2021.1966853Search in Google Scholar
 Timm BC, Mcgarigal K. Fine-scale remotely-sensed cover mapping of coastal dune and salt marsh ecosystems at Cape Cod National Seashore using Random Forests. Remote Sens Environ. 2012;127:106–17.10.1016/j.rse.2012.08.033Search in Google Scholar
 Wu YJ, Liu TX, Tong X, Luo YY, Duan LM, Wang GL. Analysis of land use evolution in Horqin Sandy Land based on long-term landsat data. Acta Ecologica Sin. 2020;40(23):8672–82.Search in Google Scholar
 Zhao HW, Chen ZX, Liu J. The application and challenges of deep learning methods in crop remote sensing classification. China Agric Resour Regional Plan. 2020;41(2):35–49.Search in Google Scholar
 Cao J, Leng W, Liu K, Liu L, He Z, Zhu Y. Object-based mangrove species classification using unmanned aerial vehicle Hyperspec-tral images and digital surface models. Remote Sens. 2018;10(1):89.10.3390/rs10010089Search in Google Scholar
 Liu M, Li H, Li L, Man W, Jia M, Wang Z, et al. Monitoring the invasion of Spartina alterniflora using multisource high-resolution imagery in the Zhangjiang Estuary. China Remote Sens. 2017;9(6):539.10.3390/rs9060539Search in Google Scholar
 Ma L, Li M, Gao Y, Chen T, Ma X, Qu L. A novel wrapper approach for feature selection in object-based image classification using polygon-based cross-validation. IEEE Geosci Remote Sens Lett. 2017;14(3):409–13.10.1109/LGRS.2016.2645710Search in Google Scholar
 Xu XC, Li BJ, Liu XP, Li X, Shi Q. Global land cover mapping with 30 m resolution from 2000 to 2015. J Remote Sens. 2021;25(9):1896–916.Search in Google Scholar
 Chen YP, Li XH, Luo YQ, Wang XY, Niu YY. Study on the dynamic characteristics of vegetation in different sandy land types in Horqin and its relationship with litter. Acta Grasslandica. 2021;29(1):114–20.Search in Google Scholar
 Yao JP, Yang LK, Chen T, Song CJ. Research on continuous change monitoring of Poyang Lake wetland based on Sentinel-1,2 and Landsat 8 time series images. Remote Sens Technol Application. 2021;36(4):760–76.Search in Google Scholar
 Yang D, Zhou YN, Yang XZ, Gao LJ, Feng L. Vegetation mapping of Taibai Mountain based on time series Sentinel-1A data supported by LSTM. J Geo-Information Sci. 2020;22(12):2445–55.Search in Google Scholar
 Li Y, Niu L, Zhao QH. Analysis of land use/cover change in Fushun mining area from 1989 to 2019. Sci Surveying Mapp. 2021;46(8):96–104 + 140.Search in Google Scholar
 Xu R, Qu R, Liu XL, Ma HR, Shi ZT. Research on the recognition method and application of impervious surface based on Sentinel-2 data. Geogr Geo-Information Sci. 2021;37(5):31–7.Search in Google Scholar
 Zhang Y, Hartemink AE, Huang J, Townsend PA. Synergistic use of hyperspectral imagery, Sentinel-1 and LiDAR improves mapping of soil physical and geochemical properties at the farm-scale. Eur J Soil Sci. 2021;72(4):1690–717.10.1111/ejss.13086Search in Google Scholar
 Isa SM, Suharjito I, Kusuma GP, Cenggoro TW. Supervised conversion from Landsat-8 images to Sentinel-2 images with deep learning. Eur J Remote Sens. 2021;54(1):182–208.10.1080/22797254.2021.1875267Search in Google Scholar
 Xie Q, Xiao J, Wang D, Yu Y. Analysis of deep-layer and bottom circulations in the South China Sea based on eight quasi-global ocean model outputs. Chin Sci Bull. 2013;58(32):4000–11.10.1007/s11434-013-5791-5Search in Google Scholar
 Kuchma MO, Bloshchinskiy VD. Algorithm for the atmospheric correction of shortwave channels of the MSU-MR radiometer of the Meteor-M No. 2 Satellite. Izvestiya Atmos Ocean Phys. 2020;56(9):909–15.10.1134/S0001433820090145Search in Google Scholar
 Chen X, Zhou GX, Chen WJ. The fruit classification algorithm based on the multi-optimization convolutional neural network. Multimed Tools Appl. 2021;80(7):11313–30.10.1007/s11042-020-10406-6Search in Google Scholar
 Lu CY, Lei YF, Su Y, Huang YF, Liu MY, Jia MM. Dynamic remote sensing analysis of wetland in the low-elevation coastal zone of southeast Fujian based on object-oriented-deep learning. Remote Sens Technol Application. 2021;36(4):713–27.Search in Google Scholar
 Li JY, Chang XL, Cai MY, Zhang JP, Song YH. Analysis of relationship between desertification and landscape structure changes in Horqin Sandland. J Desert Res. 2008;28(4):622–6.Search in Google Scholar
 Wu B, Ci LJ. Temporal and spatial patterns of landscape in the Mu Us Sandland Northern China. Acta Ecologica Sin. 2001;21(2):191–6.Search in Google Scholar
 Wang ML, Bao YH, Altan T, Qin SH. Study on changes of landscape pattern of Hunshandake Sands. J Arid LResour Environ. 2007;21(5):123–7.Search in Google Scholar
 Yue XY, Hou MZ, Chang XL, Zuo XA, Liu LX, Lu DY. Influence of Horqin urbanization process on sandy landscape. Arid Zone Res. 2017;34(4):912–20.Search in Google Scholar
 Smith AB, Jackson DWT, Cooper AJG. Three-dimensional airflow and sediment transport patterns over barchans dunes. Geomorphology. 2017;278(15):28–42.10.1016/j.geomorph.2016.10.025Search in Google Scholar
 Liu BL, Courtyard TJ. Mapping the interactions between rivers and sand dunes: implications for fluvial and Aeolian geomorphology. Geomorphology. 2015;231:46–57.10.1016/j.geomorph.2014.12.011Search in Google Scholar
 Telfer MW, Hesse PP, Perez-Fernandez M, Bailey RM, Bajkan S, Lancaster N. Morph dynamics boundary conditions and pattern evolution within a vegetated linear dune field. Geomorphology. 2017;290:85–100.10.1016/j.geomorph.2017.03.024Search in Google Scholar
 Zhao HL, Zhang TH, Cui JH, Li YL. Climate change and its relationship with land desertification in the ecotone between agriculture and animal husbandry in northern China in recent 40A: a case study of Horqin Sandy Land. J Desert Res. 2000;V20(S1):2–7.Search in Google Scholar
 Wang YF, Zhang JQ, Ma QY, Zhu M. Response of desertification in Horqin sandy land to regional climate change in the early twenty-first Century. Trans Chin Soc Agric Eng. 2016;32(s2):177–85.Search in Google Scholar
 Wang T, Wu W, Xue X, Sun QW, Zhang WM, Han ZW. Temporal and spatial changes of desertification land in northern China in recent 50 years. J Geographical Sci. 2004;59(02):203–12.Search in Google Scholar
 Mao XL. Research on the application of remote sensing image supervised digital classification and GIS in the dynamic monitoring of land desertification. J Northwest Forestry Univ. 2007;22(1):145–8.Search in Google Scholar
 Yang SP, Shao J, Zhao WC. Research on land use change in mining area based on object-oriented classification. Henan Sci Technol. 2021;761(27):53–6.Search in Google Scholar
 Yuan Q, Shen H, Li T, Li Z, Li S, Jiang Y, et al. Deep learning in environmental remote sensing: achievements and challenges. Remote Sens Environ. 2020;241:111716.10.1016/j.rse.2020.111716Search in Google Scholar
 Weisberg PJ, Dilts TE, Greenberg JA, Johnson KN, Pai H, Sladek C, et al. Phenology-based classification of invasive annual grasses to the species level. Remote Sens Environ. 2021;263:112568.10.1016/j.rse.2021.112568Search in Google Scholar
 Aires F, Boucher É, Pellet V. Convolution neural networks for satellite remote sensing at coarse resolution. Application for the SST retrieval using IASI. Remote Sens Environ. 2021;263:112553.10.1016/j.rse.2021.112553Search in Google Scholar
© 2022 Huishi Du et al., published by De Gruyter
This work is licensed under the Creative Commons Attribution 4.0 International License.