Crop classification using remote sensing is the key to monitoring crop planting acreage and has great significance in further thematic monitoring. As field contains more accurate information of location and acreage than object which is the result of clustering similar pixels, it has been applied to crop classification using remote sensing increasingly. This paper summarizes the progress of per-field crop classification using remote sensing systematically, including its theories, methods and applications. Furthermore, a series of problems are analyzed and future study directions are viewed. Studies show that digitalization and image segmentation are the main approach to obtaining field boundary and more nationwide field database and bringing per-field classification a new opportunity. The strategies of per-field classification can be divided into two categories:using field features as input for the classifier and assigning field class based on per-pixel classification. The progress of features and classifiers in classification with remote sensing data are summarized further. It is indicated that combined application of multi-source data, development of field boundary detection, new features selection and improving implementation capacity of remote sensing image classification will be the crucial issues in per-field classification using remote sensing.
Multi-scale segmentation is one of the most important methods in object oriented information extraction, and the selection of optimal segmentation scale is a hot topic. Nevertheless, existing optimal segmentation scale selection methods only use spectral characteristics. In view of such a situation, this paper proposes a RMNE method, which uses textural information entropy to measure the heterogeneity between objects, uses spectral characteristics mean difference to neighborhoods to measure the object’s internal homogeneity and construct the evaluation function, and selects the optimal segmentation scales by drawing function curve. Taking 6 m spatial resolution multi-spectral SPOT6 image of the periphery of Beijing City as the multi-scale segmentation experiment example, the authors detected that the optimal scales combination is 30, 60 and 80. Compared with the multi-scale segmentation results whose optimal scales are obtained by the maximum area method and objective function method, it is shown that the effect of RMNE method is the best, which verifies the validity of the RMNE method and the applicability of the high resolution image. A comparison with Google Earth image shows that the image object’s size obtained by RMNE method is most consistent with that of the actual ground object.
In order to improve the hyperspectral images (HSI) classification accuracy and preprocess HSI by fully using the spatial and spectral information, this paper proposes a new spatial-spectral dimensionality reduction method, i.e., weighted spatial and spectral principle component analysis (WSSPCA). This algorithm reconstructs the HSI by using the physical characteristics of HSI, which can lower the influence of singular point in HSI. Principle component analysis (PCA) is utilized to reduce dimensionality of HIS, and it reduces the redundancy between bands and improves the HSI classification accuracy efficiently. The benchmark tests on PaviaU and Indian Pines demonstrate that the performance of WSSPCA is better than that of PCA and LPP when 5% and 10% samples in each class (10 samples are chosen when the total samples in every class is less than 100) are chosen randomly as train samples. The best values of Kappa coefficient obtained by WSSPCA are 0.955 9 and 0.896 1 respectively on the HSI datasets, exceeding the baseline by 0.193 8 and 0.205 0.
In order to make full use of the abundant spectral information and spatial information of hyperspectral images, this paper proposes a hyperspectral image classification algorithm based on dominant sets clustering and Markov random fields. First of all, the local spectral-spatial consistency of hyperspectral images is analyzed, the measurement of both band informativeness and independence is completed, an un-directed weighting graph is constructed and dominant sets clustering method is used to select the optimal band subset which preserves good structure information. Secondly, the local spectral-spatial consistency of adjacent pixels after the band selection is established by using Markov random fields, which makes the context information of the image space effectively used. Finally, according to the Bayesian theorem, the hyperspectral image classification problem is transformed into the maximum posterior probability which can solve the problem and yield the classification results. Experiments on two datasets, i.e., Indian Pines and Pavia University, show that this algorithm can achieve higher overall classification accuracy and Kappa coefficient than other similar algorithms.
Successful remote sensing image registration is one of the foundations of many remote sensing applications. Image high-lever features extracted by convolutional neural network (CNN) have achieved excellent performance in image classification and retrieval, and can be used to solve some problems of low-lever image registration features, such as the limitation of expression capability and easily being interfered. Hence, in this paper, the authors investigated the problem as to how to use CNN feature for remote sensing image registration. First, the authors investigated different CNN features from fully connected layers and aggregating convolutional features with different sizes from convolutional layer to register remote sensing image. Then the authors introduced the procedure by using CNN feature for image registration. Finally, the authors compared the registration performances of CNN features and scale-invariant feature transform (SIFT) features after the transformation of the image’s perspective, brightness and scale, respectively. The experimental results show that the CNN feature has better matching performance than the SIFT method in terms of matching accuracy and correct number of corresponding points. The finely tuned CNN feature has stronger robustness to the transformed image than the SIFT feature.
In order to improve the detection effect of the traditional algorithm on the ground objects in high resolution remote sensing images, this paper applies the deep learning object detection framework Faster R-CNN to the object detection task of high resolution remote sensing images. The airport and aircraft are used as the test scene and detection object for the experiment respectively, The Faster R-CNN framework is trained using the high-resolution remote sensing image data set to obtain the corresponding object detection model. The model is used to detect aircraft objects in high resolution remote sensing images and perform statistical analysis of the experimental results. The experimental results show that the Faster R-CNN model can entirely and accurately detect aircraft objects with an optimal F1 score of 0.976 3, and the same model can be used for object detection of multiple high resolution remote sensing images.
In order to improve the precision of damage assessment of post-earthquake buildings based on remote sensing images, this paper introduces a deep convolutional neural network (DCNN) model that performs well in natural image classification and target detection, and also proposes a method of using DCNN fully-connected layer features combined with support vector machine (SVM) to detect damaged building areas in remote sensing images. Firstly, neural network feed forward is used to extract the features of the training samples and the regions to be detected from the DCNN fully-connected layer; then the SVM classifier is learned based on the training samples; finally, all the blocks in the detection region are subjected to predicting and voting to determine whether they are damaged. The authors used Haiti earthquake remote sensing imagery in 2010 to do verification. The accuracy rate of damage detection in this method can reach 89%. Compared with the traditional feature extraction method, the correct rate is improved by 4%. The experimental results show that this method has a certain potential in the detection of building damage damage.
Video satellite capture color video data are obtained by using the single CMOS sensor with color filter array (CFA) Bayer pattern. To obtain the full color image sequences, researchers should perform the processing of Bayer image interpolation. Aimed at the interpolation of satellite video Bayer image, the authors propose a improved method based on the signal filter reconstruction of luminance and chrominance. Firstly, band-pass filters are used for extracting luminance and chrominance signal. And the initial reconstruction result can be obtained according to the Bayer spatial model. Moreover, median filtering in non-smooth region and edge direction interpolation in the smooth region are applied to the G-B or G-R difference band for updating the green band. In the end, the red band and blue band are also updated in the new G-B or G-R difference band. To verify the feasibility of the proposed method, the authors tested 4 video Bayer image of Jilin-1 01 and 03 and compared the obtained results with two classic methods. The experimental results show that method designed by the authors has the best comprehensive performance both in subjective evaluation and objective evaluation. The reconstructed image quality is good and has no obvious noise and pseudo-color effects; in addition, the edge is sharp and clear. The method can be used for satellite video further processing and application.
Regional and spatial continuous land surface temperature (LST) can be retrieved from satellite remote sensing data, and has an important significance in such fields as global change, ecology, environment, and agricultural production. However, the LST retrieved by remote sensing usually has missing data in time and space due to the influence of clouds, aerosols, satellite viewing angle and solar illumination angle, which limits the application of LST products. In this paper, the authors reconstructed FY-2F daily LST data of 2013 in the Yangtze River delta region using Savitzky-Golay (S-G) filter based on the characteristics of long time-series LST. The results show that S-G filter can fill the missing values effectively and ensure the spatial distribution consistency of the LST after reconstruction. The average time-series loss rate of the original FY-2F LST product is 19.43%, and then decreases to 1.69% after S-G filtering. In order to verify the reconstruction accuracy of S-G filter, the authors randomly selected some regions that are not deficient, and then made comparison with the results after S-G filtering. It is proved that S-G filter reconstructing method has obtained high accuracy, with the mean absolute error 1.35 K and the fitting accuracy 0.95. Higher quality and long time-series FY-2F LST which is reconstructed based on S-G filter offers a good foundation to the study of temporal and spatial distribution of further thermal environment.
Topographic effect is one of the main obstacles in quantitative analysis of remote sensing. For the airborne hyperspectral remote sensing, both of the impact of terrain height and angle can’t be ignored, and this causes more severe topographic effects. By taking the CASI image and LiDAR data of Qinghai Province as experimental data, the impact of elevation factor was analyzed in this paper. Firstly, on the premise that each elevation point is a horizontal Lambert body, four different elevation values were taken as reference to calculate the corresponding atmospheric radiation correction parameters by performing MODTRAN, which contain path radiance,atmospheric transmittance between the object and the sensor, atmospheric hemisphere albedo, and total downward radiance. Then an atmospheric radiation correction method with elevation factor was designed and applied to the atmospheric correction of CASI image. Finally, the CASI hyperspectral image was also processed by using FLAASH, which could only take one elevation value as reference. A comparison of two results shows that the reflectance spectrum shapes of the same ground objects are roughly the same,but the reflectance values are different. Especially, the short-wavelength reflectance values of FLAASH results are negative, and it is undoubtedly wrong. The experiment shows that the impact of elevation factors can’t be neglected. Atmospheric correction by adding elevation factors can get better results. For achieving accurate topographic correction of airborne hyperspectral image, both elevation and topographic angle factors should be considered simultaneously.
Leaf area index (LAI) is an important structural variable for quantitative study of the energy exchange characteristics of forest ecosystems. Based on field observations of LAI, 7 kinds of vegetation indexes and 5 custom vegetation indexes based on Landsat TM, LAI estimation model of different forest types were established through the model screening, in which the multiple regression model for coniferous forest and principal component analysis model for broad-leaved forest and mixed forest were used. Finally, the regional scale forest LAI distribution map was made through multiple model estimation. The accuracy of LAI is 0.829 4, 1.111 5 and 1.790 9 for coniferous forest, broad-leaved forest and mixed forest respectively. And the total R 2 is over 0.77 for all the forests. The results will provide basic data for forest ecosystem and carbon cycle studies.
Vegetation cover causes great interference in rock alteration information extraction. Forcing invariant vegetation suppression technology has achieved good vegetation suppression effect in semi-arid and open terrain area, but the effect remains to be verified in mountainous areas where vegetation is flourishing. Based on the forcing invariant vegetation suppression technology, in the southern vegetation area, the subsection leveling and programming are implemented in the key technical curve leveling steps, which can solve the contradiction between vegetation suppression, color deviation in bare land and information integrity. The vegetation information after subsection leveling is well suppressed, and the underlying bedrock information is prominent and the tone is natural. By using this method to extract remote sensing alteration information, the vegetation area’s remote sensing anomaly is obviously enhanced, the anomaly agrees well with the actual wall rock alteration, and the effect is better.
Normalized difference vegetation index (NDVI) trends can approximate the trend of “greening” or “browning” of vegetation and reflect the adaptation process of vegetation to global change. In this paper, an NDVI trend analysis method combining empirical mode decomposition (EMD) and Mann-Kendall (MK) significance test is proposed on vegetation monotone trend detection. The method includes mainly two steps: firstly, EMD is used to decompose NDVI time series into a finite number of intrinsic mode functions (IMF), and these components contain the local characteristic information of different time scales of the original signal. The first component is a high-frequency component, the subsequent component frequency gradually decreases, and the residual is a monotonic function, indicating the average trend. From the decomposition, the NDVI variation trend along with time is extracted naturally. Secondly, the MK significance test is used to detect the monotonicity of the trend varied, that is, to detect that the trend is monotonically increasing or monotonically decreasing, the monotonically increasing is corresponding to the trend of vegetation getting “greening”, and the monotonically decreasing is corresponding to the trend of vegetation getting “browning”. The test data are MODIS NDVI time series of 16 days from 2006 to 2015. The analysis of the trend detection of those NDVI time series shows that the method proposed in this paper is an effective method for time series trend analysis and has a wide application prospect.
The water extracting has the characteristics of time point effects. In view of the objective status of seasonal variation of land water, a method of geospatial correction for water extracting products is proposed. Firstly, the water land cover information is extracted based on high time resolution remote sensing image to ensure that the timeliness meets the standard time point. Then the result is used as a prior knowledge, and the refined water land cover information is extracted based on fine grid DEM data by using region growing algorithm of water seeds, whose accuracy is optimized to the high spatial resolution level and can meet the requirement. On such a basis, it achieves geospatial correction of water extracting products. With the first national geographic conditions census as an example, the Landsat 8 images of 15 m spatial resolution were obtained to meet the standard time point of the study area. The water land cover distribution was extracted based on the NDWI index, and the 2 m grid DEM data were used to optimize the precision. The results show that the geographical spatial range of the study area was corrected by 17.97% compared with the image source’s scanning time, and geographical spatial range was optimized by 1.56% caused by the spatial resolution conversion. The research shows that this method can provide a reference for the geospatial correction in the water extraction based on remote sensing technology, and has certain practical application value in the case that the images do not meet the requirements of the standard time point.
The algorithm of chlorophyll-a concentration inversion with higher universality is the key to improving the practicability of quantitative remote sensing technology. Based on the radioactive transfer mechanism, the optical characteristics of chlorophyll-a and other factors in inland lakes are analyzed, and a physical model of pixel reflectivity and factor concentration is established. The model was applied to the remote sensing data of different phases in Chaohu. The determination coefficient was 0.877 8 and the average relative error was only 11.61%. This proved that the precision of the model was higher and the universality was stronger. Then, the preprocessed Chaohu remote sensing image was applied to the model, and the spatial and temporal distribution characteristics of eutrophic pollution in Chaohu were obtained, which is consistent with the regulation of the seasonal multiplication of algae. The model used in this study has high accuracy and universality and thus can promote the application of quantitative remote sensing technology in water pollution research.
Thickness of oil slick is an important parameter of oil spill volume. In order to confirm the feasibility of oil thickness estimation with hyperspectral data,the authors used ASD FieldSpec3, quartz halogen lamp and crude oil for a laboratory experiment which simulates oil slick and spectral measurement. 27 pairs of oil thickness and reflection data were acquired. To make full use of spectral information of the hyperspectral data,the authors selected partial least square (PLS) to slick thickness and reflection modeling with 21 set model data and 6 test data set. Model result shows that PLS model expresses optimal effect when five principal components are selected which interpret 74% information of independent variables and 99.8% information of dependent variable, the prediction capability of the model runs up to 92.8%. The root mean squared error is 0.01 for modeling samples and 0.04 for validation samples. The PLS model shows better accuracy of modeling and validation error compared with traditional model, and thus it can be used in oil slicks thickness modeling with hyperspectral data.
With the Jiangle state-owned forest farm of Fujian Province as the study area, the potential of classification in tree species and age groups through GF-2 image were explored. First, the canopy spectral curve of main tree species were measured and the reflectance differences between them were analyzed. After image preprocessing and in combination with normalized difference vegetation index (NDVI) and topographic factors, multi band remote sensing images were constructed. Object-oriented multi-scale segmentation technology was applied to extracting the spectral and texture attributes, followed by attributes filter. On the basis of 7 kinds of schemes, Cunninghamia lanceolata (3 age groups),Pinus massoniana and Phyllostachys edulis were classified by random forest classifier. The role of spectrum, texture and auxiliary data in classification was quantitatively analyzed. The results show that the scheme of spectra combined with 4 directions of texture attributes has overall accuracy of 87.4% with Kappa coefficient being 0.85, and age groups in Cunninghamia lanceolate were effectively classified. Random forest classifier can achieve better classification results based on the optimal attribute set. GF-2 has great potential in tree species and age group classification and provides reliable data source for forest resources investigation and management.
At present, the area of high standard farmland has reached a certain scale in China. In the remote sensing monitoring for the utilization of high standard farmland, illegal utilization has appeared frequently. How to realize real-time and accurate remote sensing monitoring for high standard farmland has become an urgent problem for the land regulation department of the government. The national high standard farmland monitoring area is large, and the monitoring precision requirements are high. It is urgent for the government to study a set of high standard farmland automatic monitoring methods adapted to the nationwide extension. In this paper, two automatic remote sensing classification monitoring methods, i.e., object oriented and maximum likelihood, are compared. The overall precision of the object-oriented method is 98.684 7%, and the Kappa coefficient is 0.983 3. The overall accuracy of the maximum likelihood classification method is 78.587 1%, and the Kappa coefficient is 0.718 0. The research shows that the object-oriented classification method can better meet the requirements of the high standard farmland. By popularizing the method, it is the way to provide efficient and accurate decision-making information for real time supervision of high standard farmland, and can provide technical support for the national protection of cultivated land and food security.
The Loess Plateau ecological barrier area is an important part of the “ two screens and three belts” in China. It not only has an important barrier effect on local residents, but also has an important impact on the middle and lower reaches of the Yellow River. This paper is based on the typical area in Loess Plateau as the research object, takes the soil conservation quantity as the evaluation index and uses the revised universal soil loss equation (RUSLE) as the evaluation method. Meanwhile, It uses the data of land use/cover change, meteorological observation, MODIS data and some other data to assess the effect function of the grain for green and the conservation of soil and water of the typical area in Loess Plateau from 2000 to 2010. The results showed that the land use/cover type changed sharply in the study area in nearly 10 years. A large number of arable land became the forest and grassland, the vegetation coverage had increased apparently and the service function of the conservation of soil and water rose linearly. The soil erosion modulus was close to 1 986.66 t·km -2·a -1 in 2010. These data show that the ecological environment has improved and the function of the conservation of soil and water have enhanced significantly since the government implemented the grain for green.
Wetland, one of the most important ecosystems on Earth, is well known as belonging to three major ecosystems together with forest and ocean. In recent years, wetland ecosystems have been threatened by the impact of human activities and urban development. It is of great significance to carry out the study of the relationship between wetland landscape evolution and human disturbance and to protect the ecological environment of wetland in Xiong’an New Area. In this paper, the authors analyzed the wetland landscape evolution and the response to human disturbance in Xiong’an New Area based on the moving window landscape method by using the land use data from the late 1980s to 2015. Some conclusions have been reached: From the 1980s to 2015, wetland landscape changed in Xiong’an New Area. From the time series, the wetland area showed a decrease trend mainly and wetland fragmentation increased gradually. The patch shape became complicated and the degree of connection was reduced. From the spatial distribution pattern, after the year of 2000, the wetlands fragmentation gradually increased in the central and northeastern area, and the fragmentation from the core area of the central wetlands to the edge gradually increased. From the 1980s to 2015, human disturbances showed a decreased trend mainly. The human disturbance decreased in the central and southwestern area from the 1980s to 2000. From 2000 to 2015, the human disturbance increased and then slowed down, which indicated that the wetland had shifted to the low intensity and sustainable development under the implementation of wetland protection policies. The human disturbance of the surrounding areas was affected by the expansion of construction land. The human disturbance has a good correspondence with wetland distribution.
This study is based on five Landsat TM remote sensing images obtained in Nanjing City in 2004, 2007, 2010, 2013 and 2016 respectively. With landscape index method, fans analysis method and some other spatial analysis methods, the expansion characteristics of Nanjing City and its evolution were analyzed quantitatively, which revealed urban expansion in Nanjing City and morphological evolution feature in time and space. The results show that, during the period of 2004—2016, the construction land expansion of Nanjing City was mainly realized through occupying large areas of cultivated, and the speed of city expansion was faster than population growth, which was not conducive to the sustainable development of Nanjing City structure. In the growth of the construction of the city area, city expansion rate increases year by year, city expansion first strong expands and then the expansion tends to slow down, regional differences of city space growth are significant. The internal structure of Nanjing City tends to be stable, the external contour tends to be compacted, and urban area expansion is weakened. Expansion mode is mainly along the river, Nanjing City gradually forms urban spatial structure, exhibiting the pattern of the “two belts and one axis”, and is in a multi center development trend. Urban expansion in Nanjing City forms a non-differential balanced state space, and the city expansion is mainly concentrated in the north, southwest and southeast of of Nanjing, and shows obvious characteristics of various stages.
In the process of targeted poverty alleviation, the problems that traditional data statistic aperture is not unified and that nighttime light data for identifying poverty is studied in a short time usually exist. With Liupan Mountain as an example, the average light index and multidimensional poverty index (MPIstatistical) indices were constructed by using the method of invariant target area and gray relational model with the help of night light and socio-economic statistics. Poverty estimation models were constructed through average light index and MPIstatistics. MPIestimation was generated and used to explore long-term sequence of poverty identification. Some conclusions have been reached: the accuracy of poverty results based on nighttime light image was higher, which can reflect the real poverty degree of the region, and the relative error ranges between 3.14% and 3.52%. The MPI estimated averages of the contiguous special poverty areas respectively are 0.346, 0.353, 0.353, 0.357 and 0.358 in many years. The level of poverty has been reduced year by year. Between 2000 and 2012, there were 3946 counties with extremely poor conditions and 2021 counties with highly poor conditions. The Moran’s I index from 2000 to 2015 respectively were 0.49, 0.45, 0.47, 0.49 and 0.43, indicating that the poverty level in 78 counties exhibits obvious agglomeration. The pattern of poverty is presented with the spatial evolution trend of “relatively less poverty in the eastern and western regions and relatively heavier poverty in the northern and southern regions”.
Revealing the spatial differentiation characteristics and influencing factors of land surface temperature (LST) in the plateau area is of great significance for the study of local climate change. However, the existing research merely analyzes the relationship between single factor and LST, whereas the study of the spatial differentiation characteristics and the quantitative analysis of influencing factors of LST in the plateau area are relatively insufficient. Taking the Sangzhuzi District of Xigaze City as an example, the authors used Landsat8 remote sensing data to invert the LST of the study area by using radiative transfer equation algorithm and the universal single-channel algorithm. In addition, the factor detector and interaction detector in the geodetector model were used to quantitatively detect the influence of single factor and multiple factors on LST, respectively. The results show that, in the quantifiable factors, LST increases first and then decreases with the increase in the degree of aspect, and there is a significant negative correlation between LST and other factors with a difference in the descent speed. Elevation is the most important factor affecting the spatial distribution and forming the differentiation characteristics of LST in the plateau area, followed by normalized difference vegetation index(NDVI), aspect, normalized difference moisture index(NDMI), soil type, slope, and average annual precipitation; the spatial distribution and the formation of differentiation characteristics of LST in the plateau area are the result of multiple factors, all of which have a synergistic enhancement effect under interactions, such as the interaction of elevation and aspect, elevation and NDMI, and elevation and NDVI with the most significant impact.
Based on MODIS normalized difference vegetation index (NDVI) and land surface temperature (Ts) data, the authors constructed a bi-parabolic NDVI-Ts space which was verified by the filed measured soil moisture, and monitored the spatial and temporal distribution characteristics of drought conditions in Shaanxi Province from 2000 to 2016 based on the TVDI obtained from bi-parabolic NDVI-Ts space. The results show that the NDVI-Ts space was bi-parabolic and there was a significant negative correlation (P<0.05) between TVDI and 10 cm depth filed measured soil moisture. Spatially, the drought in Shaanxi Province during 2000—2016 were mainly distributed in the northwest, north of Shaanxi and the northeastern regions of Guanzhong plain; the drought area of Shaanxi Province accounted for 31.95% in 2000 and 27.65% in 2016, respectively. It is found that drought was significantly relieved in most northern part of Yulin City, the middle part of Yan’an City and the central part of Guanzhong Plain and some parts of southern Shaanxi, which accounted for 14.45 %. The drought conditions in 84.48 % of the province were changed, but the change failed to pass the significant test. 97.62% of the province had a small variation coefficient, which was between 0 and 0.8. It was mainly distributed in northern Shaanxi, south of Guanzhong Plain, and it showed that the drought conditions were stable in Shaanxi Province. There was a significant negative correlation between drought and annual precipitation, accounting for 23.74 % (P<0.1). With the increase of rainfall, TVDI decreased, and the drought was relieved. It was mainly distributed in most areas of Yulin City, central parts of Yan’an City, north and northwest of Hanzhong City, Ankang City, northern parts of Weinan City, eastern parts of Shangluo City and western and northern parts of Baoji City. It is found that the changes of drought in other areas were not significantly affected by precipitation. The annual temperature was not dominant factors that resulted in the change of drought in Shaanxi Province.
Global climate warming and human activities have caused large areas of permafrost degradation and thermal erosion gully in the Tibetan Plateau, seriously affecting the engineering construction and the ecological environment in permafrost regions. In this study, high resolution unmanned aerial vehicle (UAV) images and object-oriented classification approaches were applied to extracting the thermal erosion gullies in Eboling Mountain of Heihe River. Five kinds of object-oriented supervised learning algorithms, namely nearest neighbor, K-nearest neighbor, decision tree, support vector machine (SVM), and random forest, were analyzed for the capability and accuracy of the extraction of thermal erosion gullies in detail. The field GPS data were used for evaluating the classification accuracy. The results show that, in the object-oriented image analysis, the segmentation scale parameters have little effect on the extraction of thermal erosion gullies, wheres classification features have a greater impact, so it is important to select the appropriate classification features. The overall accuracies of the five machine learning methods are all over 90%, among which the Kappa coefficient of the SVM is higher than the other four classification methods. This means that SVM is more suitable for the thermal erosion gullies boundary extraction of UAV images in this study. The combination of high resolution UAV images and object-oriented classification methods has broad application prospects in the extraction of the thermal erosion gullies.
In this paper, on the basis of prairie biophysical characteristics and in combination with the principle of energy exchange (sensible heat and latent heat flux obtained by remote sensing and meteorological data), the fuel dry index (Fd) was proposed and applied to the Shandong prairie fire monitoring. Fd can better solve the prairie fire forecast, fire danger early warning in time and space and the estimation accuracy. It can change dynamic warning daily high fire risk areas with time in Shandong Province. Fd and fire potential index (FPI) were used to study the fire danger on April 8, 2010. Fire indicating effect of Fd is better than that of FPI. In the equidistance fire classification, data of 31 fire points in 2010 indicated by Fd fell in grade III, accounting for 87.1%, and 0 fell in grade I; the fire locations were in good agreement with areas of high fire risk early warning. In fuel dry index (Fd) graph, it can be seen that Fd has close relationship with the prairie vegetation growing season; the early development of Fd is high, but later it exhibits decreasing trend; at the medium stage, Fd is low; at the late stage,Fd is high, and shows a trend of rising. Overall, the Fd index plays an important role in fire danger forecast at the grassland growing stage.
As a slow onset geological hazard, ground subsidence could damage buildings. In particular, the settlement risk induced by subway construction on building structures has become a matter of concern to governmental authorities and the public. Spaceborne interferometric synthetic aperture Radar (InSAR) technology could acquire high-precision surface deformation information and provide technical support to evaluate the risk level of urban buildings caused by ground subsidence hazards. Based on the deformation data acquired from September 2013 to September 2016 by PSP-InSAR algorithm, the authors selected the buildings near a subway station in Shenzhen as the study object. In data analysis process, first of all, combined with the construction scheme of subway station, geological information and the property of the building, the authors carried out the corresponding research on the change of the deformation trend in different periods and the risk assessment of the settlement disaster. Then, one building in the study area was selected as the research object, and the differential deformation and inclination were analyzed based on the deformation of PS at different locations. Combined with the corresponding standards, the risk of building subsidence disaster was preliminarily evaluated. Finally, by comparing with the leveling data, the precision of the InSAR measurement results was discussed. In accordance with the field investigation, it is verified that the corresponding risk symptoms have been found on the buildings whose deformation values were identified as relatively high in the analysis process. The comparison between data analysis and field investigation results confirms that InSAR technology is capable of playing an important role in urban building risk management process in the future and the methodology can be widely applied beyond the case study area.
tability monitoring of bank slopes along the reservoirs of hydropower projects is a fundamental task for the safety of dam operation. And deformation detection is a major approach for stability monitoring. Spaceborne InSAR technique has been recognized as an effective tool for deformation detection with its high observation accuracy and capability to work independent of weather and solar illumination. The deformation information of left bank slope of Jinping hydropower station in the Yalong River Basin was obtained by processing 56 images of C-band Sentinel-1 data with small baselines time series InSAR technique. The result indicated that there was a large landslide on the left bank slope about 1.5 km away from the dam upstream of the Jinping I hydropower station, with a surface area of more than 750,000 square meters. The maximum deformation rate in the line of sight exceeded 200 mm/a from 2015 to 2018. The deformation area was mainly concentrated in the middle and upper part of the bank slope. And the maximum cumulative deformation of the line of sight in the observation period was more than 500 mm. The time series of deformation was basically a linear sliding trend without obvious periodicity. The same method was used to process 22 archived images of L-band ALOS-PALSAR data from 2006 to 2011. The results show that the left bank slope was stable before the reservoir impoundment. It is therefore inferred that the sharp rise of water level of the reservoir might be a main trigger factor for this landslide activation.
According to the relationship between the deformation position determination, the deformation gradient estimation and the coherence of the InSAR monitoring surface deformation in the mining area, the Sentinel-1A data were used to study the applicability of the InSAR technique in the monitoring of the mining area. The experimental results show that the surface type of bare land and village is maintained at high coherence throughout the year under the condition of semi - humid monsoon climate in the warm and temperate zone in Henan Province. Visual identification can effectively determine the deformation position of the mining area. The deformation range can not be accurately determined by the deformation gradient function model. The deformation gradient is also within the range of the detectable range. The field coverage type can effectively determine the deformation of the mine in summer. In addition, the deformation gradient is located on the detectable minimum deformation gradient by using the deformation gradient function model. Furthermore, the applicability of the model is proved by the leveling data.
Based on remote sensing technology, the authors investigated the distribution of tailings reservoir in Tibet, such as its mineral resources, utilization status and scale. The current mining intensity of different administrative regions, different metallogenic belts and different mine types in Tibet was analyzed. Some conclusions have been reached: for different prefectural-level divisions, the metal mines’ mining intensity in Lhasa City is the largest, the metal mines exploitation potential in Lhasa City and Naqu area are larger. For different county-level administrative regions, the metal mines mining intensity in Mozhugongka County of Lhasa City is the largest, the metal mines exploitation potential in Mozhugongka County of Lhasa City and Shenzha County of Naqu area are larger. For different important metallogenic belts, the metal mines mining intensity in Gangdise metallogenic belt is the largest, the metal mines exploitation potential in Gangdise metallogenic belt is also the largest. For different mine types, the metal mines mining intensity of nonferrous minerals is the largest, the metal mines exploitation potential of nonferrous minerals is also the largest. For different specific mine types, the metal mines mining intensity of lead-zinc mines and copper mines are the largest, and the metal mines exploitation potentialof lead-zinc mines is the largest.
The research area is located in the tropical rainforest region in Southwest China, with a humid and rainy climate, and there are many soils and vegetation in this area and few exposed rocks. These factors have brought great difficulties to remote sensing mapping. In order to minimize the impact of disturbances caused by tropical rainforest coverage areas, the authors used high spatial resolution remote sensing data, Radar remote sensing data, and elevation data. On the basis of conventional remote sensing geological interpretation methods, the authors used the correlation between vegetation and underlying bedrock, soil and underlying bedrock, topography and geological bodies, human engineering activities and lithology to summarize the interpreting marks of various geological bodies in this region such as hue, morphology, geomorphology, and human engineering activities and develop remote sensing geological interpretation and geological mapping of tropical rainforest coverage areas, thus achieving good results and providing relatively reliable remote sensing geological interpretation methods for similar regional geological surveys.
The west of Xiangshan basin is an important potential area in the search for deep-buried uranium ore deposits in Xiangshan uranium orefield, and its deep alteration zoning remains to be further explored. Imaging hyperspectral core scanning technique provides a new means for revealing deep alteration information. On the basis of the imaging hyperspectral scanning data of deep drilling cores in the Niutoushan area of western Xiangshan, mapping of 5 types of altered minerals was realized by data processing. Then, pixels statistic algorithm was used to obtain the relative content logging curve of each altered mineral. The reliability of the imaging hyperspectral logging was verified by comparing the geological lithology and geophysical logging curves. According to the results of imaging hyperspectral logging of deep drill holes, the formations overlying the basement can be divided into three alteration zones. The main alteration of the first is chloritization, which is located in the upper part, and the second main alterations located in the middle part contain kaolinization, dickitization, and illitization dominated by shortwave illite, and the third main alteration is illitization that is characterized by more long wave and less short wave illite, located in the lower part. Uranium mineralizations in the upper and lower parts of the borehole have distinct features of altered mineral combinations and show that formation environments of illites with different wavelength characteristics are relatively different. The short wave illite tends to form in relatively acidic fluid environment, closely related to uranium mineralization controlled by acid alterations; the long wave illite tends to form in relatively alkaline fluid environment and is not closely related to uranium mineralization. Alteration zoning features of deep drill holes reveal that acidic fluid activity is later than alkaline fluid activity and acts on the latter. Uranium is gradually enriched with deuteric potassium metasomatism and acidic fluid activity. The action of deep fluid on the whole has the evolution characteristics of starting from sodium metasomatism to potassium metasomatism, followed by acid metasomatism with the time.
Cloud computing technology is developing rapidly and constantly expanding the application range. For exploring the cloud computing technology in the field of environmental remote sensing application, this paper discusses some key techniques of cloud computing based on virtualization and big data technology, which include architecture design, network topology and service function. 138 images of GF-1 satellite were selected for production experiments for comparing and analyzing the efficiency of cloud service platform and high performance platform in mass remote sensing data processing. Experiments show that the data processing efficiency of high performance cluster platform is about 2.5 times higher than that of cloud service platform under the existing operating environment. In general,compared with cloud service platform, dedicated high performance computing and processing platform has certain advantages in computing, communication and storage. It is more suitable for massive environmental remote sensing data processing and quantitative retrieval with efficiency.