Straw burning is an important part of biomass combustion， which not only leads to waste of straw resources， but also causes serious harm to the environment. The traditional monitoring methods of straw burning are mainly manual inspections， with limited monitoring scope and high consumption of human and material resources. Remote sensing technology， as a new means of surface information monitoring， has brought development opportunities for large-scale monitoring of straw burning. This paper introduces the basic principles， monitoring methods and research progress of remote sensing technology in straw burning fire point monitoring， burning area estimation and burned area monitoring， and analyzes the deficiency of remote sensing technology in the application of monitoring straw incineration. On this basis， the future development of remote sensing monitoring of straw burning is prospected from four aspects： multi-source data fusion and complementation， optimization and integration of monitoring methods， in-depth mining of monitoring information and decision-making service of spatiotemporal information.
Object detection has always been a hot topic in the field of remote sensing images information extraction， and has a wide range of application prospect in many fields. The development of deep learning in the field of computer vision provides a strong technical support for the extraction of massive remote sensing images， and greatly improves the accuracy and efficiency of object detection in remote sensing images. However， objects in remote sensing images have the characteristics of multiple scales， multiple rotation angles and complex scenes， deep learning technique still faces great difficulties in the application of remote sensing images object detection with limited high-quality labeled samples. According to five aspects of scale invariance， rotation invariance， complex background interference， limited training samples and detection of multi-band data， the existing algorithms of object detection based on deep learning in the field of remote sensing images in recent years are introduced and summarized. In addition， the difficulties and methods of detecting typical objects in remote sensing images are analyzed and summarized， and the common datasets of remote sensing images object detection including optical images and SAR images are also given general introduction. Finally， the future trends of object detection in remote sensing images are analyzed.
Remote sensing satellites can quickly and dynamically acquire high-resolution images of large disaster area， and thus has become one of the main technical methods for post-earthquake damage investigation. This paper focuses on the optical remote sensing data and change detection algorithms widely used in post-earthquake disaster surveys. The satellite remote sensing data and products are summarized， and then the application of change detection algorithms using both pre- and post-earthquake images for damage investigation is reviewed. The basic principles， advantages and disadvantages of pixel-based and object-oriented change detection methods are described. The existing research results are classified and reviewed， and the problems and deficiencies in practical applications are discussed and summarized， with a view to providing references and benefits for future post-earthquake damage investigation work.
Fast and accurate access to provincial carbon emission data is the premise of real-time development of differentiated carbon emission reduction policies. Based on the DMSP/OLS and NPP-VIIRS night lighting data， the statistical data comparison method was used to extract the total nighttime light value （Expressed by TDN） of provincial built-up area in China's mainland （excluding Tibet） from 1997 to 2017， and the carbon emission prediction models of provinces were established by using the TDN values of 1997 to 2014 and the carbon emissions in the same period. Then， the TDN value from 2015 to 2017 is used as the independent variable to estimate the carbon emissions of China's provinces； at the same time， the total carbon emissions of China published by four international authoritative databases （IEA， EIA， EDGAR and CEADs） are allocated to each province by using entropy method and carbon emission allocation model. Finally， the estimated results are compared with the provincial carbon emission values assigned by four typical carbon databases. The results show that the estimated provincial carbon emissions are generally consistent with the allocated provincial carbon emissions， and the Mean Absolute Percentage Error （MAPE） is only 6.45%~9.12%. Meanwhile， the provincial carbon emissions estimated based on night light data are closer to the carbon emission values assigned by IEA and EIA databases. The estimated and allocated carbon emissions of each province fall near the 1∶1 line； the MAPE value of a single province varies from 0.68% to 14.85%， and the MAPE values of most provinces are within 10.0%. The above results prove the feasibility and accuracy of estimating provincial carbon emissions by extracting TDN values based on night light data.
Amu river delta， as a typical arid land， was threatened by drought and salination， which contribute to the complexity and specificality of its ecological environment. In the Land Use/Land Cover （LULC） Remote Sensing （RS） image classification tasks， collecting large number of high quality samples at low-cost and a high efficient and robust classifier are always the crucial factors to obtain high-accuracy classification results. However， it was still problems facing RS imageries classification in some remote areas that marked samples were sparsely distributed， timely dissected or even intermittent， and manual tasks for field sampling cost high. In this end， a new frame of automatic land cover classification based on ensemble of optimum trees and sample transfer was promoted in this paper. In this frame， sample labels were marked on the historical image which is same time and source with the product， then these labels were transferred into targeted RS image. Then， OTE method classification was performed. According to the results in this paper， the OTE with sample transferring based method can extract land cover labels efficiently and update LULC map in a fine accuracy.
The cities of Xining and Lhasa are hot spots of human activities on the Qinghai-Tibet Plateau， and their development process has an important impact on the socio-economic development of the Qinghai-Tibet Plateau. This paper reconstructed urban expansion in Xining and Lhasa cities based on remote sensing images， urban planning maps and historical maps in circa 1949， circa 1978， 1990， 2000， 2010 and 2018， and urban impervious surface and green space component information since 2000. We analyzed the temporal and spatial characteristics of urban expansion in built-up area of Xining and Lhasa since circa 1949， and revealed the characteristics of the impact of socioeconomic factors and policy factors on urban land use/cover changes. The results show that： （1） Since the founding of the People's Republic of China， built-up area of Xining and Lhasa has continued to expand， showing a non-linear growth trend. The urban land area has increased from 1.98 km2 and 1.10 km2 in circa 1949 to 79.26 km2 and 77.04 km2 in 2018. The urban expansion of Xining’s built-up area presents a cross-shaped expansion trend， and Lhasa presents a circle-extensive expansion mode； （2） Since 2000， the urban greening level of Xining and Lhasa has improved significantly. From 2000 to 2018， the urban impervious surface area of Xining and Lhasa increased from 36.91 km2 and 21.56 km2 to 55.34 km2 and 48.21 km2， and the urban green space area increased from 10.78 km2 and 8.48 km2 to 19.21 km2 and 20.35 km2， and the average annual expansion rates were 0.47 km2/a and 0.66 km2/a. The percentage of impervious surface in the built-up area has dropped from 74.09% and 66.21% to 69.82% and 62.58%， and the proportion of urban green space has increased from 21.64% and 26.05% to 24.24% and 26.41%； （3） Xining and Lhasa's urban population growth， economic development and relevant national policies are closely related with the urban expansion of the main urban area and its land use/cover changes. The urban expansion stage of the built-up area is related with the population growth， economic development stage and the implementation time of relevant national policies. Land use/cover change in built-up area is highly related with urban planning-related policies， especially landscaping construction， which has significantly increased the area of urban green space， and the proportion of urban green space has increased significantly compared with 2000.
Ensemble Learning （EL） classification method has become a research hotspot of land cover classification in recent years. Boosting Ensemble Learning classification method has high classification accuracy and strong generalization ability particularly， which has been significantly applied in land cover classification. However，Boosting Ensemble classification method is sensitive to noise. If the training sample contains noise， Boosting algorithm may lose effectiveness， which is the limitation of the method. In order to solve the problems existing in Boosting Ensemble method in the classification of land cover，effectively overcome the influence of noise， reduce the salt and pepper phenomenon in the classification results and improve the classification accuracy， a Boosting Ensemble Learning classification method based on the dual-tree complex wavelet transform is proposed. In this method， the spectral band of the image is transformed by a layer of dual-tree complex wavelet to reduce the image noise. The extracted low-frequency features are taken as the input of Boosting Ensemble Learning to obtain the final classification result. Boosting Ensemble Learning GBDT， XGBoost and LightGBM algorithms are respectively compared classification accuracy and efficiency for SPOT6 and Sentinel-2A image. The results show as follow： （1）For SPOT6 image， the overall classification accuracy of the three Boosting Ensemble algorithms is higher than 90%.LightGBM algorithm after DTCWT has the highest classification accuracy.The overall classification accuracy and Kappa coefficient are 94.73% and 0.93 respectivesly.Two precision values are higher than without the transform of dual-tree complex wavelet by 1.1% and 0.01. LightGBM algorithm classification accuracy and Kappa coefficient are higher than the XGBoost algorithm by 1.99% and 0.03，and are higher than the GBDT algorithm by 2.9% and 0.04.（2） For sentinel-2A image， LightGBM algorithm after DTCWT has the highest classification accuracy.The overall classification accuracy and Kappa coefficient are 93.25% and 0.91 respectivesly.Two precision values are higher than without the transform of dual-tree complex wavelet by 1.53% and 0.01. LightGBM algorithm classification accuracy and Kappa coefficient are higher than the XGBoost algorithm by 1.14% and 0.02，and are higher than the GBDT algorithm by 2.53% and 0.03.（3） After the transform of dual-tree complex wavelet， the Boosting Ensemble Learning classification can reduce the noise of the image， reducing the salt and pepper phenomenon in the classification results， having stronger regional consistency， improving the classification accuracy.
This research classified urban land cover using the Convolutional Neural Network （CNN） model based on the fine spatial resolution remotely sensed imagery from recently launched JL1 07B satellite. We applied CNN to classify imagery using different combinations of spectral feature variables， and compared the performance of CNN with three other methods， namely maximum likelihood classification algorithm， multi-layer perceptron algorithm and support vector machine algorithm. The experimental results demonstrated that CNN consistently achieved the highest overall accuracy （>90%）， larger than that of other methods by above 12%， and reduced significantly the “salt-and-pepper” noise. The contribution of red-edge band to the classification accuracy was slight， while the near-infrared （NIR） band could increase the OA prominently. Overall， the effect of red-edge and near-infrared bands exerted a slightly impact on the OA of CNN， demonstrating the robustness and generalization of the CNN model. The high accuracy urban land cover classification map achieved using CNN based on JL1 satellite imagery can support the decision makings for land resource allocation， urban planning and regional administration.
Rapid urbanization has led to rapid change in land cover and the landsurface heat balance in megacities. Due to the complex potential nonlinear relationship between land surface temperature and surface biophysical components in megacities， the quantitative models and the response mechanism between land cover and thermal environment in megacities is not yet clear. Takesix Chinese and foreign megacities （Beijing， Shanghai， Guangzhou， London， New York and Tokyo） as the study area， Landsat images were used to comprehensively analyze the quantitative relationship between urban land cover factors and thermal environment. The single-channel algorithm was used to retrieve the land surface temperature of thesix megacities.The random forest regression model was used to establish the quantitative relationship （LCT） model between land cover types and urban thermal environment （LST）. The quantitative relationship between land cover type and LST showed that the LST was closely related to urban land surface types. The spatial pattern of the urban thermal field depends to a great extent on the spatial distribution pattern of the urban land surface types. The impervious surface will lead to the accumulation of high LST fields， while vegetation and water had a significant cooling effect. The land cover compositionin six megacities had different heating/cooling effects. In urban areas， such as Beijing， Shanghai， New York， and Tokyo， the cooling effects of vegetation and water were more pronounced than thosein Guangzhou and London. The established LCTmodel between the three land cover types， NDVI， MNDWI， and NDISI， and the urban thermal environment showed that the LCT model had higher precisionthan that was based on the multiple linear regression method. The R2 value of the LCT_RF model is 0.021~0.074， which is higher than that of the LCT_MLR model. The RMSE is 0.07℃~0.35℃， which is lower than that of the LCT_MLR model.It will be helpful for future construction of eco-cities by studying the interaction mechanism between the land cover and the urban thermal environment in megacities.
There are many researches on vegetation change in arid and semi-arid areas， but little attention has been paid on the social and economic impact of resource-based cities on vegetation change. Based on MOD13Q1 data， climate data such as rainfall and temperature， 11 socio-economic indicators such as raw coal production from 2000 to 2020， combined with GIS technology and statistical methods such as linear regression， the spatial and temporal changes of Ordos vegetation and its influencing factors were studied. The results are as follows： ①The NDVI value of Ordos ranged from 0.233 to 0.395， showing a fluctuating growth trend with a growth rate of 0.059/10 a during 2000 to 2020； the NDVI values of the eight counties under its jurisdiction also showed a fluctuating growth trend， but there were many differences among different regions. ②The vegetation in Ordos is high in the northeast and it is low in the southwest. The area of low vegetation area is 53 500 km2， accounting for 61.58% of the total area of Ordos. The area of high vegetation is only 20 000 km2. The area of the vegetation improvement is much larger than that of the vegetation degradation area. The improvement area accounts for 52.19% of the entire Ordos area， and the vegetation degradation area only accounts for 3.69%. ③The NDVI value is extremely significant positively related to rainfall， with a correlation coefficient of 0.794 （P<0.01）； the correlation coefficient between the change of NDVI and the accumulated rainfall in the month is larger， and the correlation coefficient with the temperature one month ago is larger. ④The NDVI change is extremely significantly positively correlated with the 11 socioeconomic indicators， with a correlation of 0.728~0.796 （P<0.01）. From 2000 to 2020， the restoration effect of Ordos vegetation is good. Rainfall and temperature are the main factors affecting the growth of vegetation in Ordos， of which rainfall dominates. The response of NDVI changes to rainfall has less obvious lag， and the response to temperature has a one-month lag. The positive effects of socio-economic development on vegetation cover outweigh the negative effects.
An extreme rainfall event hit Kerala in India， in August 2018， triggering a large number of geological disasters， causing serious economic losses and casualties. In order to study the impact of land use and its changes during the process of agriculturalization on the development of geological disasters， and to explore a suitable man-land coordinated development model， based on the disaster point data of Idukki， which is the most severely affected area， this paper obtained the land use data of each disaster point in 2010 and 2018 from Google Earth high-resolution remote sensing images and Landsat TM/OLI data extraction normalized difference vegetation index calculation of vegetation coverage to analyze the relationship between geological disasters and land use and its changes. The results show that： ①disasters in the study area are mainly concentrated in the north-central region， where planting forests， planting shrubs， buildings， roads， and other lands with human activities influence accounted for 80.46% of the total disasters； ②although land use change at the disaster site in the study area is few with the overall rate of 37%， however the changed land use are closely related to human activities， such as planted shrubs， planted forest land； ③the vegetation coverage decline rate in the study area was 16.70% ， while disaster susceptibility areas had a better response to areas with reduced vegetation coverage spatially.
Night fog has become an important hidden danger of frequent traffic accidents. Night fog detection is of great significance to prevent and reduce accidents and losses caused by fog， and protect the safety of people's lives and property. There is an obvious boundary between fog and surface of clear sky in the nighttime brightness temperature difference image. Canny edge detection is used to obtain the edge mixed pixel， and the separation detection threshold is automatically obtained by the average brightness temperature difference value of the edge mixed pixel to detect the night land fog. The results of 5-day H8/AHI night fog detection show that the probability of detection is 93.3%， the false alarm ratio is 29.8% and the critical success index is 67.8%. The results show that the algorithm is more suitable for large area dense fog detection， and it is prone to false alarm for special weather such as haze， rainy and snowy days， and fog developing into low cloud. If there is no weather phenomenon associated with fog， the probability of detection is 94.6%， the false alarm ratio is 0.05%， and the critical success index is 90.1%. This algorithm can automatically determine the separation detection threshold of fog and clear sky surface. Compared with the existing automatic detection algorithms of night land fog， this algorithm has higher detection accuracy. The qualitative verification results of night land fog timing detection at different times from 17：00~07：00 on November 27， 2015 to December 1， 2015 show that the algorithm is suitable for the fog detection in the night area of the remote sensing image at dawn and dusk， which can detect about 90% of fog area； for the whole image at night， the algorithm can detect more than 90% of fog area. The results of qualitative verification further prove the stability and reliability of the algorithm.
Atmospheric fine particulate matter PM2.5 is the main atmospheric environmental pollutant that affects human living environment and health. It is of great significance to study the seasonal variation and spatial distribution characteristics of PM2.5 mass concentration for the prevention and treatment of air pollutants. In this study， the MODIS L2 AOD products， MERRA-2 meteorological data and the PM2.5 measured data from ground stations from 2018 to 2020 were used to build the AOD-PM2.5 inversion model based on the improved random forest algorithm. The PM2.5 in Beijing-Tianjin-Hebei region was estimated， and the spatial distribution characteristics and seasonal variation of PM2.5 mass concentration were analyzed. The results showed that： （1） The mean values of determination coefficients （R2） of spring， summer， autumn and winter model were 0.78， 0.66， 0.83 and 0.83， respectively. And the accuracy of simulation is higher.（2） The PM2.5 concentrations of spring， summer， autumn and winter in Beijing-Tianjin-Hebei region from 2018 to 2020 showed significant spatial distribution characteristics and seasonal variation. The maximum of PM2.5 concentrations occurred in winter and the minimum value appeared in summer. （3） Compared with the same season over the years， the PM2.5 pollution range and PM2.5 concentration in the Beijing-Tianjin-Hebei region have improved. Compared with 2018 and 2019， the PM2.5 pollution range in spring and autumn of 2020 improved significantly.
Methane （CH4） is a chemically and radiologically active gas in the atmosphere. With the development of human civilization， the total amount of CH4 has been increasing. Before the industrial revolution， the global CH4 concentration was 700 ppbv， and it reached 1714 ppbv by the 1990s. This paper describes the development of CH4 ground-based detection， space-based detection， satellite-based detection and inversion algorithms. During 1979~1983， the SAMS on Nimbus-7 satellite observated CH4 concentration in the stratosphere for the first time. After that， many sensors have been launched to observe CH4. The satellite-based instruments mainly observe CH4 in the infrared band by nadir viewing， limb sounding or occulation observating. In recent years， the spatial resolution and spectral resolution of the sensors are greatly improved. The TROPOMI has a spatial resolution of 7 km×7 km； the spectral resolution of ACE-FTS、AIUS reach 0.02 cm-1. In the near infrared band， WFM-DOAS is the main algorithm for inversion of CH4 concentration， and OEM is mainly used in the mid and far-infrared band. With the development of satellite observation technology and the improvement of retrieval algorithms， the accuracy of CH4 retrieved has gradually improved. The deviation between CH4 concentration retrieved by TROPOMI in the nadir viewing and the CH4 concentration provided by ground-based stations is 14 ppbv （0.8%）； The accuracy of CH4 profile retrieved by ACE-FTS in the limb sounding/occultation observing is within 10% from troposphere to lower stratosphere.Then， the trends and distribution characteristics of global CH4 concentration at 300 hPa and 150 hPa are described in this paper. The global CH4 concentration increased by ~50 ppbv from 2010 to 2020， with an average annual growth rate of ~0.29%； At 300 hPa， the global CH4 concentration was 1 767 ppbv in 2010 and increased by 55 ppbv in 2020； At 150 hPa， the global CH4 concentration was less than 1 700 ppbv in 2010 and reached 1 745.6 ppbv in 2020； The global CH4 concentration is higher in the north and lower in the south， the reason lies in that there are numerous methane emission sources （freshwater wetlands， rice cultivation， fossil fuel combustion and biomass combustion， etc） in the northern hemisphere； The high concentrations of CH4 are distributed in the mid-Atlantic， northern Africa， the Middle East and western China.
With the intensive study of global climate change， the monitoring of atmospheric aerosols have been gradually attracting people's attention. Polarization remote sensing technology has the advantage of observing aerosol microphysical characteristics on a large scale， which serves as an effective means of monitoring and managing aerosol pollution in large areas. Nowadays， this technique can obtain more dimensional aerosol information， which can solve the problem that traditional ground stations cannot monitor on a large scale. In this paper， the retrieval of fine-mode aerosol optical （AODf） is performed， based on the multi-angle， multi-band， polarization and intensity measurements from Directional Polarimetric Camera （DPC） onboard GF-5 satellite. The ground surface bidirectional polarization distribution function （BPDF） model was used to estimate the polarized ground surface reflectance， and the AODf is retrieved with the Look-Up Table （LUT） method. The comparison between the retrieval results and the Aerosol Robotic Network （AERONET） ground-based observation data produced a correlation coefficient （r） of 0.903， 90% of the errors are concentrated between 0.0~0.3， the Mean Absolute Error（MAE）， the Mean Relative Error （MRE） and the Root Mean Square Error（RMSE） are 0.026， 0.43% and 0.060， respectively， which proves the retrieval results are generally reliable and the retrieval method is feasible.
Time-series InSAR technology provides an effective method for monitoring and controlling land subsidence in Kunming city. However， the defects of PS-InSAR and SBAS-InSAR technology limit the monitoring accuracy， especially the low coherence of PS points caused by complex terrain. In this paper， we proposed InSAR technique combining PS， SBAS and DS to monitor the subsidence in Kunming urban area， and the results of the proposed method are compared with PS+SBAS-InSAR. The results show that the proposed method is in good agreement with the results of the Kunming subsidence rate inversion by PS+SBAS-InSAR， and the proposed method can enhance the spatial distribution density of the points in the observation area and obtain more effective surface deformation information. From the perspective of the whole study area， the subsidence rate of the urban surface of Kunming city is -22~8 mm/a， and the serious subsidence areas are concentrated in Guandu District， Xishan District and Wuhua District， and several subsidence funnels have been formed. Since 1989， Xiaobanqiao and Hewei Village are still the two most serious subsidence funnel centers， while Jiangjiaying in the northeast is the new subsidence point found in this study. Combined with the analysis of historical data， it is shown that the ground subsidence in Kunming is mainly affected by groundwater pumping， building load， engineering construction and tectonic movement of faulted basin.
MODIS dailyLand Surface Temperature （LST） products are seriously contaminated by weather effects and the effective pixel information missed. It is sincerely important in areas where data is sparse. An approach to downscaling LSTs from AMSR-2 vertical polarizations multi-brightness temperature and vegetation index observations was preliminarily investigated in the Gurbantunggut Desert， and then the downscaled LSTs were used to fill the gaps due to clouds in the MODIS of 2018.（1） In this study， four machine learning methods（Cubist、DBN、SVM、RF）， two training spatial resolution（5 km、10 km ）， ten band combinations， were applied to train the model. The 10-fold cross-validation results show that the RF model and C09 band combination have the best simulation effect. （2） Two robust downscaling methods of land surface temperature using Random Forest algorithm （5 km|RF|09/10 km|RF|09） were developed to retrieve a 1km-resolution land surface temperature product from Advanced Microwave Scanning Radiometer 2 （AMSR2） data. The validation results with MODIS and station LSTs show that 5 km|RF|09 downscaled LSTs has a better performance than 10 km| RF|09. Comparisons of the retrieval results with MODIS LSTs and ground measurement data from Fukang stations yielded that R2 respectively is 0.971、0.761， RMSE is 3.380 K、7.614 K and MAE is 2.509 K、6.695 K， which indicated that the accuracy of the 5 km|RF|09 LST retrieval model was high. （3） The downscaling results fill the gaps due to clouds in the MODIS， which can be applied to long-term LST sequence analysis in Gurbantunggut Desert. The method of LSTs downscaled provided scientific reference for data acquisition in data sparse area.
WRF is an important tool for urban near-surface wind field simulation and land use and land cover data is an important input of WRF. It is one of the important factors which affects the accuracy of WRF simulation. USGS and MODIS data are defaults land use and land cover data of WRF. They both have poor timeliness and low spatial resolution which often cause poor accuracy of simulating near-surface wind field in rapidly developing urban areas. Extraction of land use and land cover data involves many land features. It will be time-consuming and labor-intensive， if all land features are updated. It can’t meet the needs of pratical application. Among all land features， urban built-up area changes fastest and it also has the greatest impact on regional climate. In this paper， a method of extracting built-up area from SAR data and Night Light （NL） data is proposed to update land use and land cover data rapidly. Then we use updated WRF to simulate wind field. The research in Beijing-Tianjin-Hebei shows that this method can extract the boundaries of built-up areas quickly and accurately. And the accuracy of WRF simulation of near-surface wind speed improves significantly.
Leaf Area Index （Leaf Area Index， LAI） is one of the important indicators to characterize the changes of land surface characteristics， as well as an important parameter of land surface and hydrological models.This dataset was based on GLASS LAI （8 d/500 m）， combined with MOD13A1， MYD13A1 and Landsat 7-ETM+，Landsat 8-OLI data. First， ESTARFM model was used to synthesize LAI at 8 d/30 m resolution， and then LAI with high spatial and temporal resolution （1 d/30 m） was obtained by time linear interpolation. The spatial and temporal characteristics of LAI with high spatial and temporal resolution （1 d/30 m） were compared based on GLASS LAI products to verify the accuracy of the data set. The results show that the distribution features of this data set are basically consistent with that of GLASS LAI in space， and the contour and texture are clearer. In terms of time， they have the same intermonthly variation characteristics， and the regional monthly average LAI and regional 8-day average LAI estimated by 1 d/30 m LAI have a significant positive correlation with the original GLASS LAI， R2 are 0.95 and 0.94， respectively. Pearson product moment correlation coefficients are 0.97， P values are all less than 0.01.
The gridded precipitation is an important support for the studies of meteorology， hydrology and ecology on the Tibetan Plateau. However， due to the limitations of topography and observation conditions in this region， conventional gridded precipitation products are usually difficult to reflect the actual statistical parameters of precipitation within the grid-box. Also the measurement loss casing by wind are not took into account. To solve this problem， this paper introduces a new gridded precipitation with measurement undercatch correction and frequency distribution optimization. The dataset has a temporal resolution of 1 day and a horizontal spatial resolution of 10 km， and ranges from 1 January 1980 to 31 December 2009. The dataset can be used as a reference data source for numerical model precipitation correction and as an input parameter for various ecological hydrological models. Compared with the similar gridded precipitation datasets， this dataset considers the wind-induced measurement undercatch an is more accurate in the precipitation frequency distribution.
To fully fuse the feature information in the spatial and frequency domains of hyperspectral image （HIS）， a spatial-spectrum fusion HSI ground object recognition model that integrates multiscale features of Gabor and LPQ （Ms_GLPQ） is proposed. Firstly， the Gabor filter bank is used in the spatial domain to extract the multiscale and multidirectional spatial neighborhood information of various ground objects in HSI to describe the spatial structure of its edge and texture. Secondly， the Local Phase Quantization （LPQ） operator is utilized in the frequency domain to extract the multiscale frequency domain texture features， and the phase invariant feature description of HSI is obtained. Then the Principal Component Analysis （PCA） algorithm is used to reduce the dimensionality for the problem of feature redundancy， and the features in the spatial and frequency domains are fused to obtain the feature vector that fully describes the HSI information. Finally， the classifier based on Boosting tree （XGBoost， CatBoost， etc.） are utilized for recognition. Experiments on Indian Pines， Salinas， and tea farm datasets acquire accuracy rates of 85.88%， 94.42%， and 92.61%， respectively. The experimental results show that the Ms_GLPQ model can extract effective features in HSI and obtain more discriminative multi-featured region descriptors than traditional methods， and it performs better by using boosted tree model for ground object recognition and achieves higher accuracy than other classifiers.
A statistical analysis of the HY-2B satellite radar altimeter near-shore waveform in Hong Kong is carried out. The coastal non-ocean waveforms were divided into three categories： ocean-like waveform， land-like waveform and single cone waveform. On this basis， this paper proposes a coastal waveform processing approach based on neural network classification. Seven waveform features were extracted as the inputs for classification， and then a BP （Back propagation） neural network classifier was constructed. Different types of waveforms are handled differently. For ocean-like waveforms， an innovative method of removing abnormal samples before retracking was proposed， which effectively improved the waveform retracking success rate. Experiments with Jason-2 satellite radar altimeter near-shore data show that the success rate of retracking and the validity of the data have been improved. Experiments are performed using the sea area data of Zhiwan Island from the HY-2B satellite radar altimeter， and the tide gauge station data is used as a reference standard. The sea level height obtained by retracking is compared with the sea level height of the tide gauge station. The standard deviation of the difference sequence obtained by the method in this paper is 6.5 cm， which is better than the 14.5 cm of the traditional processing method.
An asymmetric fingerprinting scheme for remote sensing images based on the homomorphic public key encryption algorithm is proposed for solving the problems of high computational complexity and low bandwidth efficiency in asymmetric digital fingerprinting. In this scheme， the data provider encrypts the remote sensing images via the DCT spread spectrum scrambling method， and the Bresson homomorphic public key encryption algorithm is used to realize asymmetric distribution of the decryption key. The copies of remote sensing images with the fingerprint are decrypted by the client， and different copies with the fingerprint come from different decryption keys. The remote sensing images are not directly encrypted by the public key encryption algorithm， so the complexity of the algorithm is greatly reduced and the encryption efficiency is dramatically improved. Meanwhile， the demand of bandwidth will be reduced and the efficiency of bandwidth will be improved， because data owners not only need to generate one copy with the fingerprint for different users， but also data copies are distributed to different users via multicast transmission. The experiments have shown that the scheme can effectively improve the efficiency of bandwidth and the efficiency of encryption when the number of users is relatively large， which can greatly reduce the computing load of the data server and reduce the waiting time of the users.
