Incorporating Endmember Variability into Linear Unmixing of Coarse Resolution Imagery: Mapping Large-Scale Impervious Surface Abundance Using a Hierarchically Object-Based Spectral Mixture Analysis

Comparison of methods for estimation of absolute vegetation and soil fractional cover using MODIS normalized BRDF-adjusted reflectance data

Evaluation of spectral unmixing techniques using MODIS in a structurally complex savanna environment for retrieval of green vegetation, nonphotosynthetic vegetation, and soil fractional cover

Impervious surfaces have been widely considered as the key indicator for evaluating urbanization and environmental quality. As one of the most widely applied methods, spectral mixture analysis (SMA) has been commonly used for mapping urban impervious surface fractions. When implementing SMA, the original multispectral remote-sensing reflectance images are served as the foundation and key to successful SMA. However, the limited spectral variances among different land covers from the original reflectance images make it challenging in information extraction and results in unsatisfactory mapping results. To address this issue, a new method has been proposed in this study to improve urban impervious surface mapping through integrating statistical methods and SMA. In particular, two traditional statistical methods, principal component analysis (PCA) and minimum noise fraction rotation (MNF) were applied to highlight the spectral variances among different land covers. Three endmember classes (impervious surface, soil, and vegetation) and corresponding spectra were identified and extracted from the vertices of the 2-D space plots generated by the first three components of each of the statistical analysis methods, PCA and MNF. A new dataset was generated by stacking the first three components of the PCA and MNF (in a total of six components), and a fully constrained linear SMA was implemented to map the fractional impervious surfaces. Results indicate that a promising performance has been achieved by the proposed new method with the systematic error (SE) of −3.45% and mean absolute error (MAE) of 11.52%. Comparative analysis results also show a much better performance achieved by the proposed statistical method-based SMA than the conventional SMA.

Urban environments are characterized by high spectral and spatial heterogeneity and, as a consequence, most urban pixels in moderate-resolution imagery contain multiple land-cover materials. Despite these complexities, virtually all urban land cover can be generalized as a combination of vegetation, impervious surfaces, and soil (V–I–S components), in addition to water. Previous work has demonstrated the potential of multiple endmember spectral mixture analysis (MESMA) to model the subpixel abundance of V–I–S components. Here, the authors test whether the technique is sufficiently robust to map V–I–S components for a diverse set of cities, selecting 10 urban centers in the state of Rondônia, Brazil, to represent a range of populations, development histories, and economic activities. For each urban sample, a 20 km × 20 km region centered over the built-up area was subset from Landsat Enhanced Thematic Mapper Plus (ETM+) imagery. MESMA was applied to all subscenes using the same spectral library, model constraints, and selection rules. Accuracy of the modeled V–I–S fractions was assessed using high-resolution images mosaicked from digital aerial videography. Modeled fractions and reference fractions were highly correlated, with R2 values exceeding 0.75 for all materials in multiple cities across a region. Model complexity, or the number of endmembers required to accurately model each pixel, was correlated with the degree of human impact on the landscape. Built-up areas, as delineated by model complexity, exhibited a strong fit to the well-established relationship between the built-up area of a settlement and its population. Finally, this work demonstrates that the V–I–S components as modeled by MESMA can capture both inter- and intraurban variability, suggesting that these data products could contribute to comparative studies of urbanizing areas through time and across regions.

Evaluation of potential of multiple endmember spectral mixture analysis (MESMA) for surface coal mining affected area mapping in different world forest ecosystems

In this paper we evaluated the potential use of Multiple Endmember Spectral Mixture Analysis (MESMA) applied to EO-1 Hyperion hyperspectral data to separate land covers (soil = dunes and dry fields; green vegetation = pinus, eucalyptus and grasslands; water = without sediments, with sediments, and with chlorophyll; and shade), in the southern state of Rio Grande do Sul, Brazil. The approach involved (a) preprocessing and atmospheric correction of Hyperion image; (b) sequential use of Minimum Noise Fraction (MNF), Pixel Purity Index (PPI) and n-Dimensional Visualizer techniques in the visible to shortwave range for the initial selection of a group of endmember, and another group of pixels for model validation; (c) use of the software Visualization and Image Processing for Environmental Research (VIPER) Tools to perform the final selection of endmembers based on the spectral library, and to obtain MESMA models; and (d) evaluation of resulting fraction images and RMSE values to determine the optimal number of endmembers of the MESMA model. Results showed that a four-endmember MESMA model described the diversity of the scene components, including that of materials within the same class (e.g., pinus and eucalyptus) and produced the largest fractions and the lowest RMSE values on a per-pixel basis. Results also showed the performance of MESMA applied to Hyperion data to discriminate properly land covers in the coastal plains, even considering the low signal-to-noise ratio of the instrument.

In order to improve the accuracy of multispectral image inversion of soil and plant analytical development (SPAD) of the cotton canopy, image segmentation methods were utilized to remove the background interference, such as soil and shadow in UAV multispectral images. UAV multispectral images of cotton bud stage canopies at three different heights (30 m, 50 m, and 80 m) were acquired. Four methods, namely vegetation index thresholding (VIT), supervised classification by support vector machine (SVM), spectral mixture analysis (SMA), and multiple endmember spectral mixture analysis (MESMA), were used to segment cotton, soil, and shadows in the multispectral images of cotton. The segmented UAV multispectral images were used to extract the spectral information of the cotton canopy, and eight vegetation indices were calculated to construct the dataset. Partial least squares regression (PLSR), Random forest (FR), and support vector regression (SVR) algorithms were used to construct the inversion model of cotton SPAD. This study analyzed the effects of different image segmentation methods on the extraction accuracy of spectral information and the accuracy of SPAD modeling in the cotton canopy. The results showed that (1) The accuracy of spectral information extraction can be improved by removing background interference such as soil and shadows using four image segmentation methods. The correlation between the vegetation indices calculated from MESMA segmented images and the SPAD of the cotton canopy was improved the most; (2) At three different flight altitudes, the vegetation indices calculated by the MESMA segmentation method were used as the input variable, and the SVR model had the best accuracy in the inversion of cotton SPAD, with R2 of 0.810, 0.778, and 0.697, respectively; (3) At a flight altitude of 80 m, the R2 of the SVR models constructed using vegetation indices calculated from images segmented by VIT, SVM, SMA, and MESMA methods were improved by 2.2%, 5.8%, 13.7%, and 17.9%, respectively, compared to the original images. Therefore, the MESMA mixed pixel decomposition method can effectively remove soil and shadows in multispectral images, especially to provide a reference for improving the inversion accuracy of crop physiological parameters in low-resolution images with more mixed pixels.

Water bodies are a fundamental element of urban ecosystems, and water mapping is critical for urban and landscape planning and management. Remote sensing has increasingly been used for water mapping in rural areas; however, when applied to urban areas, this spatially- explicit approach is a challenging task due to the fact that the water bodies are often of a small size and spectral confusion is common between water and the complex features in the urban environment. Water indexes are the most common method of water extraction at the pixel level. More recently, spectral mixture analysis (SMA) has been widely employed in analyzing the urban environment at the subpixel level. The objective of this study is to develop an automatic subpixel water mapping method (ASWM) which can achieve a high accuracy in urban areas. Specifically, we first apply a water index for the automatic extraction of mixed land-water pixels, and the pure water pixels that are generated in this process are exported as the final result. Secondly, the SMA technique is applied to the mixed land-water pixels for water abundance estimation. As for obtaining the most representative endmembers, we propose an adaptive iterative endmember selection method based on the spatial similarity of adjacent ground surfaces. One classical water index method (the modified normalized difference water index (MNDWI)), a pixel-level target detection method (constrained energy minimization (CEM)), and two widely used SMA methods (fully constrained least squares (FCLS) and multiple endmember spectral mixture analysis (MESMA)) were chosen for the water mapping comparison in the experiments. The results indicate that the proposed ASWM was able to detect water pixels more efficiency than other unsupervised water extraction methods, and the water fractions estimated by the proposed ASWM method correspond closely to the reference fractions with the slopes of 0.97, 1.02, 1.04, and 0.98 and the R-squared values of 0.9454, 0.9486, 0.9665, and 0.9607 in regression analysis corresponding to different test regions. In the quantitative accuracy assessment, the ASWM method shows the best performance in water mapping with the mean kappa coefficient of 0.862, mean producer’s accuracy of 82.8%, and mean user’s accuracy of 91.8% for test regions.

Rapid urbanization leads to rapid land use/cover change that can cause a severe deterioration of living environment in urban areas. Although numerous research projects have explored land use/cover change and the urbanization process with remote sensing data, only a few have focused on the analysis of their temporal and spatial variations in a visualized method using sub-pixel imaging techniques. In this study, Wuhan, the largest developing city in Central China, was chosen as the experimental region. Six scenes from Landsat images were selected, covering a lengthy period. To solve the problem of mixed pixels in urban areas, the spectrum of each image was selected individually and multiple endmember spectral mixture analysis (MESMA), a sub-pixel imaging soft classification technique, was applied. The land-use dynamic and time-series land-cover fractional maps were produced to evaluate the land use/cover change situation. Next, the land-cover change intensity was calculated and the dominant land-cover change intensity is proposed as a virtualized method of analyzing the urbanization process during each period covered by the Landsat images. The results of the study demonstrate that the land-cover categories and fractional information were effectively acquired following the implementation of MESMA. This information shows that the built-up areas in Wuhan have expanded at an average rate of 4.36 % rate in the last 20 years. The urbanization process has resulted in the transformation of the natural landscape to anthropogenic urban built-up areas. It is surprised that these transformations have not been smooth. In addition, the urbanization process exhibits different patterns during each period.

Endmember selection is the key to success in pixel unmixing which plays an important role in urban impervious surface abundance extraction. During the extraction, however, there has been a problem for the discrimination of impervious surfaces and soils because of their similarity in spectral. This increase the difficulty in distinguishes impervious surface and soil in endmember selection. To address this issue, in the current study, the biophysical composition index (BCI) and soil adjusted vegetation index (SAVI) were introduced to enhance the information of impervious surface and bare soil in the study area. Then, by selecting high albedo, low albedo, soil and vegetation endmembers with the utilization of the histogram of the indices and minimum noise fraction (MNF) scatter plot, we applied spectral mixture analysis (SMA) to extract impervious surface abundance. The scene of multispectral Landsat TM image was acquired allowing for the interpretation and analysis of impervious surfaces distribution. Experiments and comparisons indicate that this method performs well in estimating subpixel impervious surface distribution with relatively high precision and small bias.

A pre-screened and normalized multiple endmember spectral mixture analysis for mapping impervious surface area in Lake Kasumigaura Basin, Japan

Prior-knowledge-based spectral mixture analysis for impervious surface mapping

Quantifying the temporal and spatial patterns of impervious surfaces (IS) is important for assessing the environmental and ecological impacts of urbanization. In order to better extract IS, and to explore the divergence in urbanization in different regions, research on the regional differentiation features and regional change difference features of IS are required. To extract China’s 2013 urban impervious area, we used the 2013 night light (NTL) data and the Moderate Resolution Imaging Spectroradiometer Normalized Difference Vegetation Index and enhanced vegetation index (EVI) temporal series data, and used three urban impervious surface extraction indexes—Human Settlements Index, Vegetation-Adjusted NTL Urban Index, and the EVI-adjusted NTL index (EANTLI)—which are recognized as the best and most widely used indexes for extracting urban impervious areas. We used the classification results of the Landsat-8 images as the benchmark data to visually compare and verify the results of the urban impervious area extracted by the three indexes. We determined that the EANTLI index better reflects the distribution of the impervious area. Therefore, we used the EANTLI index to extract the urban impervious area from 2003 to 2013 in the study area, and researched the spatial and temporal differentiation in urban IS. The results showed that China’s urban IS area was 70,179.06 km2, accounting for 0.73% of the country’s land area in 2013, compared with 20,565.24 km2 in 2003, which accounted for 0.21% of the land area, representing an increase of 0.52%. On a spatial scale, like economic development, the distribution of urban impervious surfaces was different in different regions. The overall performance of the urban IS percentage was characterized by a decreasing trend from Northwest China, Southwest China, the Middle Reaches of the Yellow River, Northeast China, the Middle Reaches of the Yangtze River, Southern Coastal China, and Northern Coastal China to Eastern Coastal China. On the provincial scale, the urban IS expansion showed considerable differences in different regions. The overall performance of the Urban IS Expansion index showed that the eastern coastal areas had higher values than the western inland areas. The cities or provinces of Beijing, Tianjin, Jiangsu, and Shanghai had the largest growth in impervious areas. Spatially and temporally quantifying the change in urban impervious areas can help to better understand the intensity of urbanization in a region. Therefore, quantifying the change in urban impervious area has an important role in the study of regional environmental and economic development, policy formulation, and the rational use of resources in both time and space.

Land surface temperature (LST) plays an important role in characterizing surface energy fluxes as well as climatology. High-spatial-resolution LST values are needed to profoundly understand urban climatology through identifying its detailed thermal distribution patterns. In this paper, a simple method that uses hyperspectral spectral mixture analysis (SMA) to estimate fine resolution urban LST is presented. The proposed approach takes account of the universal existence of thermally mixed pixels in complex urban areas and includes mainly two steps. In the first step, multiple endmember SMA (MESMA) is applied to calculate land composition fractions. With the retrieved component abundances and associated thermal responses, high-resolution urban LST is estimated through a linear combination of each thermal response component multiplied by its respective abundance in the second step. In particular, hyperspectral data were used to enhance the spatial information extraction through improved accuracy in retrieving land cover abundance. Evaluation of the proposed method was performed at four spatial scales (4, 8, 16, and 32 m) over two scenes with different land cover characteristics. Results show a relatively modest agreement between the estimated LST and the reference LST by both visual interpretation and quantitative accuracy metrics, with a root-mean-square error (RMSE) of 4.18 and 1.71 K as well as the mean absolute error (MAE) of 3.62 and 0.89 K for the two scenes separately. Overall, this approach enhances the application of hyperspectral data for estimating high-spatial-resolution LST; therefore, it can serve as a promising technique in urban applications.

Multiple Endmember Spectral Mixture Analysis (MESMA) to map burn severity levels from Landsat images in Mediterranean countries