Downscaling plays an important role in comparison,integration of disparate datasets and calibration,validation of models in a range of applications.Firstly,the general formulation of geostatistical-based downscaling through the use of area\|to\|area,area\|to\|point and point-to-point covariances in the kriging system will be briefly descrebed.In the next section,a basic idea of conducting point support variogram from geographical units with different sizes and shapes is given.After that the principles and cases about geostatistical-based downscaling methods were developed in recent years that would be introduced.Area\|to\|area and area\|to\|point kriging could be used without auxiliary data and efforts to find a method which provides better prediction for supplementary information is tried.Finally,trends and some basic questions of methods proposed recently are discussed in the conclusion.
Results of a new methodology for the retrieval of surface particulate matter concentration (PM10) from MODIS (MODerate resolution Imaging Spectroradiometer) top\|of\|atmosphere reflectance measurements in Beijing is presented in this paper.Basis of the PM10 retrieval methodology is ngstrm-α derived from multi\|spectral aerosol optical depth (AOD).Based on properties of 6 aerosol models from OPAC (Optical Properties of Aerosols and Clouds) database,AOD is retrieved from MODIS top\|of\|atmosphere reflectance at 500 × 500 m2 resolution.Particle effective radius is related to ngstrm\|α and can be estimated by ngstrm\|α.By integrating satellite derived AOD,effective radius and ECMWF (European Centre for Medium\|Range Weather Forecasts) forecast meteorological data,therefore PM10 can be retrieved.AOD was validated by ground based AERONET measurements (Beijing and Xianghe) and PM10 was validated by surface PM10 which was estimated from AQI.The results showed that retrieved AOD (r=0.923,RMSE=0.149) and PM10 (r=0.794,RMSE=48.34 (μg/m3) )have a good consistency with ground observations.
In order to dig up information from multi\|spectral remote sensing data and standardized and quantitative interpretation knowledge for geological hazards,the characteristic charts of typical earthquake secondary geological disasters on remote sensing images are studied.Landslides,debris flows and collapses induced by ““5·12”” Wenchuan earthquake were selected to build characteristic charts.Firstly,disaster units were extracted from remote sensing images and twenty characteristic parameters of spectral,texture,color for the images were calculated.Then,the parameters were normalized and the first three principal components were obtained by principal component analysis which were used to represent the features and build characteristic charts for various types of earthquake secondary geological disasters on remote sensing images.It verified that the characteristic charts could express the image features for different types of earthquake secondary geological disasters and distinguish them from through being tested with new samples of geological disasters images.The characteristic charts revealed the relationship between remote sensing image features of different seismic secondary geological disasters,and provided quantitative and visual knowledge for geological hazard identification by remote sensing images.
With the influence of the monsoon and climate,high frequent droughts have heavy effect on the grain production in Huang\|Huai\|Hai Plain,which is one of the most important grain base in China.So it is significant to real\|timely monitor drought conditions in this plain for the improvement of our grain yields.Based on MODIS reflectance,land cover products,and rainfall over the weather stations in the study area,the Modified Normalized Difference Water Index(MNDWI),Vegetation Health Index (VHI) and standardized precipitation index (SPI) were adopted to monitor drought conditions from 2001 to 2012.This paper analyzed the temporal and spatial variations and their underlying reasons.The results indicated that VHI had a coincident variation with MNDWI,Winter monsoon from Siberian,spring vegetation and fast water evaporation influenced significantly spring drought over the Huang\|Huai\|Hai plain,especially for Yanshan and Taihang piedmont.Crop were planted from 25 to 100m height above sea level which is more sensitive to dought than others;the weakest drought was happened in 2003.MNDWI and VHI are sensitive to vegetation moisture and land surface temperature,and can be used to monitor drought in real time.While SPI is suitable for drought monitoring or forecasting over a long time series.
The study area is located in Lima South Central,Peru,and the south of Andean porphyry belt northern,with good copper polymetallic mineralization geological conditions.Based on ASTER multispectral satellite data and the spectral characteristics of altered minerals region,using characteristic band combination of principal component analysis method completed the altered minerals assemblage information extraction.Using this method,extracted Al\|OH altered minerals information,such as sericite,kaolinite,montmorillonite,illite,alunite ,et al,and propylitization alteration minerals information,such as chlorite,epidote and carbonate (calcite and dolomite).While using interactive interpretation technique in the study area to carry out geological interpretation of remote sensing,which combined with regional geological characteristics,had a comprehensive analysis of remote sensing indicator mineral information including ore\|controlling linear structures,circular structures,ore\|bearing strata and alteration mineral assemblages.In addition,remote sensing indicator mineral information conducted a comprehensive prospecting prediction,and delineated favorable prospecting area.Comparing with the ground stream sediment geochemical mapping and high\|resolution satellite images,which indicates that the delineation of remote sensing prospecting area provides important reference basis for exploration in the development of ground mineral.
The objective of this study was to estimate the area planted with paddy in Jiangning district of Nanjing City,Jiangsu Province.In order to study the paddy identification in rainy and cloudy region where cropping rice was cultivated with the machine or sowed and distributed brokenly,this paper extracted the rice\|planted fields by using the method of hierarchical classification based on different data.The data involved multi\|temporal SAR data,the fusion data with TM and multi\|temporal SAR,and the fusion data with TM and single temporal SAR.Then we compared the results,including the accuracy and the area,and analyzed the reasons according to field investigation.The result showed that the accuracy of paddy identification of multi\|temporal SAR or the fusion with TM and multi\|temporal SAR were not lower 72%,and higher than the fusion with TM and single temporal SAR;the both paddy\|planted area were close.The main factors were that feature distribution,paddy with the different planting patterns and broken distribution.
The Three\|River Headwater Region is one of the most important ecological function areas in China.In recent years,due to the global warming and effect of frequent human activities,the degradion of grassland ecosystem becomes serious in the Three\|river Headwater Region.Based on HJ\|1A HSI data,field spectrum data and sample investigation data of different degradation level grasslands,using MLC and SAM methods,this paper studied the classification of different degradation level grasslands in the northeast area of Qingshuihe Town of Chengduo in Three\|River Headwater Region.The results showed that the overall classifiction accuracy of SAM method was above 75% based hyperspectral image data and proved the feasibility of the classification approach.It performed effectively discriminating different degradation level grassland with similar vegetation coverage based on hyperspectral image data.SAM performed better than MLC in classification accuracy and fine degree.SAM worked best in distinguishing medium level degradation grassland,while performed weaker in distinguishing other degradation level grassland,which could be improved by introducing auxiliary parameters.
Dark Dense Vegetation(DDV) method is most widely used for retrieving the atmospheric aerosol optical thickness from remote sensing images in recent years,in which automatically detecting the dark dense vegetation pixels is the primary basis.However,it is quite difficult to detect DDV pixels for the visible and near\|infrared (VIS\|NIR) images without the mid\|infrared band.A new method was named red band histogram threshold method that was proposed in this paper to automatically detect the DDV pixels from VIS\|NIR images at typical mountainous areas.Using the low reflectance of forest pixels in visible bands,the method searched for the first peak of red band reflectance histogram,and detected the DDV pixels by threshold value defined by the first peak.Results of the proposed method were presented for omlyLandsat\|5 Thematic Mapper (TM) imagery using only the first four bands.The performance of the method was compared with mid\|infrared threshold method and multiple thresholds method,which were widely used in the imagery with mid\|infrared band and VIS\|NIR imagery respectively.Then the proposed method in this paper was applied to HJ satellite CCD imagery to detect DDV pixels and retrieve aerosol optical thickness.Results show that the proposed method evidently is better than the multiple threshilds method,and its accuracy is very close to the Mid\|infrared threshold method.The proportion of DDV pixels whose similarity index is less than 2 and less than 3 were 83.12% and 93.48% respectively.It is a new robust method applicable for automatic detection of DDV pixels for VIS\|NIR images at typical mountainous areas,and meets the need for the DDV method to retrieve aerosol optical thickness.
There are two shortcomings in Compact Polarimetric(CP) SAR H/α mpirical characteristic space:(1) the problem that scattering entropy in compact modes is generally higher than that in fully polarimetric (FP) mode,which is not taken into account;(2) the classification efficiency of the existing compact H/a space is relatively weak when the scattering mechanisms overlap,especially for multiple scattering mechanisms.In this paper,the relationship between scattering angle in DCP and FP modes is quantitatively analyzed;parameter ED is proposed as an alternative to entropy H,by analyzing data acquired seven different platforms,and the steady condition of H/α decomposition (including entropy H,scattering angle α and alternative parameter ED) is analyzed using Monte Carlo simulation;a new ED/α characteristic space for compact polarimetric SAR is then proposed through the statistics of density map in ED/α pace of each scattering mechanism.Experimental results show that ED in compact polarimetric mode and H in fully polarimetric mode have a strong correlation,and ED/α haracteristic space improves the classification accuracy of the scattering mechanisms.
In order to promote the application of high spatial resolution remote sensing image in traffic industry and enhance the levels of planning,investigation and management in transportation network for city\|clusters,the applications of HSRRS to monitor the transportation network condition is proposed paper in this.On the basis of the applications of the partition man\|machine interactive SNAKES algorithm and the edge detection combined with Support Vector Machine algorithm respectively in the information extractions of the road and the vehicle in high resolution remote sensing image,application technologies and methods in traffic industry are elaborated,such as the transportation network and the vehicle vector information were used to judge city\|clusters traffic queue state and the demand analysis of new residential bus station,and a foundation is established for future in\|depth application.
The hyperspectral data has very high spectral resolution,but low spatial resolution with the trade\|off between the spectral and spatial resolution,as well as the physical limits.Therefore,in many cases,the spatial resolution of the hyperspectral image system is lower than the multispectral image system that has the less spectral channels.The fusion of the hyperspectral and multispectral data can produce fused data with high spatial and spectral resolution which can contribute to the identification and classification of the material at a finer spatial resolution.NMF (Nonnegative Matrix Factorization) algorithm is used to perform the fusion of low spatial resolution hyperspectral data and high spatial resolution multispectral data.Firstly,VCA (Vertex Component Analysis) is used to perform the factorization of hyperspectral data to obtain endmember spectra matrix and abundance matrix; Secondly,NMF(Nonnegative Matrix Factorization) is used to unmix the hyperspectral and multispectral data alternatively to obtain high spectral resolution endmember spectra matrix W and high spatial resolution abundance matrix H.Finally,fusion data with high spectral resolution and high spatial resolution can be obtained by multiplying the two matrices,.Sensor observation models of the data are built in the initialization matrix of each NMF unmixing procedure.The experiments with AVIRIS data and HJ-1A data have shown that NMF method can be used to improve the spatial resolution on all wavelength regions,and the higher qualities of the estimated data by NMF can be used for the classification of the materials and identification at a finer spatial resolution.
A dense image matching method based on iterative dynamic programming is proposed to solve the mismatching points diffusion effect caused by the overall integrity in the dynamic programming dense image matching.The parallax direction consistency and parallax mutation are introduced as an iterative decision criterion in the algorithm.By analyzing a set of mismatching point candidates which are against iterative condition in dynamic programming matching result,the mismatching points are identified and removed from the set.The iteration continues until the iterative decision criterion is fulfilled and the mismatching points diffusion effect will be solved.The algorithm proposed in the paper has been successfully applied to Chang’e\|3 project.The results in orbit show that the dense imagery matching algorithm based on iterative dynamic programming is able to greatly reduce the mismatch rate of corresponding image points in stereo image match.
Remote sensing technology has become the main means to achieve the extraction of ground information.Based on high\|resolution remote sensing image,this paper adopted multi\|scale segmentation algorithm and measured the characters of object spectrum and shape to realize the object\|oriented classification method.The classification method was a combination of object\|oriented method and comprehensively analyzes the features of objects,give full paly to the advantage of the fine features classification of high\|resolution remote sensing image,so that high precision classification results were obtained.The method was tested in Nabanhe River Watershed Nature Reserve,Xishuangbanna,then the method was checked by the total accuracy of classification result reached to 88.58% and Kappa coefficient was 0.77.The method was a high\|precision and rapid classification method.
The topographic correction of remote sensing images is an important factor of improving the precision of quantitative remote sensing data processing.Some topographic and atmospheric correction algorithms have been integrated in the current data\|processing software systems.However,there are still some problems for their operational applications.We are unable to obtain the important terrain parameters ( i.e.shadow factor,sky view factor),so need to provide an accurate DEM.The correction algorithms are also based on the assumption that the land surface is Lambertian in the applications.To meet the requirement of professional users of Remote sensing,this paper designed and realized a topographic and atmospheric correction system for remote sensing images.The system aims to provide radiation correction,obtaining DEM and its correlative terrain parameters,topographic and atmospheric correction for remote sensing image.This paper presented the design of function modules and a prototype version of the system.Based on the topographic and atmospheric correction methods showed in this paper,the reflectance of HJ/CCD and LANDSAT/TM images were obtained.The results showed that the BRDF method of the system can eliminate the effect of topography.This system can support the research and application of remote sensing science.
Inscribed by UNESCO as world cultural heritage site,Fujian Hakka earth building is a wonderful work of ancient Chinese architecture with high architectural,cultural and tourism value.It is of critical importance with highly precise,realistic three\|dimensional (3D) models of cultural heritage to fulfill the goals for their protection,preservation,communication and inheritance,and to promote virtual culture tourism.Hakka earth buildings are characterized as large size,with complex geometry and irregular shape,rich detail features,as well as fragile,vulnerable to manmade and natural damages.It is difficult to obtain complete and accurate geometric information by traditional survey and measurement methods.And it may induce harm to the cultural heritage with frequent contact in the measurement process.Terrestrial laser scanning and other technologies were used to create realistic,fine 3D models of Hakka earth building.Firstly,this study proposed a complete procedure for 3D high\|precision reconstruction of Hakka earth buildings based on point cloud;Secondly,some core steps in point cloud data processing and three\|dimensional reconstruction,in particular point cloud data registration,simplification and de\|noising,components contour extraction,three\|dimensional geometric modeling and texture mapping are discussed in detail.In addition,we discussed the difficulties and specific solutions to texture optimization in the process of modeling analysis.Application results show that 3D laser scanning technology is well suited for precise,realistic modeling of Hakka earth buildings with high complexity geometric features,as provides high\|precision data base for future development and protection,building damage monitoring,restoration and virtual display,etc.At present,the technical approach has been applied to the Fujian Hakka earth building digital tourist information services with excellent results.
Urban is one of the important types of land use/cover mapping.The spectral character of barren lands and urban information in arid and semi\|arid region are similarity in the moderate resolution remote sensing images.It is very difficult to extract the urban information using the traditional classification methods based on pixels.For exactly extracting the urban information of the Middle reaches of the Heihe River basin located at the typical arid and semi\|arid region in northwestern China using Landsat TM image,a multi\|layers classification procedure combining object\|oriented classification method for precisely extracting urban information was is proposed in this paper.As one critical step of object\|oriented classification,image segmentation must ensure to get the best segmentation result.In order to choose the optimal segmentation scale,divergence is applied to measure the segmentation result.The urban information extracted using the new method rejects the influence of barren lands for extracting urban information,which is better than traditional supervised classification using SVM classifier and time series classification method.Comparing the urban extraction results using three different methods,the new method works better during more confusing regions and the boundary information is more precise than the other two results.The accuracy estimation result shows that the overall accuracy is about 94.51% and the kappa coefficient is about 0.89 using the new method.
Mixed pixels are widely presented in remotely sensed images.Soft classification techniques can avoid the loss of information comparing to hard classification methods while handling mixed pixels.However,the assignment to these classes by soft classification does not specify the location in the pixel.Sub-pixel mapping (or super-resolution mapping) is a technique which designed to use the information obtained by soft classification to get a sharpened image and it can incorporate benefits of both hard and soft classification techniques.In this paper,a variation of genetic algorithm,named as partheno-genetic algorithm (PGA),is developed to accomplish the sub\|pixel mapping.With the sub-pixel/pixel attraction model,PGA can achieve sub-pixel mapping in a straightforward one-pass process.It is evaluated with artificial and degraded land cover images by visual and quantitative classification accuracy indices.The results show this method can increase accuracy while compared to hard classification.
An improved classification algorithm is proposed to use multinomial logistic regression with DFP quasi\|newton to solve the problem of slowly improving classification accuracy and time consuming when used multinomial logistic regression classifier.The second order Hessian matrix is replaced by secant method in newton method and using DFP to improve the performance.After two different hyperspectral image experiments by using presented method,the result shows that this method has a big improvement in both accuracy and time consuming than the traditional multinomial logistic regression method.
Human or animal activity information on snow relates to applications in various fields animal tracking such as,in winter,animal migration tracking,public safety and defense,etc.Remote sensing technologies can play key roles in monitoring the tracks of human or animal activity traces through image enhancement,e.g.filter techniques with very high spatial resolution images,especially linear disturbance traces from human or animal activities in snow\|covered area,and grasp the animal migration,human activities on the frozen river and other information timely to further assist in decision\|making.In this paper the principle and characteristics of the Difference of Gaussian filter(DoG) were proposed to identify linear disturbance traces in the snow,and experiments showed that the accuracy of traces extraction reached 83.67% with an adaptive filter(σ=1.5).The results were more reliable and effective than that from laplacian,sobel and prewitt operators.DoG filter is more suitable for human activity traces identification on the frozen river,and will play a key role in further services for recognizing human or animal activity type,trace route,as well as security patrols of relevant departments.
The paper introduced the morphological building/shadow indexes derived from high spatial resolution image,incorporated with object\|based information extraction routine,to extract the city buildings and its shadow to accurately and further estimate the height of buildings.Firstly,morphological building index with the multi\|scale and multiple directional features were used to separate buildings from the neighboring roads with the similar spectra.Secondly,a dual\|threshold strategy was applied to further improve the extracted accuracy of buildings.In the end,the height of buildings was estimated by the model,where the relationship among the length of buildings shadow,the azimuth and elevation of sun,and satellite is constructed.A case study with ZY-3 dataset,located in Xiamen Software Park,was used to validate the algorithm.Experimental results demonstrate that the proposed method can significantly improve the accuracy of buildings extraction,when compared with the supervised classification method with Support Vector Machine (SVM).Moreover,the mean square errors of buildings height eatimation is about ±1 m.
Microwave Intensity Correlation Imaging (MICI) breaks the limitation of antenna aperture to reconstruct image by random radiation field and echo signals.As a new type of statical imaging technique,the influence of model parameters was not clear to the image radiant performance.The MICI theory was demonstrated in the paper,and then model parameters were changed to analyze the radiant performance of measurement matrix and reconstructed image.The results showed that the image radiant performance can be improved by increasing the number of antennas and antennas’ space and bandwidth,especially for large and high resolution area.However,it drops if antennas are in a higher platform or girds are larger.The lower quality system is acceptable to well\|distributed area.This study is conducive to improve the MICI system.
The simplified UV Radiation Model was applied to OMI satellite sensor data to map the UVB irradiance in the Western Pacific Ocean,and the characteristics of the UVB radiation distribution in the Western Pacific Ocean were investigated.Firstly,the parameters derived from OMI data,such as total ozone amount,optical thickness of aerosols,reflectance,and etc.,were applied to the simplified UV Radiation Model to calculate the UVB radiation.Secondly,the comparison of the UVB radiation between the satellite inversion and ground measurement was carried out to evaluate the performance of the simplified UV Radiation Model.The results show that UVB radiation derived from satellite data is highlyconsistent with the in situ measurements,this means that the simplified UV Radiation Modelhas a good performance in deriving UVB radiationin the western Pacific Ocean.Finally,the UVB radiation in the western Pacific Ocean derived from OMI satellite sensor was mapped from 2007~2010,the distribution of the UVB radiation with temporal and spatial scales was analyzed,the result is consistent with the existed observations which have previously been done.
Using the RTTOV10 forward model and atmospheric parameter profile datasets,a numerical simulation scheme was developed based on artificial neural networks,which provides a reference for the selection of channels,requirements analysis,development of retrieval algorithm and evaluation of the retrieval products for atmospheric soundings from satellite\|borne,airborne or ground\|based microwave radiometers.Then,the impact of retrieval method,channel selection and observation error of bright temperatures were analyzed by a series of retrieval experiments.The results show:(1) The neural networks provide better reproductions of the profiles than the statistical inversion,and especially show less sensitivity to the bright temperature observation errors;(2) The humidity detection channels near 183.31 GHz can provide some information for temperature retrievals,which is about 0.2 K between 400~800 hPa.The channels near 118.75 GHz can provide additional information for temperature soundings in the whole atmosphere,and the largest positive impact is about 0.4 K occurring near 200 hPa;(3) The temperature sounding channels near 50 GHz and 118.75 GHz are not only important but complementary for humidity profile retrievals with 183.31 GHz channels;(4) The retrievals of temperature and humidity profiles are sensitive to the brightness temperature observation errors and the surface emissivity assumptions.
Using three kinds of winter (November to February)snow data between 2010~2012 in Altay region study snow depth retrieval.Study data include retrieval of snow depth by FY-3/MWRI data,snow cover data from U.S.NATIONAL ICE CENTER Interactive Snow and Ice Mapping System ( IMS),and measured snow depth in and around Altay region.Through combining advantages of these three types data built a correction model of snow depth retrieval,eventually got a better result of snow depth retrieval in the study region.Meanwhile achieved the corresponding model platform by programming,for operational monitoring snow depth of the study area in the future.The result indicates that the model is able to improve the accuracy and reliability of snow depth retrieval by FY\|3/MWRI data,eliminating the drawback that retrieval of snow depth by FY\|3/MWRI in Altay region is lower than that measured,narrowing the gap between microwave and measured average snow depth from 21.7~12.1 cm before modification,to 3.7~1.5 cm after modification.
In recent years,hyperspectral remote sensing technology has been widely used in the inversion of soil organic matter contents.In this paper,in order to inverse soil organic matter contents,indoor spectral of 146 soil sampling data,which is collected from Yushu and Maduo County in Three\|River Headwaters Regions,was measured by ASD FieldSpec4;Thereafter,soil organic hyperspectral forecast models for soil matter contents in four kinds of transformation forms were respectively established using Partial Least Squares Regression (PLSR) and Artificial Neural Network (ANN) method.
The results showed that the overall accuracy of using ANN model to estimate soil organic matter content is higher than that of using PLSR model,and its total root mean square errors were all less than 17.51,however,The different soil types have showed different optimal estimation models and the optimal indicators.The optimal retrieval models and indicators of Alpine meadow and marsh soil are both the ANN model and BD (Band Depth),and their total root mean square errors are 10.29 and 3.29 respectively;PLSR model is the optimal estimation model of Alpine steppe soil,the optimal indicator is REF,with the model’s total root mean square error of 5.59;PLSR model is the best inversion model for Mountain meadow soil and the optimal index is BD index,with the model’s total root mean square error of 4.68.The results further indicated that the ANN model and the PLSR model can better predict SOM contents of four soil types inThree\|River\|Headwater Region,and
