The nonparametric model of estimated the sea state bias with the local linear regression estimation,the spherical Epanechnikov kernel function and the local adjustable bandwidth is established based on HY-2 altimeter data.The performances of this model are evaluated by the explained variances,the correlations between the sea state bias and the significant wave height or wind speed and residuals.Compared with the estimations of parametric model under the same dataset,the results show that the correlations between the sea state bias of the nonparametric model in this paper and the significant wave height or wind speed are higher which means the model is more effective.The nonparametric model and the parametric model both have their own merits in different latitudes.The nonparametric model performs better in high-latitudes of the northern hemisphere.
For studying the effect of sea surface wind on sea surface salinity retrieval,the reasonable forward model of brightness temperature simulation should be established.The calculated results of different main sea roughness models used in sea salinity retrieval at L\|band are compared with the roughness data of Aquarius product.The result shows that the two scale & foam model agrees best with the Aquarius data.Forward model of the brightness temperature simulation of wind\|induced sea is built based on the flat sea model and the two scale & foam model.The effects of sea surface wind on seawater permittivity and the effects of sea surface wind error on salinity retrieval are investigated.The results show that 2 m/s wind velocity error has a great effect on the salinity retrieval and fails to reach the required precision when the sea surface temperature is low or the wind velocity is high.20°wind direction error has a weak effect on the salinity retrieval and reaches the monthly average precision requirement when the incident angle is not large.
To enhance the ability of weather radar to estimate precipitation quantitatively,the coefficients a and b in the relationship k=aZ b between the specific attenuation k and the reflectivity factor Z of an X-band weather radar were retrieved using the revised least residual error method.In the method,the path integrated attenuation was obtained by a microwave link which was set up along the radial of the X-band weather radar.At the same time,the attenuation correction of the reflectivity factors along the corresponding radial was realized.Experiment on real data shows that the mean of the optimal a is 5.2×10-5with a standard deviation of 2.9×10-5 and the mean of the optimal b is 1.068 with a standard deviation of 0.118.Comparing with the results in the literatures,the a is relatively smaller and b is larger.Assisted by the microwave link,the radar reflectivity factors can be corrected for attenuation effectively and the instability problem is addressed.After correction,the radar reflectivity factors are more reliable.The estimated rainrate using the corrected reflectivities are more closer to the raingauge measurements than using the uncorrected reflectivities.The results of this experiment can be applied in attenuation correction and quantitative precipitation estimation.
The traditional parametric models of estimating the sea state bias may result errors,so a segmented modeling method was presented to improve the estimations of the sea state bias.Based on the latitudinal distributions of the significant wave height and wind speed,the global area was divided into 6 different latitude ranges at the segmented interval of 20°.According to Taylor expansion,the parametric models as the function of significant wave height and wind speed were established in each latitude range based on the Jason\|2 altimeter data。The performances of this model were evaluated by the explained variances and the residuals.Compared with the estimations of the traditional parametric model under the same dataset,the results showed that the segmented parametric model of latitude was superior to the latter,especially in low\|latitude regions.And the precision of this model is near the nonparametric estimations in Jason\|2 altimeter GDR.These suggested that the segmented parametric model of latitude can effectively increase the precision of the sea state bias estimations,further improve the correction level of the sea surface height.
Firstly,obtaining 215 long time series ASAR Level 1B data which are from different ASAR measurement modes.After the process of geometric rectification,radiometric calibration,polarization conversion,noise removal,image resampling,converting all the Level 1B data into the normalized radar backscatter coefficient images with the same pixel spacing of 1 km× 1 km.Secondly,wind directions from cross-calibrated muti\|platform (CCMP) were used as the initial wind directions to retrieve wind speeds from SAR images by using three most commonly used geophysical model functions (GMF):CMOD4,CMOD-IFR2,CMOD5.The comparison of wind speeds retrieved from SAR with interpolated CCMP wind speeds and with in\|situ meteorological stations wind speeds showed that wind directions from CCMP could be used as the initial wind directions for wind retrieval using SAR images,and using CMOD4 mode function could get the highest accuracy retrieval wind speeds.Since CCMP data valid for a long time,covering a wide range and easily accessible,the combination of CCMP wind direction and SAR images have an advantage for operational wind field retrieval in the future,a useful try has been carried out in this experiment.
Wave numerical simulations of East China Sea,Eastern Pacific and central Pacific waters near Hawaii are carried out based on third-generation wave model WAVEWATCH Ⅲ,using NCEP/QuikSCAT wind field as model mixed input.In East China Sea,it is obtained that significant wave height and the reference value of buoy has a strong correlation coefficient,and the root mean square error of wave height is about 0.5 m by simulation;In the Pacific waters near the east coast and Hawaii,significant wave height values are generally lower than the reference value of buoy,root mean square error of wave height of different months is between 0.4 and 1.2 m,but the simulation of wave height and the reference value of buoy still has strong correlation.The results show that it is feasible that taking advantage of WAVEWATCH Ⅲ combined NCEP/QuikSCAT mixed wind field to simulate the wave height of East China Sea.However,it is still necessary to consider some non-wind factors in simulation of wave height lies at Pacific Ocean and the eastern open deep waters.
Vegetation water content is one of the important factors that influence and evaluate of vegetation growth.This paper aims to the property of hyperspectral data,the spectrum characteristics of vegetation were accurately extracted.Based on the continuum removal,a novel model called bi-inverted Gaussian model for the extraction of absorption characters was explored.Firstly,according to the vegetation spectral absorption characteristics,bi-inverted Gaussian model was founded.Then,in order to verify the validity of this approach,the ground data and the Hyperion hyperspectral remote sensing data was used.The results showed that the depth and the symmetry had the linear correlation with vegetation water content,and the coefficient of determination reached to 0.86 and 0.76,RMSE reached to 0.797 and 1.112.The model was verified and this paper proved the feasibility for classification of vegetation by using hyperspectral data.
Forest disturbance play an important impact on terrestrial ecosystems.Remote sensing technique has become the most important way to detect the forest disturbance at regular intervals and in a sequential manner because of the capacity of obtaining large area synchronous forest observation data at regular intervals.Forest disturbance monitoring based on time series data is becoming the main method.Fujian Province is taken as a case study.Five kinds of forest disturbance indices of DI,IFZ,NBR,NDMI and NDVI,and the different disturbance types spectral response capacity are studied,and the classification accuracy is evaluated by using MODIS time series data set from 2001~2013.The results show that extraction capacity of DI for forest cutting,plant diseases and insect pests,and afforestation is strong,and NBR is most sensitive to forest fire,in addition,spectral response capacity of NDVI for four disturbance types is relatively weak.The separability index(SI) of DI and IFZ are higher than 1 for different disturbance,which indicate that these two indices can be used to monitor multiple disturbance types.The accuracy assessment shows that DI among the indices,has the highest extraction capability.Its total accuracy to monitor the different disturbance is the highest of 92.97% and its kappa coefficient reaches to 0.92,followed by IFZ,which has the total accuracy of 89.66% and kappa coefficient of 0.88.The monitoring accuracy of NBR and NDMI nearly are the same,and are higher than NDVI.
Due to the traditional building 3D modeling method existing time consuming and low accuracy,adopts the contact measurement,and only can get the building a small number of feature points and line data.Compared with the traditional measurement method,the ground three\|dimensional Laser scanning technology (Terrestrial Laser Scanner,TLS) method can be fast,efficient and non-contact high\|precision 3D building surface information,so its more traditional building 3D modeling method advantage is obvious.based on thou farmland meeting site as the research object,this paper first introduces the main characteristic of the research target and point cloud data collection scheme,after the building is modeled as a demand with high complexity,the point cloud data preprocessing are introduced in detail,and building 3D model reconstruction related core technologies and methods,and focus on the relevant point cloud data registration splicing,simplified denoising,two\|dimensional contour extraction,three-dimensional entity reconstruction and so on,the paper finally realized the high complexity of thou farmland conference site three\|dimensional geometric model,USES the advanced 3D printing press 1∶40 scale production of miniature 3D printing point cloud data.Compared with the field measurement data analysis,the ground laser radar measurement method is used to collect the cloud data structure modeling precision is superior to the traditional measuring method.This article research results can be applied to thou farmland conference site,such as the restoration of cultural heritage,deformation monitoring,virtual reproduce,etc.
Unlike the conventional spaceborne SAR,the integration time of L\|band geosynchronous circular SAR (GEOCSAR) can be as long as 24 hours,the orbit altitude is about 36 000 km and the coverage may reach 1/3 of the Earth’s surface,the spatial and temporal variation of the trospheric refractivity and the ionospheric electron content may impose significant effects on repeat pass interferometric performance of GEOCSAR.Based on the spatial and temporal variation characteristics of the atmospheric medium,and the interferometic signal model,we analyze the effects of the trosphere and ionosphere on GEOCSAR interferometric performance.The effects from four aspects are mainly discussed including the phase delay caused by the troposphere and the ionosphere,the image shift on XY plane caused by the ionospheric horizontal gradient,focusing performance degradation caused by the diurnal variation of the trosphere and the ionospheric turbulence,and the Faraday rotation effect.The analysis and the simulation results show that the troposphere and ionosphere both impose very important influence on the GEOCSAR interferometric processing,the temporal variation of the ionospheric electron content during the synthetic aperture and the ionospheric turbulence may cause image defocused and then severely reduce the interferometric coherence.
Taking advantage of the adaptive detection feature of the random forests method,a classification method for coastal tidal flat area has been developed on basis of the HJ satellite images.Two typical areas covering coastal tidal flats,namely the ecological demonstration area of coastal tidal flats and the elk reserve,in the city of Dafeng,Jiangsu Province were selected in this study to showcase the classification effectiveness.The parameter settings and its accuracy were discussed.The results show that the random forests method performs obviously better than other traditional supervised classification methods such as parallelepiped,minimum distance,and maximum likelihood methods,and presents high stability in classification performance.
Target detection is one of the important content in remote sensing imagery information extraction,however,with the increase of image size and the interference of similar objects,the false alarm rate of target detection increase obviously.This paper built a multispectral remote sensing imagery target detection method (LCLCM) by combining the linearly constrained minimum variance (LCMV) with the local contrast method (LCM):first,using the correlation matrix of some targets to partial unmix image,then,adding the spatiality to enhance the target information and inhibit the background information,finally,normalizing and segmenting the image.Taking the boat in Landsat 8 multispectral imagery as the target to test this method,the false alarm rate of LCLCM is 1.07% and better than that of LCMV and LCM,which are 12.39% and 11.26%,respectively,showing that the method could detect target effectively and robustly.
In the process of hyperspectral image classification using the tri_training algorithm,the labels of unlabeled samples have error labels when the amount of initial training samples is small.In this paper,we propose a novel tri_training based on spatial neighborhood information(tri_training_SNI) to solve the problem for the tri_training algorithm.Firstly,we choose three basic classifiers from MLR(Multinomial Logistic Regression),KNN(k\|Nearest Neighbor),ELM(Extreme Learning Machine) and RF(Random Forest) classifier based on disagreement measure and disagreement\|accuracy.These classifiers are redefined using unlabeled samples in the tri_training_SNI process.In detail,in each round of tri_training_SNI,unlabeled samples are labeled for a classifier by the following two steps.Step 1:the first selection of unlabeled samples is constructed under certain conditions that the other two classifiers have the same labels.Step 2:spatial Neighborhood Information of initial training samples based on 8\|neighborhood is applied in this proposed approach to construct the secondary selection of unlabeled samples and the labels of unlabeled samples.Then the final classification results are produced via majority voting by the classification results of three classifiers.Experiments on two real hyperspectral data indicate that the proposed approach can effectively improve classification performance.
Accurate extraction of crop acreage has been one of the main concerned issues of agricultural remote sensing.Timely information about crop acreage at regional and national scales is also essential for predicting crop yields,and agricultural planning.In this paper,a new crop identification method is proposed combining with medium\|resolution and low\|resolution remote sensing data.Firstly,based on the differences of NDVI time series curve for various types of vegetation,we analyze the identifying characteristics of maize on the seasonal rhythm,and build an identification model.Then,the maize pure pixels are identified according to the closeness with the standard NDVI curve of maize.For the mixed pixels from maize and other vegetation,their sub\|pixel NDVI time series are extracted based on pixel unmixing method and the sub\|pixels are identified according to the model above;further,the identification results are repositioned to the medium\|resolution scales according to the spatial relationship.The mixed pixels area from maize and other crops are identified based on spectral differences in TM remote sensing image.Finally,the identified results are integrated into the medium resolution scale.In the dominating agricultural area of the Yiluo basin,the identified results of maize show that the acreage is 132 704 hm2,and the accuracy is 90.33%.The method proposed by this paper improved the identification accuracy and provided a new perspective to solve problems for extraction of crop cultivation information.
To accurately extract the growing area of paddy rice is a significant premise of paddy rice production and food security under the background of climate change.based on the current situation that paddy rice extraction is beset with difficulties in southern china,where clouds and rain appear in high frequency during growing seaon,how to take full advantage of the limited images to obtain accurate paddy rice planting area is a desiderated problem.In this study,we combined remotely sensed data from two different dates and brought out D\|value bands,using object\|oriented Random Forest to achieve the goal of rice extraction.The D\|value bands,indicating the difference between a character derived from two different time phases,can be generated from traditional characteristic bands including vegetation indexes,water index,prominent component analysis and Tasseled Cap results,as well as the original bands.We applied this method to extract paddy rice planting area in Dingcheng District,Changde,Hunnan Province,China,and results show that,the accuracy of paddy rice extraction was improved to 93% by 3 percent compared with single\|phased method,and the kappa coefficient reaches 91 in the study area.To further analyze the effect of D\|value bands in other classifiers,we compared the accuracy of combination of D\|value bands with decision tree and Random Forest,separately.Results show that the D\|value bands provides infromation in both subject segementation and classification,which can effectively improve the accuracy of paddy rice planting area extraction.
Understanding the link between greenspace patterns and land surface temperature is very important for mitigating the impact of urbanization on Urban Heat Island (UHI) and also useful for professionals and decision\|makers during the design phase of urban planning.Previous most researches were conducted to examine the cooling effects of greenspace amount,composition and configuration (spatial arrangement).Few have examined the cooling effects of spatial layout of greenspace (referring to space combination form of greenspace).Five scenarios of greenspace patterns are selected to examine them how to affect the horizontal and vertical distribution of the temperature coupling remote sensing data with a Computational Fluid Dynamics (CFD) model(ENVI\|met model).The simulation results demonstrate that zonal pattern always has the best cooling effect;dotted pattern has the best cooling effect on local ambient temperature.The temperature distributions have many similarities between radial and wedge patterns,which can become the “cold source” of urban thermal environment and play an important role in the improvement of the urban ecological environment.
Taking HJ\|1A and MODIS as data source,the comparison study of HJ\|1A NDVI and MODIS NDVI time series for extracting the vegetation phenology based on the method of dynamic threshold is analyzed qualitatively and quantitatively.Through compared results,some issues of application of HJ\|1A NDVI on vegetation can be analyzed which will improve the application of relatively high spatio\|temporal remote sensing data on the urban vegetation.The results showed that the standard deviation of MODIS NDVI time series is small on the main phenological points of SOS,EOS,LOS and TOMS.There is a big difference from the standard deviation of HJ NDVI time series which is relatively big,however,on the main phenological points of POS,BOS and AOS,the standard deviation and the degree of discrete of HJ NDVI time series is small.For the different types of vegetation,HJ\|1A NDVI time series can be used to extract the vegetation phenology.
Based on GPS reference data since 2010 surrounding the epicenter of 2016 Menyuan M6.4 earthquake,the characteristics of crustal deformation before the earthquake are discussed on the basis of the analysis results of the displacement time\|series,the processing to remove the periodization,baseline time\|series and strain time\|series.The results show that the displacement time\|series,especially the EW component of each GPS station appear a synchronous abnormal change since a month before the earthquake,the abnormal range is about 6~10 mm within 5 days.In addition,the results of remove the periodization of QHME and QHQL station show a biggest tendency change about the EW and NS component in the latter half of the year 2015.The baseline shortening rate smaller significantly near the epicenter,which shows the regions near the epicenters may in an obvious background of strain accumulation.Moreover,the EW strain and plane strain rate began to weaken since 2014.It indicates that the existence of deformation loss region in the past two years,a certain strain accumulation background may exist before the earthquake as well.
Recently,the citation of scientific data is gradually developed and focused by many researchers around the world.Earth sciences involve multiple research fields,and hence accumulates a large amount of scientific data from many research studies.This study illustrates the current situation and shortcomings of scientific data sharing system based on the data sets provided by the Cold and Arid Regions Science Data Center,which has citation results from 2006 to 2013 (the data of 2013 is not complete).The results show that it takes about 2 years for the data set to be widely cited since being public,and two years later the citations are in substantial growth.The remote sensing data get the top position in the citation statistics,followed by land and meteorological data.The fact that remote sensing and meteorology data are cited at the highest frequency reflects users’ great demand for this type of data.
In recent years,oil spill events in the oceans have been increasing,mainly caused by oil exploration and tank transportation.It has become one of the key environmental issues in the world.This study uses the full\|polarization data of Radarsat2 to detect onshore oil spills in the northern part of the South China Sea (SCS).The data was processed by three steps:①Compare of three full-polarization filter methods by BOX,GAUSS,and LEE,and find LEE is the best;②Apply LEE filter and six polarization parameters to process the oil spill images;③Use the threshed method to determine the scope and estimate the area.The result show that LEE filter is the best,the depolarization\|index parameter is best.Otsu is selected to determine the scope and estimate oil spills affected area.This study show:the key of process image is to elect effect polarization parameters;the depolarization\|index is the best parameter,it not only further remove speckle but also effect distinguish oil spills and the ocean water for next step extract oil spills.
Based on satellite derived NO2 column data from OMI,the characteristics of spatial and temporal distribution of tropospheric NO2 column density and its impact factors over China during 2005~2015 is analysed.Results demonstrate:①Tropospheric NO2 column density had a small fluctuation during 2005\|2009,and had a larger increase during 2010~2011,and declined in 2012,and had a sharp drop during 2014~2015.②The area of the high tropospheric NO2 column density changed significantly during 2005~2015,the area of the highest level had been on the rise during 2005\|2011,and reached a peak with 37.2 million square kilometres in 2011,and kept stable during 2011~2013,and dropped sharply during 2014~2015,and shrunk to 6.1 million square kilometres in 2015.③ Tropospheric NO2 column density over Shanghai and Tianjin was highest and in the fifth level.Shanghai is the city with the highest NO2 concentration of China,and Shandong is the province with the highest concentration of China.④Pollution sources are determined by the industrial and energy structure to a large extent.Atmospheric environment should be improved by optimizing industrial structure to reduce the proportion of secondary industry.It is an important reason for high NO2 concentration that is long\|term dependence on the energy structure of coal high pollution,and is urgent to develop new energy to replace coal fuel.Other reasons for NO2 emissions increase are the rapid increase of the number of vehicles,and vehicle standards and oil dropping behind the international development level.
Fractional Vegetation Cover (FVC) is a vital indicator for ecological environment monitoring.However it is quite difficult to obtain the high\|accuracy FVC in the mountainous regions,especially in southeastern China,due to the topographic effects.In this research,a mountainous area in Yongding county Fujian Province was selected as study area,and Landsat\|8 OLI was employed for the FVC retrieval.In order to choose a suitable model for the FVC retrieval in this area,the Dimidiate Pixel Model(DPM) and the Linear Spectral Mixture Analysis(LSMA) were performed to quantitatively evaluate the effects of mountainous topography.Then,the Normalized Difference Mountain Vegetation Index(NDMVI) was selected to calculate the FVCs of Yongding county in 1992,2002 and 2014,respectively.The results show that the model using NDMVI is found to be the less sensitivity in the hilly areas after testing the four models for retrieving FVC,and it is quite suitable for retrieving FVC in southeastern China.What’s more,the average value of FVC in Yongding county is greater than 77.99%,and more than 59.73% of the county are cover with high-density vegetation.Additionally,The FVC increase markedly during the period from 1992 to 2002,and decrease from 2002 to 2014.Some special regions such as Gaokanfu,Jinfeng and the west of Yongding county are cover with relatively low\|density vegetation and change obviously during these years.And the values of FVC are greater than before in Jinfeng region,but the vegetation are seriously damaged by mining in Gaokanfu region and the FVC are quite smaller in last 12 years.
In this paper,a systematic research on key technologies in the web sharing of remote sensing image was conducted.Functions of visual querying、browsing and download for massive remote sensing data are analyzed and designed.At last,we developed massive remote sensing data sharing prototype system and provided the service of data sharing based on the Web and other related computer technologies.
Remote Sensing Technology and Application has been published for 30 years,which is concentrated on remote sensing,GIS and the related issues.This article uses the detailed bibliographic data obtained from database(CSCD,CNKI and ISI WOS database) to analyze the publication trends of Remote Sensing Technology and Application in 30 years.Through analysis on articles' number,discipline classification,core-authors group,research hotspots,the annual variation of these indices have been illustrated,so do the research content and direction,research institutes and researchers' distribution change.The results indicate that in recent 30 years,the number of articles published presented a swift growth tendency;Landsat and MODIS are the two highest frequencies' remote sensing data.The results of bibliometric analysis on cited literatures about Remote Sensing Technology and Application shows that the citation frequencies by CSCD and SCI and the influence factors all increase unceasingly,which further explains the attention of researchers to remote sensing and the related issues is strengthened more and more.The development of technology,such as computer technology,data science,Big data will promote the development of remote sensing science;In addition,natural disasters,atmospheric environmental problems,food security,sustainable development and the Future Earth strategy,all need further development of remote sensing science,which need to develop more data source,systemic platform,and the process for more intelligent decisions.
A series of serious problems caused by climate change have becoming a focus.In the early 21th century,a new core plan of World Climate Research Program (WCRP),Climate and Cryosphere (CliC) was initiated,as marked Cryosphere research as an international research hotspot.Due to sensitivity of glacier and snow on Qinghai\|Tibet on climate change,possible influence of geological disasters on the surrounding countries,China,India,the United States,Japan and Germany arranged and carried out a series of research plans and projects in recent years,and published lots of papers about mechanism and application of spatial observation of glacier,snow and geological disasters on the Tibet Plateau.Bibliometrical analysis can reveal related research progress and trend.In this paper,SCIE retrieved the relevant papers,research reviews,conference papers and other relevant documents from 2001 to 2015.The overall of study,the major research subject (countries and institutions) and hot spots in different periods,distribution of subjects,research cooperation and future development trend and other features were analyzed.
