In consideration of within-class endmember variability, it is realistic to use multiple endmembers to model a pure class. We propose an advanced multi-endmember unmixing algorithm based on twin support vector machines (UTSVM), which derives the abundances based on the distances from the mixed pixels to each classification hyperplane. Unmixing uncertainty, an issue often neglected in multi-endmember unmixing, is also analyzed quantitatively for UTSVM. Two types of unmixing uncertainty, abundance overlap (i.e., different mixed pixels have the same abundances) and model overlap (i.e., one mixed pixel may be unmixed into different abundances), are introduced. Abundance overlap angle and abundance variability scale (AVS) are defined as two uncertainty indexes to measure abundance overlap and model overlap, respectively. The relationship between within-class endmember variability and unmixing uncertainty is discussed. When the unmixing uncertainty is high, we propose to use the mean value of abundances within AVS as the estimation of abundance to obtain the best compromised results. Experimental results show the feasibility and effectiveness of our study.
In VHF pulse Ground Penetrating Radar(GPR) system, the echo pass through the antenna and transmission line circuit, then reach the GPR receiver. Thus the reflection coefficient at the receiver sampling gate interface, which is at the end of the transmission line, is different from the real reflection coefficient of the media at the antenna interface, which could cause the GPR receiving error. The pulse GPR receiver is a wideband system that can't be simply described as traditional narrowband transmission line model. Since the GPR transmission circuit is a linear system, the linear transformation method could be used to analyze the characteristic of the GPR receiving system. A GPR receiver calibration method based on transmission line theory is proposed in this paper, which analyzes the relationship between the reflection coefficients of theory calculation at antenna interface and the measuring data by network analyzer at the sampling gate interface. Then the least square method is introduced to calibrate the transfer function of the GPR receiver transmission circuit. This calibration method can be useful in media quantitative inversion by GPR. When the reflection coefficient at the sampling gate is obtained, the real reflection coefficient of the media at the antenna interface can be easily determined.
Snow depth parameter inversion from passive microwave remote sensing is of great significance to hydrological process and climate systems. The Helsinki University of Technology (HUT) model is a commonly used snow emission model. Snow grain size (SGS) is one of the important input parameters, but SGS is difficult to obtain in broad areas. The time series of SGS are first evolved by an SGS evolution model (Jordan 91) using in situ data. A good linear relationship between the effective SGS in HUT and the evolution SGS was found. Then brightness temperature simulations are performed based on the effective SGS and evolution SGS. The results showed that the biases of the simulated brightness temperatures based on the effective SGS and evolution SGS were −6.5 and −3.6 K, respectively, for 18.7 GHz and −4.2 and −4.0 K for 36.5 GHz. Furthermore, the model is performed in six pixels with different land use/cover type in other areas. The results showed that the simulated brightness temperatures based on the evolution SGS were consistent with those from the satellite. Consequently, evolution SGS appears to be a simple method to obtain an appropriate SGS for the HUT model.
Soil surface temperature (Ts) is an important indicator of global temperature change and a key input parameter for retrieving land surface variables using remote sensing techniques. Due to the masking in the thermal infrared band and the scattering in the microwave band of snow, the temperature of soil surfaces covered by snow is difficult to infer from remote sensing data. We attempted to estimate Ts under snow cover using brightness temperature data from the special sensor microwave/imager. Ts under snow cover was underestimated due to the strong scattering effect of snow on upward soil microwave emissions at 37 GHz. The underestimated portion of Ts is related to snow properties, such as depth, grain size, and moisture. Based on the microwave emission model of layered snowpacks, the simulated results revealed a linear relationship between the underestimated Ts and the brightness temperature difference (TBD) at 19 and 37 GHz. When TBDs at 19 and 37 GHz were introduced to the Ts estimation method, accuracy improved, i.e., the root mean square error and bias of the estimated Ts decreased greatly, especially for dry snow. This improvement allows Ts estimation of snow-covered surfaces from 37 GHz microwave brightness temperature.
The precision of snow parameter retrieval is unsatisfactory for current practical demands. The primary reason is because of the problem of mixed pixels that are caused by low spatial resolution of satellite passive microwave data. A snow passive microwave unmixing method is proposed in this paper, based on land cover type data and the antenna gain function of passive microwaves. The land cover type of Northeast China is partitioned into grass, farmland, bare soil, forest, and water body types. The component brightness temperatures (CBT), namely unmixed data, with 1 km data resolution are obtained using the proposed unmixing method. The snow depth determined by the CBT and three snow depth retrieval algorithms are validated through field measurements taken in forest and farmland areas of Northeast China in January 2012 and 2013. The results show that the overall of the retrieval precision of the snow depth is improved by 17% in farmland areas and 10% in forest areas when using the CBT in comparison with the mixed pixels. The snow cover results based on the CBT are compared with existing MODIS snow cover products. The results demonstrate that more snow cover information can be obtained with up to 86% accuracy.
To improve snow depth (SD) inversion algorithms using passive microwave data, it is important to objectively assess their accuracy and to analyze their uncertainty. Some previous studies validated the inversion algorithms only using spatial data at a fixed time node, which is not objective or convincing. The spatiotemporal analysis of the SD inversion based on the FengYun-3B MicroWave Radiation Imager is performed in Heilongjiang Province, China. Based on the temporal analysis, the results show that the accuracy of SD inversion algorithms is different at different time phases throughout the winter. In cropland areas, the variation in snow properties, particularly the increase in snow grain and the presence of depth hoar, leads to underestimation and overestimation at the earlier and later phases, respectively. The spatial analysis shows that the SD in the high forest coverage regions is seriously overestimated due to the addition of a forest correction factor using the Chang algorithm. In addition, the complex underlying surfaces and hilly terrain are also influencing factors that result in the low accuracy for several regions. Therefore, the analysis and identification of these uncertainties are benefits not only in understanding the influential factors of SD inversion algorithms but also in developing better algorithms for the next generation of SD retrieval.
In recent years, the bio-optical model has been paid more and more attention. In order to validate its applicability in the
near-infrared wavelengths to Case II waters, two simply parameterized equations employing reflectance at 808nm and
873nm were established to estimate total suspended matter (TSM) concentrations in the Shitoukoumen Reservoir that
represented a turbid inland water condition. It was showed that both equations gave out comparative good performance
with coefficient determination (R2) larger than 0.85 and root mean squared error (RMSE) much lower than data span for
both training and test data. Based on the transfer of radiation in waters, the bio-optical model could integrate well
apparent optical properties (AOPs) with inherent optical properties (IOPs). However, further investigation is needed to
upgrade the bio-optical dataset and to refine the model for the universal applications.
Access to the requested content is limited to institutions that have purchased or subscribe to SPIE eBooks.
You are receiving this notice because your organization may not have SPIE eBooks access.*
*Shibboleth/Open Athens users─please
sign in
to access your institution's subscriptions.
To obtain this item, you may purchase the complete book in print or electronic format on
SPIE.org.
INSTITUTIONAL Select your institution to access the SPIE Digital Library.
PERSONAL Sign in with your SPIE account to access your personal subscriptions or to use specific features such as save to my library, sign up for alerts, save searches, etc.