To explore the application potential of airborne point cloud and UAV visible light image in tree species identification and classification， a single-tree scale tree species classification and recognition method based on UAV hybrid fusion of multi-modal features and decision was proposed. Firstly， Kendall Rank correlation coefficient method and Permutation Importance （PI） were used for feature selection， and Efficient Low-Rank Multi-Mode Fusion Algorithm （LMF） was used to fuse the selected point cloud and visible image features. Ensemble learning was introduced to input point cloud， image， and fusion features into eXtreme Gradient Boosting （XGBoost）， Light Gradient Boosting Machine （LightGBM）， and Random Forest （RF） base classifiers integrated by Stacking. Finally， the meta classifier， Naive Bayes， is used for decision fusion. The experimental data show that the independent test accuracy of the proposed algorithm is 99.4%， which improves 22.58% compared with the Random Forest classifier by traditional feature concatenate fusion. In addition， the Kappa coefficient also increased by 28.54%. The comparison experiment with Convolutional Neural Network（CNN） shows that the proposed algorithm has obvious advantages in small sample training and better generalization ability.

Water storage is a critical component of the global and regional hydrological cycle， which can be used to analyze the spatio-temporal evolution of regional water resources and drought. Traditional methods to monitor water storage are usually based on in-situ groundwater level data. However， challenges arise due to the limited placement and distribution of monitoring stations in large-scale research and exploration. The Gravity Recovery and Climate Experiment （GRACE） satellite have provided large-scale monthly data on Earth's gravity field variation. Several scholars have applied the water storage anomalies data retrieved by GRACE in hydrology research， which has facilitated the progress and development of hydrology. However， the current systematic elaboration of research on inversion of water storage based on GRACE data is not comprehensive enough， and few studies have systematically summarized the status of monitoring drought， interpolation， and reconstruction based on GRACE data. Firstly， this study briefly introduces the application fields of GRACE data， and discusses the advantages and disadvantages of the two data processing methods. Then， the application status and existing problems of GRACE data in the verification and uncertainty of inversion results， terrestrial water storage anomalies， groundwater storage anomalies， drought evolution and response， and interpolation and reconstruction were analyzed and summarized. Finally， further research about GRACE was suggested to carry out in the aspects of exploring the impact of changing environments on water storage anomalies， reducing the uncertainty of GRACE data， constructing a suitable drought index for drought monitoring， improving the accuracy of interpolation and reconstruction， and improving spatio-temporal resolution. The study is aiming to provide reference and insight for related research using GRACE data.

Speckle is an inherent property of SAR image， but its existence seriously interferes with the quality of SAR image and affects the high-quality application based on SAR image， so it is urgent to suppress it. The accuracy of the edge detection model of the traditional AD （Anisotropic Diffusion） filter still has room for improvement， and the noise suppression effect is often limited by the problem that it is difficult to accurately estimate the diffusion threshold. To solve the above problems， a novel AD filter based on Multidirectional Sobel （MSAD） is proposed. MSAD filter is an improved algorithm of SRAD. It builds a new edge detection model based on Multidirectional Sobel templates. Based on this， a new AD diffusion coefficient is established by integrating Gaussian kernel function， which can effectively solve the limitation of traditional AD diffusion coefficient by parameter estimation and improve the accuracy of speckle anisotropy suppression. Three real SAR images are selected for filtering experiments. In experiments， SRAD， DPAD， EnLee， and PPB filters are selected as the comparison algorithms； ENL， SSI， ESI， and M-Index are selected to evaluate the performance of proposed algorithms. Experiments show that MSAD filter can effectively improve the edge detection ability and obtain better speckle suppression effect.

Global warming results that glaciers retreat rapidly. Monitoring and mapping glacier boundary are extremely significant for research on global climate change and predicting related disasters. However， snow covering is the main barrier all the time. Selecting Karakoram subregion as study area， the Landsat 8 OLI， and Senitnel-1 images and DEM data in spring （March 24th， 2019） were utilized. The spectral reflectance of green， red， near-infrared and short-wave infrared bands in Landsat 8 OLI images were selected as the optical image features. The backscattering coefficient of VH polarization channel， the coherence coefficient of VV polarization channel， local incident angle， polarization entropy H and scattering Angle α after polarization decomposition were gained from SAR data and used as SAR features. Topographic features included DEM and slope. These characters were employed as input of models. First， based on U-Net model， experiments compared the accuracies using different-size samples. The 256×256-pixel-size samples were imported to U-Net network model based on different backbone networks （MobileNetv2， VGGNet， ResNet and EfficientNet） and DeepLabv3+ model. Finally， the best one among the above networks was employed to import samples with different feature combinations. Results show： ①Using the bigger training sample with the richer spatial context information can obtain the higher segmentation accuracy and the glacier terminal boundary is more accurate. ②Among the different backbone networks， VGG19 backbone network exhibits the highest accuracy， which is higher than that of DeepLabv3+. Its F1-value is 0.899 6， and the mean intersection over union（mIoU） is 0.875 4， and the overall accuracy is 0.948 4. The recognition effect of shadow， snow melt-water， mist covering and frozen lake area is comparatively good. ③With the decrease in the number of training features， the accuracy also drops. Topographic features can improve the precision rate， while SAR features can increase the recall rate by 4% or so. This study proves the feasibility of the deep learning methods on the identification of mountain glaciers covered by a large amount of snow and provides reliable basis on model selection and parameters setting for rapid and large-scale mountain glaciers mapping.

In recent years， in order to improve the classification accuracy of ground objects， break through the technical system of single sensor， and make up for the limitations of single data source application， multi-source remote sensing data fusion has become a research hotspot concerned by many scholars in the field of remote sensing. The fusion technology of optical image and LiDAR point cloud data of hyperspectral remote sensing technology provides a feasible scheme to improve the accuracy of ground object recognition and classification at the technical level， breaks the technical upper limit of single sensor， and provides a new solution for the integrated acquisition of target three-dimensional space-spectral information. At the same time， it lays a foundation for the research of hyperspectral LiDAR imaging technology. This paper reviews the development history of LiDAR and hyperspectral imaging data fusion， discusses the main fusion methods and research progress at the feature level and decision level， introduces the commonly used feature level fusion and decision level fusion methods in detail， summarizes the latest research algorithms and discusses their challenges and future development and application prospects. Finally， the future development of LiDAR and hyperspectral imaging data fusion is prospected systematically.

Surface deformation is a geological phenomenon caused by natural or artificial factors， and its disaster-causing process is slow and irreversible. It is also a geological disaster with destructive solid power. Therefore， real-time and high-precision surface deformation monitoring is one of the most critical tasks in maintaining urban safety. However， due to the complex causes， long duration， wide range， and many triggering factors of surface deformation， there are many difficulties in monitoring surface deformation using single technology such as leveling， GNSS， INSAR， and optical remote sensing. Considering the characteristics and complementarities of InSAR and GNSS， the combination of InSAR and BeiDou/GNSS can improve the surface deformation monitoring capability in space and time at the same time. Unluckily， the traditional GNSS-InSAR data fusion method is simple to fuse and cannot dynamically reflect surface deformation characteristics， leading to insufficient data use and low accuracy of deformation features. A new fusion method is proposed based on the Kalman filter algorithm GNSS-InSAR correction values. The method mainly consists of two sequential processes， i.e.， the a priori processing of GNSS and INSAR data and the fusion process of GNSS-InSAR correction values based on the Kalman filter algorithm. The a priori processing of GNSS and INSAR data is to obtain the a priori deformation results using the fitted estimation model to correct the systematic errors in the InSAR observations. The fusion process of GNSS-InSAR correction values based on the Kalman filtering algorithm is to fuse the two data through Kalman filtering based on the spatial and temporal correlation between the time-series GNSS observations and the InSAR correction observations. The experiment was processed using 103 views of sentinel-1A data from November 15， 2018， to June 3， 2022， and 13 GNSS point data during the same period. The experimental results show that the fusion result of the corrected InSAR observations and GNSS observations by the Kalman filter is 45% more accurate than the fusion result of the uncorrected InSAR observations and the GNSS observations， which is 45% 57% higher than the accuracy of InSAR observations. Therefore， the fusion method model based on the Kalman filter algorithm of GNSS-InSAR corrected values proposed in this paper improves the accuracy of InSAR deformation monitoring and expands the breadth and depth of InSAR applications.

As an important part of terrestrial ecosystems， vegetation is often used as an indicator to assess the effectiveness of climate change and ecological restoration. In this study， the Shiyang River Basin is taken as an example， Theil-Sen and Mann-Kendall models， and the Hurst index were used to analyze the change characteristics of vegetation cover. The correlation analysis， residual analysis and Geodetector were used to explore the influencing factors of vegetation cover change. The results showed that the vegetation NDVI demonstrated a fluctuating but upward trend from 2001 to 2020， with a rate of increase of 0.023/10 a. Areas with significant increased and significant decreased accounted for 72.32% and 2.4%， respectively. Areas with sustainability （Hurst>0.5） accounted for 63.84 % of the entire area， among which 47.37% showed continuously significant increasing trend. The correlation results between NDVI and climatic factors indicated that the impact of precipitation was particularly significant， and the impacts of temperature， solar radiation and saturated vapor pressure deficit were relatively weak. The area of NDVI_pre showed a significant increase trend accounted for 21.59%， while the area of NDVI_res showed a significant increase trend accounted for 60.07%， so interannual variation of NDVI in Shiyang River Basin was greatly affected by human activities. Geodetector results showed that the spatial distributation characteristics of water-heat conditions. It is noted that the spatial distribution of NDVI of cultivated land is greatly affected by population density. The results of this study are helpful to deepen the understanding of the driving factors of vegetation change and provide scientific reference for ecological protection and restoration of Shiyang River Basin.

Estimation of Aboveground Biomass （AGB） in Spartina alterniflora （S.alterniflora） can provide an important basis for ecosystem stability evaluation and regional carbon sink assessment in coastal wetlands. Using typical coastal wetlands in Zhejiang， China as an example， this study used 48 measured aboveground biomass data of S.alterniflora to extract vegetation index and band characteristics reflecting aboveground biomass information of vegetation based on Landsat8 OLI images， constructed a Univariate Regression （UR） model， a Multiple Linear Regression（MLR） model， a multi-scale geographically weighted regression model （MGWR）， and a Partial Least Squares Regression （PLSR） model to estimate the AGBof S.alterniflora from actual field measurement data. The results showed that： （1） The AGBof S.alterniflora was significantly correlated with the 29 selected remote sensing variables， and the correlation coefficients were all between 0.5 and 0.8（P<0.01）. （2） The model constructed by PLSR method was the optimal model of S.alterniflora AGB inversion in the Zhejiang coastal wetland （R²=0.767； RMSE=130.576 g/m²； MAE=100.801 g/m²）. （3） The average AGB of S.alterniflora in Zhejiang coastal wetland was 6 607.01 g/m²， and the total AGB was 1.36×10³ t. The distribution pattern of S.alterniflora AGB in the coastal area of Zhejiang Province was high in the south and low in the north. This study could provide the scientific basis for the rational development and utilization of coastal wetland resources， carbon sink monitoring， and ecosystem function evaluation.

The Eastern Tianshan regions of Xinjiang were characterized by severe climate and exposed bedrock. The lithological boundaries were not accurate enough in the existing 1∶200 000 geological map， and the application of high-resolution remote sensing technology was lacked. GF-1 and Landsat 8 satellite images were used for lithology identification and geological mapping in Changji Area， Eastern Tianshan， Xinjiang. Image processing technology such as IHS and Gram-Schmidt were executed to obtain high spatial resolution images. And the image processing process was based on image enhancement using False Color Composite （FCC）， Principal Component Analysis （PCA） and Minimun Noise Fraction （MNF）. Image interpretation markers were established by the combination of 3D image constructed from DEM data and available geological records. Additionally， an updated 1∶50 000 lithological map was generated for study area by fieldwork verification， sample thin section identification and reflectance spectrum characteristic analysis. The results showed that， in Changji area with good outcrop of bedrock， the integrated application of multi-source remote sensing data could identify the lithological units which were missed in the 1∶200 000 geological map and corrected the lithological boundaries. It improved the efficiency of geological mapping and guided the follow-up geological survey and geological prospecting.

The vegetation greening period is one of the important stages in the vegetation growth process， so it is of great significance to explore the change of vegetation greening period and its influencing factors. At present， there are few studies that combine snow cover with temperature and precipitation to explore the influencing mechanism of vegetation greening period. Therefore， based on MODIS NDVI， snow cover products and ERA-5 reanalysis data， this paper adopts the Logistic curve curvature extreme value method， trend analysis， correlation analysis and sensitivity analysis of accumulated NDVI. To explore the spatio-temporal variation of vegetation greening period in Mongolia Plateau from 2001 to 2018 and its response mechanism to climate， snow cover and soil water change. The main results show that the average greening period of the Mongolian plateau in recent 18 years is about 123d， and the overall trend is not significantly delayed. The earliest greening stage was found in the southwest region and the Greater Khingan Mountains， and the latest was found in the Sayan Mountains and Hangai Mountains. The regions with significant positive correlation with snow cover， snow cover date and snow cover total day were all >30%， while the regions with significant negative correlation with temperature during snowmelt period were >25%， and the correlations with precipitation， soil moisture and snow cover first day were weak. The results showed that snow cover had a significant effect on the greening stage of vegetation， and its factor sensitivity was ranked as follows： snow cover （0.467） > snow cover date （0.184） > temperature during snowmelt period （0.113） > snow cover day （0.028）.

The Gaofen-6 medium resolution wide-width camera （GF6-WFV） is designed with two red-edge bands， which has the potential to monitor chlorophyll a concentration in water. In this study， six typical lakes in eastern China， including Guanting Reservoir， Luhun Reservoir and Baiyangdian Lake， were selected as the study area， and measured spectrum and chlorophyll a concentration data were obtained from 141 sampling points. Based on the measured data， the parameters of four kinds of commonly used semi-empirical inversion models of chlorophyll a concentration were optimized and the model accuracy verified， and the optimal inversion model was selected. The results show that the red edge band Ⅰ （B5：710 nm） and red band （B3： 660 nm） are newly added in GF6-WFV data. Which construct a two-band ratio 2BDA model with high inversion accuracy， correlation coefficient square （R²） is 0.89， the Mean Relative Error （MRE） is 34.71 %， and the Root Mean Square Error （RMSE） is 13.29 mg/m³. The results show that the chlorophyll a concentration in water body can be effectively retrieved by using GF6-WFV image data. The inversion model of chlorophyll a concentration in water body established in this paper based on multi-lake and multi-temporal data has good applicability in typical lake repositories in eastern China.

Studying the topographic differentiation characteristics and driving mechanism of land use landscape pattern is of great significance for land use optimization and landscape dynamic management. Yongding District of Zhangjiajie City， with complex terrain， various types of coverage and tourism interference， was selected as the research object. The landscape type maps of multiple years in the study area were superimposed one by one with elevation， slope and aspect classification maps， and classified and counted. Eight landscape indices such as Patch Density（PD）， Aggregation Index（AI）， and contagion index（CONTAG） were selected from the landscape level index and type level index to calculate the annual change of the index and explore its topographic differentiation law and driving mechanism. The results show that ： （1） The land use landscape types in the study area have obvious altitude gradient characteristics. More than 80 % of the land area is concentrated in the area with an altitude of 300 ~ 800 m and a slope of 6 ° ~ 35 °. （2） Whether the landscape level index or the type level index， the topographic differentiation characteristics are obvious， and the differentiation of elevation and slope is significantly higher than that of slope direction. （3） The evolution of land use landscape pattern in the area with large terrain gradient （ high altitude steep slope area ） is dominated by natural ecological evolution， while the evolution of the area with small terrain gradient （ low altitude gentle slope area ） is obviously disturbed by social and economic factors.

Based on Sentinel-2 remote sensing data， we selected three methods， including Supervised Classification （Maximum Likelihood Classification）， Machine Learning Classification （Random Forest Classification） and Phenological Feature Classification based on time-series NDVI， to extract Potamogeton crispus L.community in Nansi Lake in early May 2021. By using the measured area and distribution data of the Potamogeton crispus L. community in Nansi Lake， we analyzed the classification accuracy of the three methods during the same period， and analyzed the extraction effects of the three methods for Potamogeton crispus L. in combination with the Fractional Vegetation Cover （FVC）. The results showed that （1） there was a significant difference in the total area of the Potamogeton crispus L. extracted by three methods. The areas of the Potamogeton crispus L. community extracted by both Supervised Classification and Random Forest Classification were less than 100 km²， which were 98.97 km² and 75.92 km² respectively. While the area extracted by the time-series NDVI method was 207.44 km²， which was closest to the measured area of Potamogeton crispus L. （2） Both the whole lake and the core area， the extraction accuracy of Supervised Classification and Random Forest Classification was just about 75%， the Mean Relative Error （MRE） was about 0.5， and Mean Error （ME_area） was about 20~30 km²， while the accuracy of the time-series NDVI method was above 90% and the MRE and ME_area were also the lowest. （3） Comparing the fractional vegetation cover， we found that Supervised Classification and Random Forest Classification could only extract the Potamogeton crispus L. with high fractional vegetation cover near the lake shore and poorly with low cover in the lake core area， while the time-series NDVI method was more sensitive to the low fractional vegetation cover Potamogeton crispus L. community and could extract it well in different areas of the whole lake， which is a potential method for Potamogeton crispus L. remote sensing extraction. This study has some theoretical value for innovative remote sensing extraction methods of submerged vegetation and guiding remote sensing monitoring of lake ecological environment.

From text analysis to image interpretation， the Topic Model （TM） consistently plays a pivotal role. With its robust semantic mining capabilities， topic model can effectively capture latent spectral and spatial information from Remote Sensing （RS） images. Recent years have seen the widespread adoption of topic models to address challenges in RS image interpretation， including semantic segmentation， target detection， and scene classification. Thus， clarifying and summarizing the present application status of topic models in remote sensing imagery is pivotal for advancing remote sensing image interpretation technology. This paper initially presents the foundational theory of topic models， followed by a systematic overview of their typical applications in remote sensing imagery. In addition， experimental comparisons and analyses are performed across various typical remote sensing image interpretation tasks， illustrating the extensive applicability of topic models in the realm of remote sensing and the efficacy of distinct topic models in enhancing our comprehension of remote sensing imagery. Subsequently， we have outlined the limitations of topic models and explored the potential and prospects of integrating them with deep learning.

Remote sensing images have low detection accuracy due to the characteristics of different object angles， generally arranged densely， high proportion of small objects and complex background. In view of the inapplicability of the horizontal detection algorithm for remote sensing rotating object detection， and the periodicity and edge interchangeability of angle in the mainstream five-parameter method， a VR-CenterNet is proposed， which used the vector representation to detect the rotating box and design the loss function to avoid the problem of angle regression， and to optimize the high displacement sensitive problem of slender objects. For the high redundancy problem of shallow feature fusion， self-adaptive channel activation is introduced to automatically filter impurity information. In order to strengthen the key point information， an improved global contextual self-adaptive layer activation attention block is introduced in the output of backbone. First， the performance of different algorithms is compared on HRSC2016 and UCAS-AOD data sets. Then， the module ablation experiment is conducted on the two data sets to verify the effectiveness of each improved method. Experimental results show that： 88.48% and 90.35% accuracy are obtained on HRSC2016 and UCAS-AOD data sets respectively. The improved algorithm can improve the detection accuracy of remote sensing rotating objects， and provide another problem-solving idea for the accurate detection of remote sensing rotating objects.

Some of the data of the second Chinese glacier inventory（CGI2.0） are replaced by the first Chinese glacier inventory， and these data are concentrated in southeastern Tibetan Plateau. Where the terrain is steep， the climate is harsh， and it is covered by clouds all the year round. There is no systematic glacier survey due to the inability to obtain effective optical images. Aiming at the problems that the traditional threshold segmentation method is influenced by noise， and the standard Unet has a large amount of computation， which leads to slow operation， compressedUnet model is designed to improve model training efficiency and glacier extraction accuracy by modifying model parameters such as sample size， number of convolution kernel and optimizer. Using the polarization characteristics and topographic features of glaciers， 45-scene ENVISAT ASAR images and NASA DEM are selected to carry out deep learning based on Unet and compressed Unet. By referring to optical images and other auxiliary data， the misclassified and missed glaciers are visually interpreted one by one. Finally， the extraction and correction of the glacier boundaries without update are completed， and their attributes are updated. The results show that deep learning based on SAR images and topographic features can effectively identify glaciers in cloud-covered areas. In the areas where the CGI2.0 is not completed， there are 8 374 glaciers with a total area of 5 622.65±303.58 km²， and the error accounts for 5.4% of the total glacier area， most glaciers are retreating and fragmenting. The dataset updates the alternative data in CGI2.0， and provides reliable data support for related studies of glacier changes and mass balance in southeastern Tibetan Plateau.

UAV remote sensing technology can quickly obtain the Canopy Height Model（CHM） of the survey area. How to identify tree vertices more accurately from CHM is key to tree height extraction. This paper discusses the influence of different window types， window sizes， and stand canopy density on the extraction of tree vertices. Using the university campus as the study area， two local areas of dense and sparse forest land were selected based on canopy density. GIS rectangular neighborhood analysis， GIS circular neighborhood analysis， and local maximum algorithm are used to extract tree vertices. The results show that the accuracy of tree vertex extraction is not only affected by the window size and canopy density， but also closely related to the window type， and the result of GIS rectangular neighborhood analysis to extract tree vertices is more stable and accurate， and the highest F-Measure value is 78.13% in dense forest， and 96.94% in sparse forest. Comparing the extracted tree heights corresponding to the tree vertices obtained based on this result with the tree height values measured in the field， the RMSE is 37cm for dense forest and 39cm for sparse forest. The results proved the feasibility of extracting tree heights of broad-leaved forests with higher canopy density based on the visible light remote sensing technology of small UAVs， which provided a reference for the subsequent identification of tree vertices based on the canopy height model and improved the accuracy of tree height extraction.

Namcha Barwa region is located in the core tectonic deformation zone of the Eastern Himalayan Syntaxis with a complicated geological-tectonic environment and frequent geohazards. Therefore， it is of great significance to strengthen the research of surface deformation monitoring in this area for local disaster prevention and mitigation and sustainable economic development. This study aims to monitor surface deformation using Sentinel-1 SAR images in this region. Using PS-InSAR technique， the surface deformation rates distribution and deformation time series on LOS （Line-Of-Sight） were acquired. Then the status of surface deformation distribution and coseismic deformation caused by Mainling M6.9 earthquake in 2017 were discussed. It is revealed that the deformation in Namcha Barwa is greatly affected by Cenozoic tectonic deformation. Tectonic deformation in the study area mainly included coseismic deformation， postseismic relaxation deformation and thrust deformation in plate boundary. The deformations were quite different on both sides of the Yarlung Zangbo River. A slow negative deformation trend is shown on the north side， while the south side is positive deformed at a high rate caused by thrust faults. The coseismic deformation of Mainling earthquake showed a spatial distribution trait of negative deformation on the southeast side of the epicenter， positive deformation on the northeast side， positive and a larger deformation on the southwest side. This study demonstrated that， InSAR can provide high spatial and temporal resolution surface deformation data for hazard monitoring and scientific research on Qinghai-Tibet Plateau.

Sea ice classification using Synthetic Aperture Radar （SAR） images is a crucial aspect of sea ice monitoring. Existing methods have mainly relied on spatial features of SAR images， but rarely consider temporal features， which can potentially provide additional information. A novel approach called SE-ConvLSTM has been developed to combine both spatial and temporal features for sea ice SAR image classification. Firstly， ConvLSTM is used to extract the spatial-temporal features of HH and HV polarization SAR images respectively. Then， the spatial-temporal features of different layers and channels are concatenated， and the channel feature response is adaptive recalibrated by using SE channel attention. Finally， SoftMax function is used for image classification. To evaluate the effectiveness of the SE-ConvLSTM method， six time-step image blocks of SI-STSAR-7 dataset were used for comparison with other classification methods. The results indicate that SE-ConvLSTM achieved an overall accuracy of 97.06% and 90.01% for the thick one-year ice which is difficult to classify. This suggests that adding temporal information can significantly improve classification accuracy. Additionally， the proposed network has better recognition ability for regions with low density of main ice types and for boundary positions of SAR images， making it an effective tool for generating sea ice distribution maps.

To reveal the long-term variation trend of vegetation cover and further determine the main climatic driving factors affecting vegetation variation in Hainan Island. Also， to provide scientific evidence related to the impact of climate change on vegetation and scientific basis for achieving vegetation optimum development in island regions. The spatiotemporal variation trend of vegetation in Hainan Island from 1982 to 2015 was explored by applying trend analysis method to the GIMMS NDVI data. The effects of temperature， precipitation， and solar radiation on vegetation variability in Hainan Island were investigated by partial correlation analysis and principal component regression analysis over the 34 years. Results show that： ①Spatially， vegetation exhibited a significant increasing trend in the northern and coastal regions of Hainan Island while displayed a degeneration trend in Sanya city and its surrounding areas. ②Temporally， we found the vegetation in Hainan Island showed a slowly increased trend in most areas with a speed of 0.019/10 a and its intra-annual variability was obvious. ③In general， temperature and solar radiation jointly dominate the vegetation growth in 88% areas of Hainan Island in a significant way. Solar radiation was the most important climate driving factor to control vegetation variability in Hainan Island， followed by temperature， and precipitation had a small impact on the vegetation variability. ④Temperature dominated the vegetation variability in the northern and western areas of the Hainan Island. By contrast， solar radiation dominated the vegetation variability in the southern areas of the Hainan Island. Precipitation was the dominant climate driving factor to explain the variability of forests in the middle of the Hainan Island. Overall， this study found that temperature and solar radiation were two major climate driving factors which affected vegetation growth in the Hainan Island.

As black and odorous water bodies in rural areas have negative influence to environment， it is important to monitor the rural black and odorous water bodies by high resolution remote sensing. While， the spectral curve from remote sensing of rural black and odorous is similar to some vegetation， green roofs and greenhouses， which bring difficulties to identify the rural black and odorous in remote sensing images with satisfactory repeatability and accuracy， and automation， by using the color purity on a Commission Internationale de L’Eclairage （CIE） model and spectroscopic method. Thus， we collected and interpreted 325 rural black and odorous water bodies by GF1/2/6， covering several counties in Xi’an and including various type of polluted object， to train the model using DeeplabV3+ with ResNet101 as the backbone to identify the rural black and odorous water bodies， in which we imported the Efficient Channel Attention （ECA） and pre-processed the samples by increasing the brightness and correcting the color difference. The F1-score， MIoU （Mean Intersection over Union）， IoU （Intersection over Union） and FOR （False Omission Rate） of the model were 0.931， 0.935， 0.935 and 0.085 respectively， which indicated that the model could efficiently， accurately， and repeatedly identify rural black and odorous water bodies from high-resolution remote sensing images and offer assistance for government departments to regulate rural black and odorous water bodies.

Sichuan is one of the regions with frequent extreme precipitation in China. Based on the grid precipitation dataset （CMPA） provided by China Meteorological Administration as the reference of surface precipitation. We systematically evaluated the accuracy of extreme precipitation monitoring in Sichuan province by using a variety of extreme precipitation index including GSMaP （NRT， MVK， Gauge） and IMERG （Early， Late， Final） satellite precipitation products. The results are as follows： （1） The frequency and intensity of extreme precipitation in the basin and its surrounding areas are significantly higher than those in other areas. There is an obvious boundary line of extreme precipitation along the basin and the plateau. （2） IMERG products can detect rainstorm， continuous precipitation events and drought events more accurately， of which Final product have the highest precision， and the precision of extreme precipitation index of IMERG products is significantly better than GSMaP. （3） The results of probability density distribution function （PDF） show that the PDF characteristics of IMERG-Final products and CMPA are the most similar， followed by GSMaP-Guage， However， the Gauge product corrected by the site completely changed the PDF characteristics of NRT and MVK， ignoring many heavy rain and rainstorm events， which will be extremely unfavorable to the monitoring of extreme precipitation. In general， the detection accuracy of GPM products for extreme precipitation has great regional differences. IMERG products show higher extreme precipitation detection accuracy than GSMaP. We also found that there are large precipitation errors in IMERG and GSMaP products in the Western Sichuan Plateau with complex terrain.

A brief review has been conducted on the progress of typical spaceborne and airborne polarimetric Synthetic Aperture Radar（SAR） systems at home and abroad， for which the implementation of radiometric and polarimetric calibration accuracies has been focused and surveyed. First the general requirements for the polarimetric SAR data calibration accuracy have been drawing from literature research， and then the status quo of representative polarimetric SAR systems in the word and the system data calibration accuracy achievements have been systematically presented， including the relative radiometric calibration accuracy， the absolute radiometric calibration accuracy， the polarization channel crosstalk accuracy， the polarization channel amplitude imbalance accuracy， and the polarization channel phase imbalance accuracy， etc. Finally， the key factors affecting the calibration accuracy of polarimetric SAR data have been analyzed， and the future calibration tasks meeting the new polarimetric SAR system design has been briefly discussed. This paper comprehensively describes the calibration accuracy information index of polarimetric SAR systems and their development status， and provides relevant researchers with timely， comprehensive and systematic information on the development requirements of polarimetric SAR systems and the research progress of calibration accuracy achievements.

Forest is a valuable non-renewable resource， but the ecological environment of forest is seriously threatened by many natural or man-made factors such as fire， flood， and deforestation interference. Accurate grasp of forest resource changes can provide effective information for forest resource management and protection. In the task of forest change detection， traditional machine learning change detection methods have difficulty in capturing deep semantic information due to large differences in forest categories and tree species， and suffer from poor adaptability of extracted features， weak recognition ability， and pseudo-change due to seasonal phases. We propose to build a deep learning model with Siamese neural networks for forest change detection experiments. Three different feature extraction methods， ResNet50 （Residual neural network）， CBAM （Convolutional Block Attention Module） and SE （Squeeze and Excitation） with different lightweight attention mechanisms are used as backbone feature extraction modules， respectively. All three backbone feature extraction networks are trained based on pre-trained weights， which improve change detection by fusing the extracted multi-scale feature maps so that the coarse and fine details of information in different feature maps complement each other. It also has the advantage of sharing weights with the same number of parameters. Taking Jiande Forest Farm in Zhejiang province as the experimental area， two phases of GF-2 images in 2015 and 2020 are acquired to construct a forest change detection dataset with a resolution of 1m. The results of Siamese neural network change detection are compared with the true change labels （Ground truth）， where the backbone feature extraction network SE-ResNet50 has the best combined results with Precision （0.91）， Recall （0.78） and F1-score （0.83）， which is better than mainstream change detection models FC-Siam-conc， FC-Siam-diff. It is proved that Siamese neural networks can accurately capture forest changes in the task of forest lad change detection from high-resolution remote sensing images， and provide a new forest change detection method for forest resource management departments.

Remote sensing Visual Question Answering （VQA） is to answer natural language questions related to image content based on a given remote sensing image， which is essential for fast investigating and monitoring global resources. With the complexity and diversity in remotely sensed imagery， the scale variation is unequivocally challenged in the observation of images from understanding global scenes to identifying local objects. To address the problem of scale variations in the remote sensing visual question answering system， in this paper， a new model Multi-scale Remote Sensing Visual Question Answering（MRS-VQA model） and a dataset （MRS-VQA dataset）， which include multi-scale scenes of question-answer pairs of remote sensing images， are created. In addition， the attention mechanism is employed in the fusion module of the MRS-VQA model to show the visualization results of the combination of two modalities， which effectively improves the accuracy and interpretability of the model. Experimental results illustrate that the proposed MRS-VQA model with two attention layers （96.82% accuracy） outperforms other remote sensing visual question answering models （81.36% accuracy on RSVQA）， which means that multi-scale feature fusion is of great significance in remote sensing VQA.

Baihetan Hydropower station is the second largest hydropower station in China after the Three Gorges Hydropower Station. With the rapid change of water level during the reservoir impoundment period， the geoenvironment of the reservoir area is changed， which is easy to cause sudden geological disasters. In addition， the maximum reservoir level of the transformed hydropower station is 825m， which has a large storage capacity and higher sudden-onset. In order to ensure the normal impoundment of the hydropower station and the safety of residents' life and property， it is necessary to quickly and dynamically identify geological hazards in the reservoir area. Therefore， based on the short baseline DInSAR method， this paper carried out the identification of geological hazards in the key reservoir bank section of Baihetan Reservoir area （Hulukou-Xiangbi Ling section） during the water storage period of the Jinsha River Basin. The sentinel-1A /B data were used for joint monitoring， and the re-entry period was increased to one scene every 6 days. Finally， a total of 66 scenes of data were obtained during the study area's water storage period from April 2021 to November 2021. DInSAR processing based on the same main image was performed combined with SRTM DEM data. The fast and dynamic deformation monitoring can be achieved by combination analysis of interference pair. Don't out of the 92 consensus exists significant deformation of strong deformation zone， are distributed in Jinsha River two sides， including eight strong deformation area have been inundated by the highest water level 800 m， 38 strong deformation area near the water， part of the small bank collapse， at the same time has chosen three continuous deformation area in combination with the actual topography analysis， found signs of deformation is more obvious， The location of strong deformation area is consistent with the monitoring results of DInSAR method， which proves the feasibility of using DInSAR method to quickly and dynamically discover new hidden points. The disaster analysis and monitoring and early warning during water storage period is of great significance to ensure the normal water storage and power generation of Baihetan Hydropower Station. The method provided in this paper can be used for large-scale， efficient， rapid and dynamic geological disaster monitoring， and the hidden danger points can be found in the first time and qualitative analysis can be carried out， which provides a new idea for the identification of potential geological disaster caused by the change of water level during the reservoir bank impoundment period.

Mangrove forests are among the ecosystems with the highest net primary productivity in the world， and they play an important role in the study of global climate change and the evolution of coastal zone geography. Rapid and accurate acquisition of the spatial distribution of mangroves on a large scale is vital for effectively managing and exploiting mangrove resources. Landsat satellite images have become an important data source for extracting large-scale and long-period mangrove distribution information. Yingluo Bay and Pearl Bay along the coast of Guangxi， China are selected as the study sites in this study. Landsat-8 OLI images are used to construct five indices to extract the distribution of mangroves， including Normalized Difference Mangrove Index （NDMI）， Combined Mangrove Recognition Index （CMRI）， Modular Mangrove Recognition Index （MMRI）， Mangrove Index （MI） and Mangrove Vegetation Index （MVI）. This study compared the efficiency of different indices used for mangrove extraction to determine the optimal mangrove extraction index. Optimizing the mangrove distribution information extraction is proposed by combining Normalized Difference Water Index （NDWI） index. The aim is administrator improve the remote sensing classification accuracy of mangroves. It is also applied to the extraction of coastal mangroves in Guangxi. The results showed that： Mangrove distribution can be effectively extracted based on Landsat-8 OLI satellite images and index method. By comparing the extraction accuracy of five indices of mangroves， we found that the MVI has the best extraction effect and the CMRI has the worst extraction effect. The combination of NDWI can better optimize the extraction accuracy of mangroves， and the optimized MVI applied to Guangxi coastal mangroves showed the best extraction results with an overall accuracy of 97.10%. The research strategy and the range of mangrove index thresholds in this paper can provide reference and decision support for large-scale mangrove distribution extraction.

The Tibetan Plateau （TP），with its unique climate characteristics and geographical pattern， plays an important role in global climate change. As an important part of the earth atmosphere system， cloud is key to affecting climate change. Cloud cover can more directly reflect the change of cloud. Therefore， it is of great significance to reconstruct a cloud cover product with longer time series and higher accuracy in the TP. In this paper， Considering the complex underlying surface types and geographical elevations in the TP， We select the cloud cover of MOD06， ERA5 and CRA40 from 2001 to 2020. We take the cloud cover of MOD06 from March to November as the true value and evaluate the applicability of the two reanalysis data in the TP through methods such as climate tendency rate and correlation coefficient.Based on ERA5 and MOD06， the improved auto-encoder model is used to reconstruct the cloud cover of the plateau from March to November of 1950 to 2020. The results show that the cloud cover of ERA5 is higher than that of MOD06， while that of CRA40 is lower than that of MOD06， and the correlation between ERA5 and MOD06 is obviously better than that between CRA40 and MOD06；The improved auto-encoder model evaluated by four evaluation indicators of correlation coefficient（R）， bias， Mean Absolute Error（MAE） and Root Mean Square Error（RMSE） has a good effect on cloud amount reconstruction. The correlation coefficient between cloud amount reconstructed by the improved autoencoder model and MOD06 cloud amount data increases by more than 20% on average from March to November， and can simulate the change trend of cloud amount over the TP. The results provide reliable long time series data for studying the temporal and spatial evolution of cloud cover over the TP.

The existing detection research of abnormal water bodies is usually carried out for specific regions， specific data sources and specific time phases. Anomaly recognition algorithm testing is often a backtracking of the water body anomaly events that have occurred， rather than real-time monitoring of the anomaly events， which cannot serve the requirements of rapid detection and identification of water body anomalies. In this paper， a method of extracting water body abnormal information based on unsupervised isolated forest plus decision rule （U-IForest-SD） is proposed. We selected Landsat and Sentinel as the test data， and tested the accuracy of U-IForest-SD with the black and smelly water body of Qingdao Enteromorpha， Songya lake and the oil spill in the Gulf of Mexico as research cases. We also compared U-IForest-SD with SVM and supervised isolated forests. The results show that the overall accuracy of the proposed method for the three types of anomalies is above 90%， and the kappa coefficient is above 0.8. The overall accuracy is higher than that of supervised isolated forest but slightly lower than that of SVM. This algorithm only needs to input single phase images， and does not need training samples. It has the advantages of good portability， strong universality and high automation. In addition， it can effectively avoid the occurrence of "wrong alarm" and "false alarm". Therefore， the newly proposed method has a good application prospect in the rapid detection and identification of abnormal water bodies.

In order to reduce flood related disaster risks， it is necessary to quickly and accurately obtain the flood area， and use relevant data and information of flood events to analyze flood susceptibility areas， which can provide a scientific basis for flood prevention decision-making and management. Based on the Sentinel-1 SAR data， we analyzed the flood inundation status and susceptibility of the concentrated areas of the flood storage and detention areas in the middle and lower reaches for the Yangtze River in July 2020. The research results showed that the water body area during the flood period in 2020 reached 3 747 km² in the flood storage and detention area. Compared with the normal water period， increased about 1 301 km² water body in the flood period， accounting for 19% of the new water body area in the entire study area. In all flood storage detention area the largest new water body area is Honghu and Huayang River storage flood detention area. Judging from the changes of land use since 2010 within the new water bodies area， the area of cultivated land and man-made land has expanded， the area of wetland， woodland grassland， and bare land has decreased. So the shrinking area of wetland may reduce the capacity of water storage and flood reception. The area of high flood susceptibility in the study area accounts for 23.33% of the entire study area， and the extremely high level of flood susceptibility accounts for 22.55% of the entire study area； the area of high flood susceptibility in the flood storage and detention area accounts for 38.97%， the extremely high level accounts for 52.05% of the entire flood storage and detention area. The research results can provide scientific basis and theoretical reference for flood control， planning and construction in flood storage and detention areas.

This study， focused on the area of Beibu Gulf， explores the remote sensing inversion method for chlorophyll concentration based on the Sentinel-3A satellite's OCLI water color sensor. The study partitions the Beibu Gulf by using measured spectral data and then combines the measured chlorophyll-a concentration with Sentinel-3A remote sensing data of which aims to build the remote sensing inversion model for chlorophyll-a concentration. The results show that （1） the remote sensing reflectance curves exhibit distinct partition characteristics， dividing the area into nearshore， transitional， and offshore water types based on the spectral features； （2） Different water types require different inversion factors for model construction， and all of them got relatively good fitted result. Among them， the fitted inversion factor is R_rs（764.375）/R_rs（681.25） that could be used in the nearshore water， for the transitional water， ［1/R_rs（620）-1/R_rs（708.75）］/R_rs（753.75） is the most suitable， and for the offshore water， R_rs（708.75）-R_rs（764.375） achieves the best fitting performance， with corresponding R² values of 0.67， 0.80， and 0.8， respectively； （3） The partitioning method effectively improves the applicability and accuracy of the remote sensing inversion model for chlorophyll concentration in the Beibu Gulf. This study successfully realizes the remote sensing inversion of chlorophyll concentration in the Beibu Gulf by using a partitioning model based on Sentinel-3A satellite's OCLI data. The result provides the important scientific support for the remote sensing monitoring of chlorophyll concentration in the Beibu Gulf， and enhances the management and protection of marine ecological environments.

In recent years， the abundance of hydro fluorocarbons （HFC） has been increasing， which has huge greenhouse potential value. It has an impact on global warming and also indirectly causes the destruction of the ozone layer. Scholars at home and abroad have carried out a wide range of in-situ ground measurements to obtain global abundance. At the same time， remote sensing technology can monitor the changes of HFC gas in a large range， for a long time and quickly， and has become an important means for the inversion of the gas concentration. The contents of in-situ measurement method， tracer ratio method， satellite inversion sensor development and satellite inversion method are described， and the advantages and disadvantages of different inversion methods are compared in combination with load characteristics analysis. Finally， discusses and prospects the existing problems and future development trend of current inversion.

The concentration of suspended matter in water body is an important parameter to describe the optical characteristics of water body. Satellite remote sensing has the advantages of a large range， fast and high-frequency word dynamic monitoring， which helps to strengthen the monitoring of water environment quality of Qinghai Lake and reduce the monitoring cost. And ZY1-02D satellite hyperspectral camera with high spatial resolution and high spectral resolution provides the possibility of high-precision monitoring of water quality in Qinghai Lake. In order to verify the applicability of the ZY1-02D hyperspectral camera in the application of remote sensing monitoring of water quality， this paper uses the ZY1-02D hyperspectral camera as the remote sensing data source， and also assists the actual measurement data to construct an inversion model of the suspended matter concentration in Qinghai Lake， and conducts accuracy verification to evaluate the accuracy of the inversion results. The results show that the average relative error of the Qinghai Lake suspended concentration inversion model is 21.1%， and the root mean square error is 0.296 mg/L. The accuracy is good， and the inversion results of Qinghai Lake suspended concentration show the characteristics of low in the center of the lake and high on the shore， compared with the retrieval results of Sentinel-2 and Landsat-8 in the same period， the retrieval results of Sentinel-2 and Landsat-8 in the same period， the inversion results remain consistent， results remain consistent， which indicates that the ZY1-02D hyperspectral image can retrieve the water quality parameters.

Snow Water Equivalent （SWE） is defined as the height of melting snow， regarded as the most important variable describing the amount of snow. Space-borne passive microwave （PMW） remote sensing is currently viewed as a attractive option for SWE estimation at global scale. One challenge for SWE estimation is that most inverse algorithms were built based on a non-comprehensive RTM in which it neglects other components， e.g.， forest. Forest canopy not only attenuates the microwave radiation from soil， but also emits some radiation into the sensor. Therefore， forest canopy increases the uncertainness of SWE retrievals. This paper aims to study the sensitivity of forest parameters （transmissivity， cover fraction， and single albedo） to brightness temperature based on a proposed systematic RTM， and further analysis the robust of SWE algorithms to brightness temperature noise.To better describe the transmission of electromagnetic wave in a real complex contracture， a systematic RTM was built firstly， considering the soil， snow， forest and atmosphere. The Advanced Integral Equation Model （AIEM） model was applied to simulate soil-snow boundary reflectivity. A semi-empirical radiative transfer theory （HUT） model was applied to simulate snow microwave emission. A zero-order tau-omega model was used to describe the interactions between snow and forest canopy. Then a database based the built RTM was generated， and applied to conduct sensitivity analysis of forest and snow parameters to brightness temperature. Meanwhile， noise tests of brightness temperature to SWE retrieving algorithms were done based modeling data and satellite observations.The results indicate that canopy transmissivity is the most sensitive factor among three forest parameters， secondly forest fraction， and lastly single albedo. Meanwhile， the brightness temperatures raise with increasing forest fraction， but decline with increasing canopy transmissivity and snow grain size. Namely， there is ‘neutralize effect’ among these three parameters （forest fraction， canopy transmissivity， and snow grain size）. This is because forest canopy attenuates the radiation from snow and the brightness temperature （Tb） of snow is typically lower than canopy radiation. Thus， the Tbs in forested areas are higher than those in open areas.The noise analysis based on modelling data and satellite observation shows that the influence of brightness temperature uncertainness on AMSR2 SWE retrieving algorithm is more serious than FY-3B method， especially in forest areas. This is maybe because a polariton index charactering grain size evolution is empirical. The influence of brightness temperature uncertainness on FY-3B algorithm is very small， even can be negligible.The paper first quantitatively investigated the sensitivity of forest （transmissivity， fraction， single albedo） and snow （grain size） parameters to microwave brightness temperature based on the built systematic ourselves. What's more， this study determined the key parameters affecting SWE retrievals， including canopy transmissivity， forest fraction and snow grain size. Additionally， the noise analysis reminds us the robust of algorithms to brightness temperature uncertainness must be considered， not just their accuracy. This study provides the theoretical basis and direction for improving SWE in the near future.

In the ' post-poverty era ' of 2020， the four prefectures in southern Xinjiang are still underdeveloped areas in the development pattern of the whole country and in Xinjiang.Therefore， it is of great significance to carry out long-term economic measurement and development analysis of the four prefectures in southern Xinjiang. However， the traditional measurement methods using socio-economic data have great limitations.The nightlight remote sensing data is used to objectively invert the economic development characteristics of the poverty-stricken area.This paper selects four prefectures of southern Xinjiang as the study area， and corrects NPP/VIIRS and DMSP/OLS data for noise and supersaturation respectively， based on the integration of two kinds of night light data， the correlation between the total night light amount of 33 counties （cities） and the secondary and tertiary industries was used， the spatial and temporal pattern of economic development in the four regions of southern Xinjiang from 2005 to 2020 was studied by using standard deviation ellipse and Molain index.The results show that ： （1） From 2005 to 2020， the economic center of gravity moved to the northwest. The total economy is dominated by the northeast-southwest direction， the economic development trend is more and more concentrated and contiguous.（2） There has been a high spatial autocorrelation and aggregation during the study period， mainly showing H-H and L-L gathering areas. The higher the economic level of the region， the more prone to aggregation.（3） The probability of occurrence of cold spots and hot spots in regional economic development in the four prefectures of southern Xinjiang has obvious local characteristics. The coordinated economic development among the four regions of southern Xinjiang should be the policy focus.

Snow albedo plays a very important role in the study of global radiation and energy， water cycle and climate change. It is of great significance for the application and improvement of snow albedo products to make clear the accuracy and uncertainty of snow albedo. In the past， the evaluation of albedo products mainly focused on the snow-free surface or the permanent snow-covered surface， but the seasonal snow cover is also widely distributed in the world， and has the characteristics of dynamic change. Therefore， the accuracy of the new version of MCD43 albedo product under the seasonal snow cover is verified and analyzed by using the adequate spatial representative snow albedo data of FLUXNET ground site.Results in the study area indicate that compared with non-snow surface， the missing rate of MCD43A3 albedo products under snow cover is higher， up to 24.95%， and the proportion of MCD43A3 albedo products using backup inversion algorithm is also higher， up to 78.06 %. Compared with the non-snow surface， the accuracy of snow albedo products under all stations decreased， with the minimum the Root Mean Square Errors （RMSE） of 0.101 5 and the maximum RMSE of 0.238 7. In the snow accumulation period， the accuracy of the MCD43A3 albedo product is closely related to the surface type， among which the accuracy of the evergreen coniferous forest is the lowest， with the maximum RMSE of 0.238 7 and the maximum relative bias（biasR） of 88.88%. In addition， the shortcomings of kernel function model and backup inversion algorithm used in MCD43A3 in snow albedo product production are illustrated by examples. It is found that the effective recognition and inversion ability of the product is insufficient in snowfall， snowmelt stage and "short stay of snow" event.

Due to the complex terrain characteristics of the Three Gorges Reservoir area and the shadow coverage in some areas， the detection of water body changes is prone to missed detection and false detection. In view of this situation， in order to improve the accuracy of water body change detection results， a water body change detection algorithm in the Three Gorges Reservoir area is proposed that integrates Siam-U-Net++ and scSE attention mechanism. The Siam-U-Net++ with shared weights in the encoding stage is used as the backbone network， and the ResNet34 network with residual structure is selected on the encoder network unit to quickly and efficiently obtain the characteristic information of water body changes. The scSE attention mechanism module combining channel attention and spatial attention is introduced after the upsampling of Siam-U-Net++. On the self-made data set， it is tested with LinkNet， U-Net++， DeeplabV3+ network models and NDWI， and different attention mechanisms are used for comparative experiments. The recall rate， accuracy rate and F1 value were 94.57%， 90.75% and 92.62% respectively， which were better than other models. Experimental results show that the algorithm can effectively improve the results of water body change detection.

Change detection from heterogeneous remote sensing images is an important and challenging research topic with a wide range of applications in disaster assessment， urban planning and environmental monitoring. However， the direct comparison of heterogeneous data for change detection always has a poor detection accuracy. To address this issue， a multioutput adaptive regression and association-based feature fusion method for heterogeneous remote sensing change detection is proposed. Firstly， the proposed method determines the adaptive regression direction according to the information entropy， which utilizes the difference of information between heterogeneous data. To transform heterogeneous data into a common feature space， the regression image will be obtained via a multioutput multilayer perceptron image regression algorithm. Then， the fuzzy local information C-means algorithm is used to identify the fuzzy region in the difference image， which further ensures the reliability of significant sample pairs. Finally， an association-based fusion method was applied to the heterogeneous remote sensing change detection dataset by simultaneously exploiting the high-order information of heterogeneous data and the association information between features. The binary change map is obtained via training a classification model with the boosting dataset. Experiments conducted on three real datasets （Sardinia， Yellow River and Texas） show the effectiveness of the proposed method by comparing it with seven related change detection methods. Experimental results indicate that the proposed method owns the best change detection results on both three datasets， which proves its effectiveness， and it can suppress the influence of noise and improve the accuracy of change detection.