Mapping of debris-covered glaciers in the Garhwal Himalayas using ASTER DEMs and thermal data

Mapping of debris-covered glaciers using remote-sensing techniques is recognized as one of the greatest challenges for generating glacier inventories and automated glacier change analysis. The use of visible (VIS) and near-infrared (NIR) bands does not provide sufficient continual information to detect debris-covered ice with remote-sensing data. This article presents a semi-automated mapping method for the debris-covered glaciers of the Garhwal Himalayas based on an Advanced Spaceborne Thermal Emission and Reflection Radiometer (ASTER) digital elevation model (DEM) and thermal data. Morphometric parameters such as slope, plan curvature and profile curvature were computed by means of the ASTER DEM and organized in similar surface groups using cluster analysis. A thermal mask was generated from a single band of an ASTER thermal image, while the clean-ice glaciers were identified using a band ratio based on ASTER bands 3 and 4. Vector maps were drawn up from the output of the cluster analysis, the thermal mask and the band ratio mask for the preparation of the final outlines of the debris-covered glaciers using geographic information system (GIS) overlay operations. The semi-automated mapped debris-covered glacier outline of Gangotri Glacier derived from 2006 ASTER data varied by about 5% from the manually outlined debris-covered glacier area of the Cartosat-1 high-resolution image from the same year. By contrast, outlines derived from the method developed using the 2001 ASTER DEM and Landsat thermal data varied by only 0.5% from manually digitized outlines based on Indian Remote Sensing Satellite (IRS)-1C panchromatic (PAN) data. We found that post-depositional sedimentation by debris flow/mass movement was a great hindrance in the fully automated mapping of debris-covered glaciers in the polygenetic environment of the Himalayas. In addition, the resolution of ASTER stereo data and thermal band data limits the automated mapping of small debris-covered glaciers with adjacent end moraine. However, the results obtained for Gangotri Glacier confirm the strong potential of the approach presented.     

Changes in snow and glacier cover in an arid watershed of the western Kunlun Mountains using multisource remote-sensing data

Snow and glaciers in the mountain watersheds of the Tarim River basin in western China provide the primary water resources to cover the needs of downstream oases. Remote sensing provides a practical approach to monitoring the change in snow and glacier cover in those mountain watersheds. This study investigated the change in snow and glacier cover in one such mountain watershed using multisource remote-sensing data, including the Moderate Resolution Imaging Spectroradiometer (MODIS), Landsat (Multispectral Scanner (MSS), Thematic Mapper (TM), and Enhanced Thematic Mapper Plus (ETM+)), Corona, and Google Earth^TM imagery. With 10 years’ daily MODIS snow-cover data from 2002 to 2012, we used two de-cloud methods before calculating daily snow-cover percentage (SCP), annual snow-cover frequency (SCF), and annual minimum snow-cover percentage (AMSCP) for the watershed. Mann–Kendall analysis showed no significant trend in any of those snow-cover characterizations. With a total of 22 Landsat images from 1967 to 2011, we used band ratio and supervised classification methods for snow classification for Landsat TM/ETM+ images and MSS images, respectively. The Landsat snow-cover data were divided into two periods (1976–2002 and 2004–2011). Statistical tests indicated no significant difference in either the variance or mean of SCPs between the two periods. Three glaciers were identified from Landsat images of 1998 and 2011, and their total area increased by 12.6%. In addition, three rock glaciers were also identified on both the Corona image of 1968 and the Google high-resolution image of 2007, and their area increased by 2.5%. Overall, based on multisource remote-sensing data sets, our study found no evidence of significant changes in snow and glacier cover in the watershed.     

Synergistic approach for mapping debris-covered glaciers using optical-thermal remote sensing data with inputs from geomorphometric parameters

Debris cover on glacier boundaries critically impedes the global inventorying of glaciers and confounds most of the techniques developed for semi-automated mapping of glaciers. Debris on the glacier (referred as supraglacial debris) and that occurring outside the glacier boundaries (referred as periglacial debris) being derived from a common source, i.e. the valley rock, tend to have a similar spectral response in the reflection region which renders them mutually indistinguishable. However, there exist temperature differences between them. This aspect has been considered in this remote sensing based study to distinguish between the supraglacial and periglacial debris in a test area in the Chenab basin, Himalayas, by inclusion of thermal infrared (TIR) bands in remote sensing data processing. A synergistic multisensor approach for the delineation of debris-covered glacier boundaries is used here which integrates the inputs from thermal (TERRA-ASTER sensor) and optical (IRS-P6-AWiFS sensor) remote sensing data, multispectral classification techniques and the DEM derived geomorphometric parameters. The results of this study corroborate earlier findings on utilization of temperature differences as one of the parameters in glacial studies. The proposed synergistic approach therefore appears useful in accurate mapping of debris-covered glaciers in the Himalayan region.     

Land‐cover classification in the Brazilian Amazon with the integration of Landsat ETM+ and Radarsat data

Land‐cover classification with remotely sensed data in moist tropical regions is a challenge due to the complex biophysical conditions. This paper explores techniques to improve land‐cover classification accuracy through a comparative analysis of different combinations of spectral signatures and textures from Landsat Enhanced Thematic Mapper Plus (ETM+) and Radarsat data. A wavelet‐merging technique was used to integrate Landsat ETM+ multispectral and panchromatic data or Radarsat data. Grey‐level co‐occurrence matrix (GLCM) textures based on Landsat ETM+ panchromatic or Radarsat data and different sizes of moving windows were examined. A maximum‐likelihood classifier was used to implement image classification for different combinations. This research indicates the important role of textures in improving land‐cover classification accuracies in Amazonian environments. The incorporation of data fusion and textures increases classification accuracy by approximately 5.8–6.9% compared to Landsat ETM+ data, but data fusion of Landsat ETM+ multispectral and panchromatic data or Radarsat data cannot effectively improve land‐cover classification accuracies.     

Landsat-based inventory of glaciers in western Canada, 1985-2005

We report on a glacier inventory for the Canadian Cordillera south of 60°N, across the two western provinces of British Columbia and Alberta, containing ~ 30,000 km² of glacierized terrain. Our semi-automated method extracted glacier extents from Landsat Thematic Mapper (TM) scenes for 2005 and 2000 using a band ratio (TM3/TM5). We compared these extents with glacier cover for the mid-1980s from high-altitude, aerial photography for British Columbia and from Landsat TM imagery for Alberta. A 25 m digital elevation model (DEM) helped to identify debris-covered ice and to split the glaciers into their respective drainage basins. The estimated mapping errors are 3-4% and arise primarily from seasonal snow cover. Glaciers in British Columbia and Alberta respectively lost − 10.8 ± 3.8% and − 25.4% ± 4.1% of their area over the period 1985-2005. The region-wide annual shrinkage rate of − 0.55% a^− 1 is comparable to rates reported for other mountain ranges in the late twentieth century. Least glacierized mountain ranges with smaller glaciers lost the largest fraction of ice cover: the highest relative ice loss in British Columbia (− 24.0 ± 4.6%) occurred in the northern Interior Ranges, while glaciers in the northern Coast Mountains declined least (− 7.7 ± 3.4%).     

Remote sensing of glaciers in the tropical Andes: a review

While the tropical Andes of Venezuela, Colombia, Ecuador, Peru, Bolivia, northern Chile, and northern Argentina represent more than 95% of all tropical glaciers globally, they are relatively sparsely studied in general, and particularly by using remote-sensing approaches. However, studies from the 1930s used terrestrial imaging, and aerial photographs are available from the 1950s on. In this article, we review the literature on remote sensing of the glaciers in the tropical Andes, divided into the four climatic sub-regions: inner tropics, dry outer tropics, northern wet outer tropics, and southern wet outer tropics. The majority of studies used optical remote sensing, particularly Landsat and Advanced Spaceborne Thermal Emission and Reflection Radiometer (ASTER) imagery, although shadows produced by steep terrain and often dense cloud cover – within the Intertropical Convergence Zone (ITCZ) of the inner tropics – are challenges to data analyses. Microwave and lidar remote sensing have been successfully employed in some studies. The vast majority of glacier monitoring studies documented glacier recessions throughout the tropical Andes since the 1950s; most results are available for the Peruvian Andes, particularly the Cordillera Balance. Recent field tests explored the use of unmanned aerial systems (UAS) and lidar in glacier research; preliminary results are promising and have the potential to lead into new research directions.     

Remote-sensing-based study of impact of a rock avalanche on North Terong Glacier in Karakorum Himalaya

ABSTRACT

In this article, we report the results of a study on occurrence of a mass-movement event classified as ‘rock avalanche’ over the North Terong glacier (a tributary of the Siachen Glacier in the Nubra Valley), using multi-date remote-sensing data of Landsat Thematic Mapper (TM), Enhanced Thematic Mapper Plus (ETM+), and Operational Land Imager (OLI) and its effects on glacier-surface velocity. Normalized cross-correlation image matching method for displacement measurements at sub-pixel level using Computer Imaging Analysis Software (CIAS), an add-on module of Environment for Visualizing Images (ENVI) software, was used to map changes in the position of debris-deposits of the rock avalanche and assess variations in glacier-surface velocities. Well-defined debris margins gave positions of debris on different dates. Study indicated that this rock avalanche must have occurred between 21 April 2000 and 8 May 2000 (possibly on 3 or 4 May). It induced a kinematic wave or a ‘mini-surge’ in the glacier with its surface velocity showing almost twofold increase in post-event period.     

Differential response of glaciers with varying debris cover extent: evidence from changing glacier parameters

Supraglacial debris (SGD) cover on mountain glaciers is known to greatly influence various glacier processes and alter their response to climate change. In this study, vital glacier parameters of five glaciers with varying debris coverage (about 7–26%), located in Zanskar basin, Ladakh Himalaya, were monitored using Landsat imagery (from 1977 to 2013) and the Advanced Spaceborne Thermal Emission and Reflection Radiometer (ASTER) global digital elevation model (GDEM). Results reveal how varying amounts of SGD induce categorically distinct responses on glaciers, sharing same geographical settings. The clean glaciers (CG) in the area were found to have higher retreat rates (about 8–19 m year^?1), comparable areal shrinkage (about 13–15%), higher accumulation area ratios (AARs) (>0.5), rapid increase in SGD (about 1.6–3.0-fold), and association with increasing numbers of peri- and proglacial lakes (2–8 per glacier). The debris-covered glaciers (DCG) showed minimal frontal changes (about 2–5 m year^?1), higher areal shrinkage (about 14–21%), low AARs (<0.5), slow SGD changes (about ≤1.2-fold), and association with increasing numbers of peri- and supraglacial lakes (SGLs) (2–39 per glacier). Moreover, while changes in SGD had a strong negative correlation with changes in glacier area, retreat rates, and AAR, they were positively related with increase in area of SGLs.     

Monitoring Variation of Glaciers based on Remote Sensing Images in the Chenab Basin，Western-Himalaya，1993~2016

Based on Landsat TM/ETM+/OLI remote sensing images，the glacier boundaries in the Chenab basin of western Himalayas in three periods were manually delineated with visual interpretation method, and the characteristics of glacier variation were also analyzed with the temperature and precipitation of the surrounding meteorological stations and CRU reanalysis data．The results show that: ①From 1993 to 2016,the glaciers area in Chenab basin decreased 164.56±161.72 km²，accounting for 5.78% of the total area. The annul average shrinkage rate is 0.25±0.25 %·a^-1 and it accelerated shrinking after 2000. ②The glaciers in the Chenab basin have shrinked in all orientations and altitudes. Among them, S orientation glaciers has the maximum shrinkage rate, accounting for 24.35% of the total area of glacial shrinkage. The glaciers areas between 4 600~4 800 m and 4 800~5 000 m is reduced 29.93 km² and 30.91 km² near 23 a,accounting for 17.72% and 18.30% of the total shrinkage of the glacier area in the basin respectively. ③From 1993 to 2016, there were 28 different glaciers had advanced in the Chenab basin. ④Analysis of temperature and precipitation changes in the two meteorological stations of Shiquan river and Srinagar and CRU reanalysis data shows that the average annual temperature in the region increased significantly from 1993 to 2016 caused glacier retreat.     

1993~2016年喜马拉雅山西段杰纳布流域冰川变化遥感监测

基于Landsat TM/ETM+及OLI遥感影像,对喜马拉雅山西段杰纳布流域冰川面积进行提取,对冰川时空分布特征及其变化分析,并结合周边气象台站及CRU再分析资料气温、降水量资料对研究区冰川变化原因进行讨论。结果表明：①1993~2016年杰纳布流域冰川面积萎缩了164.56±161.72 km²,占总面积的5.78%,年均萎缩率为0.25±0.25 %·a^-1,且在2000年后加快萎缩;②杰纳布流域冰川在各个朝向和海拔带上均呈萎缩趋势,其中S朝向冰川面积萎缩率最大,占研究区冰川萎缩总面积的24.35%; 4 600~4 800 m和4 800~5 000 m两个海拔高度带冰川面积近23 a分别减少了29.93 km²和30.91 km²,占流域冰川面积萎缩总量的17.72%和18.30%;③1993~2016年杰纳布流域共有28条冰川末端发生不同程度的前进现象;④对狮泉河和Srinagar气象站及CRU再分析资料气温、降水量变化分析表明,1993~2016年该区域年均气温呈显著上升是杰纳布流域冰川萎缩的主要原因。     

Combining satellite multispectral image data and a digital elevation model for mapping debris-covered glaciers

Automated glacier mapping from satellite multispectral image data is hampered by debris cover on glacier surfaces. Supraglacial debris exhibits the same spectral properties as lateral and terminal moraines, fluvioglacial deposits, and bedrock outside the glacier margin, and is thus not detectable by means of multispectral classification alone. Based on the observation of low slope angles for debris-covered glacier tongues, we developed a multisource method for mapping supraglacial debris. The method combines the advantages of automated multispectral classification for clean glacier ice and vegetation with slope information derived from a digital elevation model (DEM). Neighbourhood analysis and change detection is applied for further improvement of the resulting glacier/debris map. A significant percentage of the processing can be done automatically. In order to test the sensitivity of our method against different DEM qualities, it was also applied to a DEM obtained from ASTER stereo data. Additionally, we compared our multisource approach to an artificial neural network (ANN) classification of debris, using only multispectral data. While the combination with an ASTER-derived DEM revealed promising results, the ANN classification without DEM data does not.     

基于Landsat数据的郭扎错北面冰川近20年来面积动态变化遥感研究

冰川能够敏感地反映区域环境变化,是研究全球变化的重要因素之一。昆仑山地区冰川集中,是研究冰川动态变化的理想区域。根据郭扎错北面1991~2009年Landsat TM与ETM+遥感影像,研究了该地区冰川近20 a来的变化情况。结果发现,该地区冰川变化显著,并得出以下结论：① 郭扎错北面冰川面积在1991~2009年间具有先增加后减少的波动规律;② 该区域内存在东部冰川比西部变化量大和变化率快的差异性;③ 该地区中峰冰川在2001~2004年间面积大幅增加,可能与2001年11月14日发生在昆仑山口以西的8.1级强烈地震有关;④ 该研究区内冰川面积变化主要受年均温度和年累积降水量的综合影响。     

Narrowband to broadband conversion of Landsat TM glacier albedos

In this paper we present an empirical relationship between the broadband glacier albedo (alpha) and the narrowband glacier albedos in Landsat TM bands 2 and 4 (alpha2 and alpha4, respectively). The relationship was established on the basis of multiple linear regression analysis of 112 ground-based simultaneous measurements of alpha, alpha2 and alpha4 made at 32 sites on the tongue of the Morteratschgletscher, Switzerland. The measurements were carried out over a representative set of glacier surface types ranging from completely debris-covered glacier ice (alpha=0.08)to dry snow (alpha=0.86). The regression model explains more than 99% of the variance of the broadband albedo and the root-mean-square value of the residuals is only 0.009. The relationship enables users of Landsat TM data to make an accurate estimate of the broadband albedo on the basis of narrowband albedos without having to classify the glacier surface.     

基于高分遥感数据和深度学习的石冰川自动提取研究

石冰川是以冰岩混合物为基础形成的一类具有舌状堆积纹理的冰缘地貌,了解其分布和变化对于寒区环境研究具有重要价值,遥感技术的发展为石冰川的识别提供了有效的手段。针对石冰川发育地的偏远和调查的困难,以及其光谱特征的微弱性,提出了一种基于深度学习的石冰川识别方法,以ResNet作为训练网络,得到石冰川的图像分类模型,以国产高分一号遥感影像作为实验数据,在念青唐古拉山西段展开了应用,共识别出石冰川96条。验证结果表明：该方法具有较高的识别精度（98.72%的总体精度、89.48%的生产精度和81.77 %的用户精度）,证明该方法能够有效地识别石冰川,并为在大区域开展石冰川的调查和分析提供了基础。     

A multi-sensor approach for improving biodiversity estimation in the Hyrcanian mountain forest,Iran

ABSTRACT

Modelling tree biodiversity in mountainous forests using remote-sensing data is challenging because forest composition and structure change along elevation. Topographic variations also affect vegetation’s spectral and backscattering behaviour. We demonstrate the potential of multi-source integration to tackle this challenge in a mountainous part of the Hyrcanian forest in Iran. This forest is a remnant of a deciduous broadleaved forest with heterogeneous structure affected by natural and anthropogenic factors. The multi-source approach (i.e. Landsat Enhanced Thematic Mapper Plus (ETM +), Advanced Land Observing Satellite/ Phased Array type L-band Synthetic Aperture Radar (ALOS/PALSAR), and topographic variables) allows us to propose a biodiversity estimation model using partial least square regression (PLSR) calibrated and validated with limited field data. The effective number of species was calculated based on field measurements of the biodiversity in the study area. In order to model species diversity in more homogeneous extrinsic environmental conditions, we divided data into two groups with relatively uniform slope values. In each slope group, we modelled the correlation between observed biodiversity and satellite-derived data. For that, we followed three scenarios: (A) multispectral Landsat ETM + alone, (B) ALOS/PALSAR alone, and (C) inclusion of both sensors. In each scenario, elevation and slope data were also considered as predictors. We observed that in all scenarios, coefficient of determination (R²) in gentler slopes was higher than that in areas with steeper slopes (average difference in R²: ?R² = 0.21). The highest correlation was achieved by inclusion of synthetic aperture radar (SAR) and ETM + (R² = 0.87). The results clearly confirm that the multi-source remote-sensing approach can provide a practical estimate of biodiversity across the Hyrcanian forest and potentially in other deciduous broadleaved forests in complex terrain.     

Remote-sensing assessment of glacier fluctuations in the Hindu Raj,Pakistan

Space-based assessments of glaciers across the Himalayas indicate that there is a spatial variation in glacier fluctuations due to variations in local topography, regional climate, and ice-flow dynamics. Unfortunately, limited information is available on glacier fluctuations in northern Pakistan. In this work, we quantify the glacier terminus variations in the Hindu Raj region of Pakistan, where we used Landsat and Advanced Spaceborne Thermal Emission and Reflection Radiometer (ASTER) time-series data for 1972, 1989, 1999, and 2007. Eighty-five mountain glaciers of various sizes, orientations, and altitudes were sampled. Our results show that most of the glaciers (70.6%) retreated over the last four decades, although some glaciers advanced (17.6%) or exhibited no detectable change in terminus position (11.8%). Larger glaciers with lower terminus altitudes exhibited greater retreat distances than smaller high-altitude glaciers. Long-term climate data analysis reveals that the recession of glaciers appears to be associated with the rising of summer temperatures in the Hindu Raj. Our results support a spatial trend of an increase in shrinking glaciers towards the western portion of northern Pakistan, with a greater frequency of advancing glaciers towards the east.     

Digital terrain analysis using Landsat‐7 ETM+ imagery and SRTM DEM: a case study of Nevsehir province (Cappadocia), Turkey

A three‐dimensional (3D) model of land‐use/land‐cover (LULC) and a digital terrain model of Nevsehir province (Cappadocia), Turkey, were generated and analysed using a Landsat‐7 Enhanced Thematic Mapper Plus (ETM+) multispectral image set and a Shuttle Radar Topographic Mission (SRTM) digital elevation model (DEM). Stream drainage patterns, lineaments and structural‐geological features (landforms) were extracted and analysed. In the process of analysing and interpreting the multispectral images of geological features, criteria such as colour and colour tones, topography and stream drainage patterns were used to acquire information about the geological structures of the land, including as geomorphological, topographic and tectonic structures. Landsat‐7 ETM+ multispectral imagery and an SRTM DEM of the study region were used experimentally for classification and analysis of a digital terrain model. Using the multispectral image data, the LULC types were classified as: settlement (1.2%); agricultural land (70.1%); forest (scrubland, orchard and grassland) (2.9%); bare ground (25.5%); and water bodies (lakes and rivers) (0.3%) of the study area (5434 km²). The results of the DEM classification in the study area were: river flood plain (11.3%); plateau (52.3%); high plateau (28.4%); mountain (7.6%); and high mountain (0.3%). Lineament analysis revealed that the central Kizilirmak River divides the region into two nearly equal parts: the Kirsehir Plateau in the north and the Nevsehir Plateau in the south. In terms of the danger of catastrophe, the settlements of Kozakli, Hacibektas and Acigol were found to be at less risk of earthquake and/or flooding than those of Avanos, Gulsehir, Urgup, Nevsehir, Gumuskent and Derinkuyu, which are located on river flood plains and/or the main stream drainage channels, particularly stream beds, where the lineaments are deep valleys or fracture or fault‐line indicators.     

A refined classification approach by integrating Landsat Operational Land Imager (OLI) and RADARSAT-2 imagery for land-use and land-cover mapping in a tropical area

Producing accurate land-use and land-cover (LULC) mapping is a long-standing challenge using solely optical remote-sensing data, especially in tropical regions due to the presence of clouds. To supplement this, RADARSAT images can be useful in assisting LULC mapping. The fusion of optical and active remote-sensing data is important for accurate LULC mapping because the data from different parts of the spectrum provide complementary information and often lead to increased classification accuracy. Also, the timeliness of using synthetic aperture radar (SAR) fills information gaps during overcast or hazy periods. Therefore, this research designed a refined classification procedure for LULC mapping for tropical regions. Determining the best method for mapping with a specific data source and study area is a major challenge because of the wide range of classification algorithms and methodologies available. In this study, different combinations and the potential of Landsat Operational Land Imager (OLI) and RADARSAT-2 SAR data were evaluated to select the best procedure for LULC classification. Results showed that the best filter for SAR speckle reduction is the 5 × 5 enhanced Lee. Furthermore, image-sharpening algorithms were employed to fuse Landsat multispectral and panchromatic bands and subsequently these algorithms were analysed in detail. The findings also confirmed that Gram–Schmidt (GS) performed better than the other techniques employed. Fused Landsat data and SAR images were then integrated to produce the LULC map. Different classification algorithms were adopted to classify the integrated Landsat and SAR data, and the maximum likelihood classifier (MLC) was considered the best approach. Finally, a suitable classification procedure was designed and proposed for LULC as mapping in tropical regions based on the results obtained. An overall accuracy of 98.62% was achieved from the proposed methodology. The proposed methodology is a useful tool in industry for mapping purposes. Additionally, it is also useful for researchers, who could extend the method for different data sources and regions.     

Comparing the spectral signal of burned surfaces between Landsat 7 ETM+ and Landsat 8 OLI sensors

There is a long history of the use of Landsat data in burned land mapping mainly due to certain characteristics of the Landsat imagery including the spatial, spectral, and temporal data resolution, the low cost (Landsat data are now freely available), and the existence of an almost 35-year historical archive (excluding Landsat 1–3). Landsat 8 (Operational Land Imager (OLI) and Thermal Infrared Sensor (TIRS)) was launched on 11 February 2013 and it captures data in three new bands along with two additional thermal bands. However, is the spectral signal of burned surfaces in satellite remote-sensing data of Landsat series consistent and robust enough to allow the successful application of the techniques developed so far for Landsat 8? In this article, we compare the spectral signal of burned surfaces between Landsat 7 Enhanced Thematic Mapper Plus (ETM+) and Landsat 8 OLI sensors using five case studies that correspond to five large fire events in different biophysical environments in Greece, for which both Landsat 7 ETM+ and Landsat 8 OLI data were available. From the comparative analysis using histogram data plots of burned (post-fire image) and vegetated (pre-fire image) areas, spectral signature plots and separability indices of certain land-cover types, estimated using the same sampling areas over both satellite images, a general consistency was observed between the two sensors. Slight differences between the sensors were attributed to differences in the acquisition dates and were related to the type of vegetation rather than the sensors used to record the satellite images. Neither sensor provided improved discrimination over the other.     

Land-cover classification of partly missing data using support vector machines

Land-cover classification based on multi-temporal satellite images for scenarios where parts of the data are missing due to, for example, clouds, snow or sensor failure has received little attention in the remote-sensing literature. The goal of this article is to introduce support vector machine (SVM) methods capable of handling missing data in land-cover classification. The novelty of this article consists of combining the powerful SVM regularization framework with a recent statistical theory of missing data, resulting in a new method where an SVM is trained for each missing data pattern, and a given incomplete test vector is classified by selecting the corresponding SVM model. The SVM classifiers are evaluated on Landsat Enhanced Thematic Mapper Plus (ETM?+?) images covering a scene of Norwegian mountain vegetation. The results show that the proposed SVM-based classifier improves the classification accuracy by 5–10% compared with single image classification. The proposed SVM classifier also outperforms recent non-parametric k-nearest neighbours (k-NN) and Parzen window density-based classifiers for incomplete data by about 3%. Moreover, since the resulting SVM classifier may easily be implemented using existing SVM libraries, we consider the new method to be an attractive choice for classification of incomplete data in remote sensing.