Lingli Tang

Spectral characteristics and feature selection of hyperspectral remote sensing data

Hyperspectral remote sensing data with bandwidth of nanometre (nm) level have tens or even several hundreds of channels and contain abundant spectral information. Different channels have their own properties and show the spectral characteristics of various objects in image. Rational feature selection from the varieties of channels is very important for effective analysis and information extraction of hyperspectral data. This paper, taking Shunyi region of Beijing as a study area, comprehensively analysed the spectral characteristics of hyperspectral data. On the basis of analysing the information quantity of bands, correlation between different bands, spectral absorption characteristics of objects and object separability in bands, a fundamental method of optimum band selection and feature extraction from hyperspectral remote sensing data was proposed.

Analysing the vegetation cover variation of China from AVHRR-NDVI data

In this paper, the characteristics of vegetation cover and variation in China have been studied by using the AVHRR NDVI time‐series data from 1981 to 2001. The Harmonic Analysis of Time Series (HANTS) method was successfully applied to eliminate the clouds on remote sensing data and reconstruct cloud‐free time series images. Then, the Fourier components of NDVI time series data were calculated. Finally, the physical meaning of Fourier components was analysed, and the relationship between Fourier components and land vegetation cover variation was investigated. The mean NDVI, or zeroth‐order harmonic, indicates overall vegetation cover level. The first harmonics of the HANTS summarizes the amplitude and phase of annual values of NDVI data, and the second harmonics of the HANTS summarizes those of biannual values of NDVI data. The amplitude of the first harmonic indicates the variability of vegetation productivity over the year. The phase of the first harmonic summarizes the timing of vegetation green‐up, while the second harmonic indicates the strength and timing of biannual vegetation cover variation. The Fourier components calculated by HANTS algorithm reveal the vegetation distribution and growing cycle characteristics. The physical meaning of Fourier components are significant to the land‐surface vegetation variation study of China. The methodology proposed in this paper is an effective method for the processing, analysis and application of long‐time‐series remote sensing data.

Evaluation of Temperature and Emissivity Retrieval using Spectral Smoothness Method for Low-Emissivity Materials

Land surface temperature and emissivity separation (TES) is a key problem in thermal infrared (TIR) remote sensing. Along with the development of civil applications, increasing numbers of man-made low-emissivity materials can be found around our living environment. In addition, the characteristics and variation in properties of those materials should also be concerned. However, there are still few TES methods for low-emissivity materials reported in the literature. This paper addresses the performance of the automatic retrieval of temperature and emissivity using spectral smoothness (ARTEMISS) method proposed by Borel (2008) for the retrieval of temperature and emissivity from hyperspectral TIR data for low-emissivity materials. The results show that those modeling errors are less than 0.11 K for temperature and 0.3% for emissivity as shown in the ARTEMISS algorithm if atmospheric parameters and the mean emissivity of material spectra are known. A sensitivity analysis has been performed, and the results show that the retrieval accuracy will be degraded with the increase of instrument noises, the errors of the atmospheric parameters, and the coarser spectral resolution. ARTEMISS can give a reasonable estimation of the temperature and emissivity for high- and low-emissivity materials; however, the performance of the algorithm is more seriously influenced by the atmospheric compensation than by the instrument noises. Our results show that the errors of temperature and emissivity become approximately three times than that when the instrument spectral properties are 1 cm-1 of sampling interval and 2 cm-1 of FWHM, and 4 cm-1 of sampling interval and 8 cm-1 of FWHM, respectively.

A practical SNR estimation scheme for remotely sensed optical imagery

Signal-to-Noise Ratio (SNR) is one of the basic and commonly used statistic parameters to evaluate the imaging quality of optical sensors. A lot of SNR estimation algorithms have been developed in various research fields. However, one intrinsic fact is usually ignored that SNR is not a constant value, but a quantity changing with the incident radiance received by the sensor. So SNR values estimated on different images through commonly used method are not comparable due to their distinct intensity levels between the images. Here we proposed a normalized SNR estimation scheme which can be readily applied to remotely sensed optical images. With this scheme SNR values obtained from different images can be of comparability, thus we can easily evaluate the performance degeneration of the sensor with more sufficient reliability.

Impact of spatial LAI heterogeneity on estimate of directional gap fraction from SPOT-satellite data

Directional gap probability or gap fraction is a basic parameter in the optical remote sensing modeling. Although some approaches have been proposed to estimate this gap probability from remotely sensed measurements, few efforts have been made to investigate the scaling effects of this parameter. This paper analyzes the scaling effect through aggregating the high-resolution directional gap probability (pixel size of 20 meters) estimated from leaf area index (LAI) images of VALERI database by means of Beers law and introduces an extension of clumping index, Ĉ, to compensate the scaling bias. The results show that the scaling effect depends on both the surface heterogeneity and the nonlinearity degree of the retrieved function. Analytical expressions for the scaling bias of gap probability and Ĉ are established in function of the variance of LAI and the mean value of LAI in a coarse pixel. With the VALERI dataset, the study in this paper shows that relative scaling bias of gap probability increases with decreasing spatial resolution for most of land cover types. Large relative biases are found for most of crops sites and a mixed forest site due to their relative large variance of LAI, while very small biases occur over grassland and shrubs sites. As for Ĉ, it varies slowly in the pure forest, grassland and shrubs sites, while more significantly in crops and mixed forest.

Earthquake prediction based on Levenberg-Marquardt algorithm constrained back-propagation neural network using DEMETER data

It is a popular problem that the mechanisms of earthquake are still not quite clear. The self-adaptive artificial neural network (ANN) method to combine contributions of various symptom factors of earthquake would be a feasible and useful tool. The back-propagation (BP) neural network can reflect the nonlinear relation between earthquake and various anomalies, therefore physical quantities measured by the DEMETER satellite including Electron density (Ne), Electron temperature (Te), ions temperature (Ti) and oxygen ion density (NO+), are collected to provide sample sets for a BP neural network. In order to improve the speed and the stability of BP neural network, the Levenberg-Marquardt algorithm is introduced to construct the model, and then model validation is performed on near 100 seismic events happened in 2008.

Permanent target for optical payload performance and data quality assessment: spectral characterization and a case study for calibration

Abstract To regularly evaluate the optical payload performance (geometric, radiometric, and spatial resolution) and the data quality for high-resolution airborne and satellite imaging systems, two new permanent targets (the knife-edge target and the fan-shaped target) made of gravel and with the advantages of year-round availability, lower maintenance operations, and a long lifetime were established in the Academy of Opto-Electronics Baotou site in China. The spectral properties of these targets are investigated in this study. Note that the anisotropy factor at 550 nm for the white gravel is approximately 6%, 12.5% 16.5%, 17.5%, 11.5%, and 5% at the principal plane for the observer zenith angle of 60 deg, 50 deg, 40 deg, 30 deg, 20 deg, and 10 deg (backscatter), respectively. The corresponding value for the gray gravel is 20.8%, 24.8%, 29.4%, 23.8%, 13%, and 3.7%, respectively, and 62.8%, 65.7%, 59.2%, 40.3%, 22.3%, and 9.0%, respectively, for the black gravel. The anisotropy of the black gravel is larger than that of the gray and white gravel areas. The nonuniformity of the target reflectivity is within 2.5%. Furthermore, a calibration for the optical payloads onboard the GF-1 satellite is performed with the knife-edge target, and the uncertainty analysis demonstrates that the uncertainty for this calibration is < 2.12 % when the relative error for the surface reflectance measurement, the aerosol optical depth, and the total column water vapor are approximately 1%, 10%, and 10%, respectively.

The assessment of in-flight dynamic range and response linearity of optical payloads onboard GF-1 satellite

Dynamic range and response linearity are two key parameters for impacting the quality of remote sensing image and subsequently the quantitative applications. Due to the space radiation and the degrading of electronic devices, the inflight dynamic range and response linearity of remote sensing payload are subject to change, and is essential to be assessed. Therefore, in this paper, with the aid of the permanent artificial target located in the AOE Baotou site in China, the two parameters for pan-chromatic camera (Pan) and the multi-spectral camera (Band 1-4) onboard GF-1 satellite are assessed with an extrapolation method using the in situ measurements and corresponding images acquired on November 4, 2013. The results show that the low point of the dynamic range for Pan band, Band 1, Band2, Band3 and Band4 is -24.08 W•sr-1m-2μm-1, -52.22 W•sr-1m-2μm-1, -35.20 W•sr-1m-2μm-1, -31.92 W•sr-1m-2μm-1, -24.07 W•sr-1m-2μm- 1 respectively; while the corresponding high point is 271.77 W•sr-1m-2μm-1, 401.58 W•sr-1m-2μm-1, 287.46 W•sr-1m- 2μm-1, 237.33W•sr-1m-2μm-1, 307.49W•sr-1m-2μm-1, respectively; meanwhile, it is demonstrated that all the sensors have a response linearity error of lower than 1%. Moreover, an analysis for this assessment is performed in terms of the uncertainties for surface reflectance measurement (1%), aerosol optical depth (10%), column water vapor (10%), MODTRAN model (1%) and solar irradiance (1%) using a simulation method with the aid of MODTRAN 4.0 model, and a total uncertainty of 2.12% is acquired.

Spectral characterization of the Dunhuang calibration/validation site using hyperspectral measurements

There is a great need to establish satellite instrument calibration consistency using vicarious calibration sites with wellknown spectral and radiometric characteristics. The Dunhuang calibration/validation site located in the Gobi desert, China, and the Dome C in the Antarctic are two of the eight reference sites endorsed by the Committee on Earth Observation Satellites (CEOS) Working Group on Cal/Val (WGCV). This paper presents results of the spectral characterization of the Dunhuang calibration/validation site using hyperspectral measurements. Its spectra are compared with those from the Dome C, as well as with ground sample measurements with radiative transfer calculations. Further comparisons are made with the spectra obtained by the HSI on Chinas HJ-1A environmental satellite. The results show that the Dunhuang and Dome C sites have distinct spectral characteristics that complement each other for satellite instrument calibration in the reflective solar bands. This study is part of the collaboration effort by the CEOS/WGCV towards consistent satellite observations for the global earth observation system of systems (GEOSS).

Land surface temperature retrieved from airborne multispectral scanner mid-infrared and thermal-infrared data.

Land surface temperature (LST) is one of the key parameters in the physics of land surface processes at local/global scales. In this paper, a LST retrieval method was proposed from airborne multispectral scanner data comparing one mid-infrared (MIR) channel and one thermal infrared (TIR) channel with the land surface emissivity given as a priori knowledge. To remove the influence of the direct solar radiance efficiently, a relationship between the direct solar radiance and water vapor content and the view zenith angle and solar zenith angle was established. Then, LST could be retrieved with a split-window algorithm from MIR/TIR data. Finally, the proposed algorithm was applied to the actual airborne flight data and validated with in situ measurements of land surface types in the Baotou site in China on 17 October 2014. The results demonstrate that the difference between the retrieved and in situ LST was less than 1.5 K. The bais, RMSE, and standard deviation of the retrieved LST were 0.156 K, 0.883 K, and 0.869 K, respectively, for samples.