Medium Resolution Spectral Imager (MERSI) is an important sensor carried by two on\|orbit Chinese polar-orbiting meteorological satellite.It enables images of the Earth with high spatial resolution and high temporal resolution.And its thermal infrared data can be used to retrieval high quality land surface temperature.In this paper,single\|channel method is developed by Jimёnez-Mu oz and Sobrino is improved in atmospheric parameter estimation and view angle correction,to more fit for thermal infrared channel spectral of FY3A/MERSI,so as to retrieval more accuracy and smooth LST.The improved algorithm is validated by simulation using MODerate Resolution Atmospheric TRANsmission (MODTRAN).The average error of LSTs at all view zenith angles is -0.6±2.2 K.The result also shows that introduction of view zenith angle factor fitting atmospheric parameters significantly improve the precision of LST.Finally,LSTs retrieved from FY3A/MERSI are compared with LSTs measured in Dunhuang Gobi site and MODIS LST products.The validation indicates MERSI LST is accurate in spatial distribution,and it has high precision.Average MERSI LST error is 1.74 K,and root mean square deviations are lower than 1.9 K relative to LST measured in site.Besides,LST difference between MERSI and MODIS in study area is 2.6307 K.Though its accuracy would be affected by the error of cloud detection and calibration of thermal infrared channel,LST retrieved from MERSI still has wide prospects for application in the related research and business.
This paper uses PROSPECT and SAIL model to simulate vegetation canopy reflectance under different chlorophyll contents,LAI and view zenith angles.The changes of NDVI with different LAI,view zenith angles and chlorophyll contents are analyzed.This study results show that chlorophyll mainly affects canopy reflectance in visible band.Canopy reflectance decreases with the increasing of chlorophyll content.Canopy reflectance increases with the increasing of view zenith angle.However,the influence of view zenith angles on canopy reflectance decreases when LAI is high.Under the same view zenith angle,with the increase of chlorophyll contents,corresponding NDVI values increase.Under constant content of chlorophyll,NDVI increases with the increase of LAI.When LAI is 1,under different chlorophyll contents,with the increase of view zenith angles,NDVI shows a trend of first decrease and then increase where the inflection point is at the zenith angle of 65°or 70°.However,NDVI presents a decreasing trend when LAI is 3.5 or 7.When chlorophyll content is higher,the influence of view zenith angles on NDVI is low.When LAI is great and chlorophyll content is low,NDVI decreases rapidly with the increase of view zenith angle (>70°).
In this paper,taking a multi\|channel microwave radiometer in Beijing for example,we analyzed the continuity of the “clear sky” brightness temperature data observed at eight and twenty o’clock (BT) during the 24 months in both 2010 and 2011 with the purpose of the brightness temperature data quality controlling.Using the atmospheric profile data from independent sources such as NCEP,we calculated the brightness temperatures through the radiative transfer equation and tested the consistency between the observed brightness temperatures and the simulated results in “clear sky” condition.This study has been found that the problem of discontinuous and inconsistency in observations is due to “Liquid Nitrogen Calibration” and “moving” which happened successively to the radiometer.And correcting for these influences based on piecewise fitting has been realized,and the continuity and consistency of the observed brightness temperature data has been improved.The result shows that quality control on the brightness temperature observations is necessary before further applications.
Brightness Temperatures(T B) for a microwave limb sounder in the millimeter and sub\|millimeter were numerically simulated by the radiative transfer model LBLRTM.A preliminary selection of wavelength ranges was made for observing trace gases in the upper atmosphere.TB sensitivity was then studied for all targeted gas species by changing their concentration profiles by 10%,50% and 100%.And dynamic ranges of TB variation with height and possible gas concentration variation were given based on the results of line position and strength analysis for 25 kinds of atmospheric species .At last,a scheme of five microwave radiometers limb sounder working in sub-millimeter to millimeter region was proposed by considering the technical feasibility and the concerns of the atmospheric species.
Forest fire is an important forest disturbance which leads to reduction in forest carbon stocks.Timely and accurately distribution information of forest burn scars is of great significance for the study of the global carbon cycle and climate change.Wuning,a county in Jiangxi Province,is taken as a study case.By using 1991 and 1992 TM images and the separability index (M),the Landsat TM 7 bands,vegetation index,burn scars index and the improvement burned index for the burned forest extraction capacity are analyzed,and the classification accuracy is evaluated.The results show that TM4,TM7 and TM6 band have higher distinguishing ability,which reflects the characteristics of the burned forest more precisely.NBR,NSTV1,NSTV2 and NSEV1 among 18 different remote sensing indexes have higher separability value,and the extraction accuracy for burned forest is higher than 85%,in addition,extraction capacity of EVI,BAI,MIRBI,SAVIT and NSEV2 is relatively weak.The improvement burned index NSTV2 is the highest as TM6 band is introduced,indicating that properly introduction of thermal infrared band can improve the extraction capacity of burned forest.
This paper mainly focused on some of the basic and critical questions of Integrated Water Vapor(IWV) inversion,including the radiative transfer simulation,hyperspectral data processing and the inversion method of the IWV.A new IWV retrieval algorithm based on the Principal Component Analysis PCA and Multi\|layer Feed\|forward back\|propagation Neural Network(MFNN) is developed,which is referred as PCA\|MFNN method.In this study,the radiative transfer simulation has been carried out with CRTM(Community Radiative Transfer Mode) combined with the latest global assimilated data which is from the CIMSS.The PCA is used to reduce the dimension,and eliminate the noise and correlation from the hyperspectral remote sensing data.A three layer forward neural network has been constructed,trained and tested,and then the IWV is retrieved using the optimized network.Compared with the AIRS L2 IWV product,the root mean square error of the retrieved integrated water vapor is 0.387 g/cm2.
The Urumqi Institute of Desert Meteorology of China Meteorological Administration (CMA) carried out an atmospheric scientific experiment on the detection of dust weather and Clear-air Turbulence using a wind\|profiling radar in the hinterland of the Taklimakan Desert in June\|August 2010.Based on the wind\|profiling data obtained from this experiment,this paper analyzes the characteristics of vertical intensity distribution of dust weather and desert summer clear air turbulence evolutionary characteristics.Studies show that:① Under the premise of a certain turbulence and atmospheric humidity,A wind\|profiling radar can effectively detect dust weather,the echo intensity of dust weather in Taklimakan is about -18~10 dBZe,indicates that the dust echo intensity is significantly weaker than that of precipitation.② The clear air turbulence development in Taklimakan Desert can reach 3 500~4 000 m altitude in summer,the turbulence is about 0.13 to 0.16 m/s lift velocity to upper air,Taklimakan Desert summer boundary layer height can reach about 3 500~4 000 m.
Leaf area index (LAI) is one of the most important biophysical parameters of vegetation and other land surface crop.In order to explore the feasibility of vegetation indices to estimate LAI,With Gonger prairie of Kesiketeng flag of Chifeng City in Inner Mongolia autonomous region as study object,several common broad\|band vegetation indices and narrow\|band vegetation indices together with 4 kinds of regression models were chosen to build up LAI inversion models.The simulation precisions for different models were analyzed and evaluated through comparing the simulated LAI and measured LAI.The results showed that PVI and MSAVI,which took soil background and environment into account,performed better than traditional vegetation indices NDVI and RVI in grass LAI inversion.It also found that narrow\|band vegetation indices showed obvious superiority compared with broad\|band vegetation indices.The R2 and standard error of the model between PVI and LAI were separately 0.65 and 0.15,which indicated that there existed good consistency between measured LAI and simulated LAI.Finally,the map of grass LAI was produced using Hyperion data and PVI mode.
The observations from the first space\|born Precipitation Radar(PR) are used to investigate the distribution of deep convection and overshooting events in the summer of Asian monsoon region to enhance the understanding for the effect of convection on atmospheric composition transportation in the Upper Troposphere\|Lower Stratosphere (UTLS).In this paper,the distribution of precipitation,convective rainfall and deep convection as well as overshooting events in the summer of Asian monsoon region are analyzed from the three\|year PR observation.In general,the occurrence frequency of those variables in the different years looks similar and is much close,which reflects that the distribution results are representable and reasonable.The distribution results over Asian monsoon region show that the higher probability of rainfall occurs around the Bengal\|bay,and the convective rainfall is the most significant in India,but is rather lower in China.For the deep convection,the occurrence probability is the highest about 15% in India,then the south\|west of China,southern China and part of South\|east of the Asia is the second high about 10%,and the most of remain area is lower than 5%.As to overshooting events,the distribution of convection overshooting 14 km is similar to that of deep convection,which shows that the highest frequency is occurred in India still with 2% probability,the second is the south\|west of China and part of South\|east of Asia with 1% occurrence,and no overshooting events in the most of remain place.For the convection overshooting the Tropopause height derived from NCEP data,its occurrence frequency is similar to the former,but the ratio decreased dramatically,such as less than 0.35%,which demonstrates that the convection overshooting events reduced dramatically with the increased referred Tropopause height.In addition,based on the preivious studies on tropopause layer (TL),our results show that the occurrence probability of the convection overshootting the lower boundary of TL is about 2%,while overshooting the whole TL is less than 0.4%。
In order to accurately analyze the dynamic change of land cover,this paper presented a change detection method based on object\|oriented classification.For two images with high spatial resolutions,the proposed approach was compared the land cover type after they are classified by object\|oriented method.According to the SPOT5 remote sensing imagines of Dongguan city,Guangdong Province between the years of 2005 and 2008,this paper adopted the change detection technique based on object\|oriented classification and further analyzed quantitatively the land cover change in the aspects of area and reciprocal transformation.The object\|oriented classification\|based change detection method is certified more accuracy than pixel\|oriented classification\|based change detection method.Moreover,using Markov chain model,the research simulated and predicted the dynamic change and evolution tendency of land cover over the study area in the next five years.Through this process of study,we provided a typical case for grasping the rules of land cover dynamic change over the study area.The results showed that construction land are still increasing,whereas cultivated land and wood land are decreasing.Furthermore,the change area of construction land and cultivated land were larger than other classes of land cover.In addition,the increasing of construction land is mainly caused by the decreasing of cultivated land,which speeds up the deterioration of the cultivated environment in the study area.In conclusion,the research could provide technical support for effectively using land and promoting the sustainable development of land and urbanization.
Ship classification and identification is of great significance for monitoring and management of marine transportation,and it’s an important part of the SAR ocean applications.In the high resolution SAR images of COSMO\|SkyMed,structural outlines of merchant ships are clear.Features of bulk carriers,container ships and oil tankers are legible,which provides support of ship classification effectively.This paper proposes an algorithm for merchant ship classification,which is based on structural feature analysis.Ship types are determined by feature extractions of Kernel Density Estimation,location of main axis,and the integration of left\|middle\|right part of the ship hull.Synchronous experiments performance in East China Sea shows that the average classification accuracy of the proposed algorithm is 89.94%.
When calculate Temperature Lapse Rates (TLR) in mountain regions,conventional meteorological stations are commonly underrepresented,poor synchronization,and easily distorted by instantaneous micro\|climate.In order to provide a difference method of TLR of land surface at the grid\|cell level,which has been developed that uses difference process on a per\|cell level with a custom neighborhoods instead of the first\|derivative operation,and considers TLR as the first derivative of temperature with elevation on the basis of temperature and Digital Elevation Model (DEM).Visual interpretation and quantitative analysis of TLR raster data at the entire regional scale and in typical landforms respectively showed that:(1) the micro\|pattern of TLR in Yunnan had great variation and represented certain rules with the strike of mountain,basin,lake or valley for the reasons of the complicated relationship between micro\|terrain factors and atmospheric circulation,among which the response of mountain with large vertical elevation to TLR was the most obvious;(2)TLR had significant distribution characters as change of undulating mountains especially that there were two ribbons of intensive change areas,of which one lied in transition zone from the valley bottom to mountain slope and another lied in transition zone from the upper part to the top of mountain;(3) The distribution zones that had intensive changes also existed in periphery of urban and large water body.This paper breaks through the limitation of traditional method with observation stations and is important for probing rules of TLR in micro\|scale.
Change monitoring of coastline is a significant task for coastline management,sea level change monitoring as well as coastal zone evolution.Remote sensing technology,with its high efficiency,wide coverage and low cost,is becoming an important means of monitoring coastline change.In this paper takes coastline of Qinhuangdao city as a case study,the data was extracted by SPOT satellite image and an integrated method considering coastal types and tidal correction.The accuracy of coastline extraction results was verified by the ground truthing data,which showed that it reached subpixel level.Based on the coastline extraction results of 2006,2009 and 2011,coastline change was analyzed.As for sandy coastline in Beidaihe area,baseline method was utilized to analyze quantitively the change.The result showed that the detected changes were quite consistent with the actual situation.
DEM (Digital Elevation Model) is an important feature for landslide recognition,but in practical application its difficult to obtain high\|resolution DEM in the study area.In order to analyze the impact of DEM resolution on landslide extraction and determine the resolution to the meet application requirements,this study resamples the 10 m resolution DEM into 20,30,60 and 90 m,and then adopts the object\|oriented method combined with high\|resolution remote sensing images and DEMs to extract the landslides information.The experimental results show that,when the DEM resolution is greater than 30 m,we can obtain better recognition and classification results for the landslide with area greater than 5 000 m2.When the resolution is less than 30 m,it is difficult to distinguish landslide types,but by adjusting parameter values we can still achieve the detection of landslides.This research provides certain guiding significance and reference value for the selection of DEM spatial resolution on the landslides information extraction.
Airship has high\|resolution,large coverage,could achieve fixed\|point observation,and deployed quickly soon.Its rapid deployment and planning research is a hot spot in mission planning,which could reduce site monitoring blindness and promote scientific decision making for disaster relief.Therefore,this paper proposed a new emergency disaster relief airship rapid deployment and planning method,which concerns both airship deployment and planning,and a deployment optimization model based on entropy information is built.Furthermore,a planning algorithm that uses maximum mission execution is given.Together with compatibility matrix and genetic algorithm,this algorithm can reduce planning complexity.Through simulation,this algorithm proved to be feasible.
Inspired by the recent success of Sparse Representation based Classification (SRC) framework in computer vision and pattern recognition areas,the HyperSpectral Image (HSI) processing community has witnessed a surge of papers focusing on the utilization of sparse prior for effective classification.In sparse representation based on HSI classification,relative difference information between similar classes can hardly be captured for the reason of randomly sampling.We first apply a novel fisher discriminative dictionary learning method,which improves the discriminative and the reconstruction capability of the dictionary.Secondly,motivated by the assumption that spatially adjacent samples are statistically related,a majority voting scheme incorporating contextual information is proposed to predict category labels.Experimental results demonstrated that this method can significantly improve HSI classification accuracy.
Extraction of road information quickly and accurately is great significant for GIS data updating.Taking Kunming city as study area,QuickBird satellite images as the data source,experiment on urban road feature semantics model and feature semantics information extraction of high resolution remote sensing image has been carried out.The results show that(1) the approach is feasible that OAR model of human cognition was introduced,cognitive framework of urban roads information from high resolution remote sensing image was presented,which can be expressed by using mathematics and logic rule languages,then feature semantic information was extracted finally.(2) the main steps of feature semantic information extraction are to use the Canny operator in edge detection of road,to refine feature of road,to track line based on node,then to implement semantic information extraction of object,semantic information extraction of spatial relationship,semantic information extraction of local context,and finally to optimize road network by GIS technology to finish the final road network extraction.The extraction result of road network is good,the length accuracy of extracted road was 89.19%,the width accuracy was 71.54%,and the intact rate was 50.32%.
Existing remote sensing image fusion algorithm has not high and poor ability of the fused image resalution to keep the spectral characteristics.So according to the disrectional wavelet transform the efficiency and direction selectivity,this paper proposed a multi\|spectral and panchromatic remote sensing image fusion algorithm based on directional wavelet.The proposed algorithm was preprocessing the images that were used for the human visual characteristics.And this algorithm was according to the wavelet decomposition characteristics.Using the Fusion rules based on energy ratio to deal with the wavelet coefficients of the low\|frequency sub-band.And it was using Fusion rules based on textures consistency to deal with the wavelet coefficients of the high\|frequency sub\|band.Experimental results show that this proposed algorithm can get the higher\|resolution image fusion reserve the multispectral image information at the same time.And it can get the best results when processed the medium-resolution remote sensing image.
The accurate information about the spatial distribution information of facility agriculture is crucial for supporting the government decision on the agricultural development and studying on the environmental impacts of facility agriculture.China has the largest facility agricultural area in the whole world.Traditionally,the information facility agricultural area is reported on the basis of census data. However,the census data cannot provide the detail spatial information.Hence,it is needed to develop a rapid method to automatically extract the spatial distribution information of facility agriculture.Object\|oriented classification shows great advantages on the objects with obvious textural and geometrical characteristic just like facility agricultural area.Therefore,this paper combined the object\|oriented classification and support vector machine classification to automatically extract the spatial distribution information of facility agriculture by using UAV aerial photo acquired at Dingzhou County,Heibei province.The overall accuracy of our classification result is 95.60% and Kappa Coefficient is 0.9022,which indicates that our newly proposed method can accurately extract the spatial distribution information of facility agriculture.
Compared with traditional LiDAR,Full-waveform LiDAR data can be digitally preserved in a very small sampling interval of the laser scattering echo pulse.Therefore the additional information reflecting the inherent characteristics of the surface features can be obtained after data processing.The accuracy of full\|waveform LiDAR data processing determines the effectiveness of target information acquisition.Among the data processing procedures,data correction is a key step to the waveform decomposition and target feature extraction.In this paper,a backscatter cross\|section correction model is proposed,which considers the differences of the same building reflection with different incidence angle.Airborne small footprint full\|waveform lidar data is used to verify the correction model.The results show that the backscatter cross\|section of building in same type can be normalization corrected by this correction model and the impact of different incidence angle is eliminated that improve the accuracy of applications such as building classification and feature extraction greatly.
This paper proposed a new Semi\|supervised dimensionality reduction method called Side information Constraint Semi\|Supervised Dimensionality Reduction(SCSSDR) based on Semi\|supervised dimensionality reduction framework.,which can preserve the intrinsic structure of the unlabeled data as well as global structure.For classification purpose,the SCSSDR take into account the discriminant information of labeled data,and finally get the projection matrix by solving an optimization problem.UCI data sets is used for quantitative analysis of the algorithm,and find that the method is superior to the traditional PCA and manifold learning algorithm,further experimental results on UCI data sets and hyperspectral data sets demonstrate the effectiveness of the method.
The new algorithm called M&H (Morphology and Hough) is aimed at extracting complicated curved roads.H&M extends the theory and techniques of Hough transform,mathematical morphology and nonlinear regression.Eventually,H&M realizes the goal of automatically extracting complicated linear features and instantly generating the vector graphics of the roads.After several tests using MATLAB,the results show that M&H has advantage in noise reduction than traditional morphological method,and is much better than Hough transform at curved road extraction.The data of accuracy tests shows that M&H decreases the rate of extraction losses by 18.9%,which compared with simply using standard Hough transform.
In this paper,we proposed an improved ISOMAP algorithm using neighborhood distance(ND\|ISOMAP).Neighborhood distance is used to successively approximate manifold distance indicating the manifold structure of high-dimensional data.In order to improve the computer rate of traditional ISOMAP,block\|matrix and auto\|color methods are applied in the ISOMAP(BA-ISOMAP).The dimension reduction of hyperspectral data is classified by Support Vector Machine(SVM),and the results show that the ND-ISOMAP performs the best overall accuracy (97.31% )than the traditional ISOMAP (75.01%).The combination method of ND\|ISOMAP and BA\|ISOMAP is with 89.61% overall classification accuracy,while its operation rate is almost 40 times than the traditional ISOMAP.
