SVM‐based segmentation and classification of remotely sensed data

Support Vector Machines (SVM) is becoming a popular alternative to traditional image classification methods because it makes possible accurate classification from small training samples. Nevertheless, concerns regarding SVM parameterization and computational effort have arisen. This Letter is an evaluation of an automated SVM‐based method for image classification. The method is applied to a land‐cover classification experiment using a hyperspectral dataset. The results suggest that SVM can be parameterized to obtain accurate results while being computationally efficient. However, automation of parameter tuning does not solve all SVM problems. Interestingly, the method produces fuzzy image‐regions whose contextual properties may be potentially useful for improving the image classification process.     

Inferences of all‐sky solar irradiance using Terra and Aqua MODIS satellite data

Solar irradiance is a key environmental control, and accurate spatial and temporal solar irradiance data are important for a wide range of applications related to energy and carbon cycling, weather prediction, and climate change. This study presents a satellite‐based scheme for the retrieval of all‐sky solar irradiance components, which links a physically based clear‐sky model with a neural network version of a rigorous radiative transfer model. The scheme exploits the improved cloud characterization and retrieval capabilities of the MODerate resolution Imaging Spectroradiometer (MODIS) onboard the Terra and Aqua satellites, and employs a cloud motion tracking scheme for the production of hourly solar irradiance data throughout the day. The scheme was implemented for the Island of Zealand, Denmark (56° N, 12° E) and Southern Arizona, USA (31° N, 110° W) permitting model evaluation for two highly contrasting climates and cloud environments. Information on the atmospheric state was provided by MODIS data products and verifications against AErosol RObotic NETwork (AERONET) data demonstrated usefulness of MODIS aerosol optical depth and total precipitable water vapour retrievals for the delineation of spatial gradients. However, aerosol retrievals were significantly biased for the semi‐arid region, and water‐vapour retrievals were characterized by systematic deviations from the measurements. Hourly global solar irradiance data were retrieved with overall root mean square deviations of 11.5% (60 W m^?2) and 26.6% (72 W m^?2) for Southern Arizona and the Island of Zealand, respectively. For both regions, hourly satellite estimates were shown to be more reliable than pyranometer measurements from ground stations only 15 km away from the point of interest, which is comparable to the accuracy level obtainable from geostationary satellites with image acquisitions every 15–30 min. The proposed scheme is particularly useful for solar irradiance mapping in high‐latitude regions as data from geostationary satellites experience a gradual degradation in spatial resolution and overall quality with latitude and become unusable above approximately 60° latitude. However, in principle, the scheme can be applied anywhere on the globe, and a synergistic use of MODIS and geostationary satellite datasets may be envisaged for some applications.     

Analysis of urban growth using multi‐temporal satellite data in Istanbul,Turkey

Uncontrolled population growth, especially in developing countries, causes serious problems, such as scarcity of food, informal settlements, environmental pollution, destruction of ecological structure, unemployment, etc. This phenomenon will require advanced methodologies, such as space technologies, to enable city planners, economists, environmentalists, ecologists and resource managers to solve these problems. In Turkey, as a result of the undesired population growth, new settlements are continuously appearing and adverse developments and changes are occurring in the presently populated areas. In Turkey's major cities, such as in Istanbul, Izmir, Ankara, Adana, Bursa and Antalya, which have seen mass migrations of people, considerable urban developments and changes have occurred. Consequently, the mostly negative impacts of uncontrolled population growth on the urban environment must be monitored continuously.

The objective of this study was to investigate the impact of urban growth on land‐use changes, especially the agricultural land in the district of Büyükçekmece in suburban Istanbul. The study is based on the 1984–1997 population database, multi‐temporal satellite data and remote sensing methods. The study revealed significant loss of agricultural and natural land areas to urban developments throughout the period 1984–1998.     

Mapping insect‐induced tree defoliation and mortality using coarse spatial resolution satellite imagery

Insect‐induced defoliation causes significant timber and carbon losses in many forested countries. The purpose of this investigation was to examine the potential use of coarse spatial resolution satellite imagery for mapping tree defoliation and mortality caused by a large insect infestation. We examined 1?km multi‐temporal SPOT Vegetation (VGT) data over a coniferous forest region in Quebec, Canada that was severely defoliated during 1998–2000 by the eastern hemlock looper. A logistic regression model based on satellite change metrics was developed to map defoliation and mortality. The results suggest that coarse imagery is effective for mapping large‐scale conifer forest mortality caused by insects, and could also be useful for near real‐time monitoring of severe defoliation, although with 2–3 times greater errors of commission.     

PCA‐based land‐use change detection and analysis using multitemporal and multisensor satellite data

Remote‐sensing change detection based on multitemporal, multispectral, and multisensor imagery has been developed over several decades and provided timely and comprehensive information for planning and decision‐making. In practice, however, it is still difficult to select a suitable change‐detection method, especially in urban areas, because of the impacts of complex factors. This paper presents a new method using multitemporal and multisensor data (SPOT‐5 and Landsat data) to detect land‐use changes in an urban environment based on principal‐component analysis (PCA) and hybrid classification methods. After geometric correction and radiometric normalization, PCA was used to enhance the change information from stacked multisensor data. Then, a hybrid classifier combining unsupervised and supervised classification was performed to identify and quantify land‐use changes. Finally, stratified random and user‐defined plots sampling methods were synthetically used to obtain total 966 reference points for accuracy assessment. Although errors and confusion exist, this method shows satisfying results with an overall accuracy to be 89.54% and 0.88 for the kappa coefficient. When compared with the post‐classification method, PCA‐based change detection also showed a better accuracy in terms of overall, producer's, and user's accuracy and kappa index. The results suggested that significant land‐use changes have occurred in Hangzhou City from 2000 to 2003, which may be related to rapid economy development and urban expansion. It is further indicated that most changes occurred in cropland areas due to urban encroachment.     

Traceability of slash‐and‐burn land‐use history using optical satellite sensor imagery: a basis for chronosequential assessment of ecosystem carbon stock in Laos

This study examined the use of satellite sensor imagery for chronosequential assessment of land use and ecosystem carbon stock in slash‐and‐burn (S/B) regions of Laos. The segmentation approach was useful because the boundaries of S/B patches are subject to change due to natural or anthropogenic factors. Polygon‐based classification using six optical bands of Landsat Enhanced Thematic Mapper Plus (ETM+) imagery showed that S/B patches could be discriminated with high accuracy (0.98). Normalized difference spectral indices, NDSI[i, j] = [R_j ?R_i ]/[R_j +R_i ], using reflectances R_j and R_i at j and i nm wavelengths for S/B polygons during four consecutive years (1999–2002) showed that NDSI[2215, 830], NDSI[1650, 830] and NDSI[660, 830] ( = the normalized difference vegetation index, NDVI) values decreased significantly in S/B years compared to those under fallow conditions (by 0.21±0.04, 0.20±0.04 and 0.17±0.03, respectively). Only slight differences were found before and after the S/B year, regardless of fallow length or biomass estimated by the allometry method. Relating reflectance signatures directly to fallow biomass was unsuitable, but these NDSIs were also useful for distinguishing S/B patches. Land‐use history, including the community age of fallow vegetation, can be traced on a pixel basis using a superimposed set of segmented classified images.     

Optimization in multi‐scale segmentation of high‐resolution satellite images for artificial feature recognition

Multi‐resolution segmentation, as one of the most popular approaches in object‐oriented image segmentation, has been greatly enabled by the advent of the commercial software, eCognition. However, the application of multi‐resolution segmentation still poses problems, especially in its operational aspects. This paper addresses the issue of optimization of the algorithm‐associated parameters in multi‐resolution segmentation. A framework starting with the definition of meaningful objects is proposed to find optimal segmentations for a given feature type. The proposed framework was tested to segment three exemplary artificial feature types (sports fields, roads, and residential buildings) in IKONOS multi‐spectral images, based on a sampling scheme of all the parameters required by the algorithm. Results show that the feature‐type‐oriented segmentation evaluation provides an insight to the decision‐making process in choosing appropriate parameters towards a high‐quality segmentation. By adopting these feature‐type‐based optimal parameters, multi‐resolution segmentation is able to produce objects of desired form to represent artificial features.     

Classification of multi‐spectral remote sensing data using a local transfer function classifier

This paper introduces a new neural network, called the local transfer function classifier (LTF‐C), for classification of multi‐spectral remote sensing data. The network structure of LTF‐C is similar to that of the radial basis function neural network (RBF), but LTF‐C utilizes an entirely different learning algorithm. In particular, the network structure of LTF‐C is not predetermined, but changes dynamically during the learning. Such a learning algorithm fits well to the classification problem, and guarantees that the size of the network is as large as is needed. The classification results show that LTF‐C evidently has a better classification accuracy than the six other classifiers in the experiment.     

DEM and bathymetry estimation for mapping a tide‐coordinated shoreline from fine spatial resolution satellite sensor imagery

Fine spatial resolution remotely sensed imagery has considerable potential for mapping a shoreline. Although fine spatial resolution imagery typically allows the instantaneous shoreline to be mapped with high accuracy, interest is normally focused on a reference shoreline, defined on a stable vertical datum, which is generally not apparent in the imagery unless acquired at a time carefully coordinated with the tidal characteristics of the region. To map a tide‐coordinated shoreline, such as the mean sea level (MSL), information on terrain topography, bathymetry and tidal characteristics is required. In this study, IKONOS imagery was used to derive topographic and bathymetric information for an extract of the Malaysian coast and combined with a tide chart for the region to map the MSL. The digital elevation model (DEM) derived had a root mean square error (RMSE), calculated on independent control points, of ～2.2 m while the bathymetric model had an RMSE of 0.87 m. The shoreline derived from the combination of the DEM, bathymetry and tidal information was mapped with an RMSE of 1.8 m.     

Spatial characterization of electrical power consumption patterns over India using temporal DMSP‐OLS night‐time satellite data

Changes in electric power consumption patterns of a country over a period of time reflect on its socio‐economic development and energy utilization processes. In the present study, we characterized spatial and temporal changes in electric power consumption patterns over India during 1993 to 2002, using ‘night‐time lights’ data given by the Defense Meteorological Satellite Program–Operational Line Scan System (DMSP‐OLS) over the Indian region. The OLS operates in two bands: visible (0.5–0.9 µm) and thermal (10.5–12.5 µm) and has a unique capability of picking up faint sources of visible–near infrared emissions (lights) at night on the Earth's surface including cities, towns and villages with a DN value ranging from 1 to 63. Night‐time light images for cloud‐free dates given by the DMSP‐OLS from 1993 to 2002 were segregated into respective years and were integrated to generate one ‘Stable light image’ per year. Changes in light scenarios over the Indian region in the decadal time frame were studied using stable lights datasets from 1993 to 2002. Information on changes in the light scenarios was integrated with demographic data to characterize developments in major cities and states of India. Results of the study suggested an increase in population by 170 million and power consumption from 44962 million kWh to 306355 million kWh over the country during 1993–2002, which was associated with an overall increase in number of night‐time lights of up to 26% in all states, indicating development in electric power consumption patterns. Correlation analysis between increase in population to the increase in night‐time lights and electric power consumption showed a coefficient of determination, R ², of 0.59 and 0.56 respectively. Increase in light intensities along the peripheries of major Indian cities was observed, which indicated increased stress on the cities and corresponding development in power consumption patterns during the decadal time frame. Certain states, however, showed a decrease in night‐time lights in some areas, which are primarily attributed to the decreased economic growth trend and poverty and accounted to the scatter observed in the correlation analysis. Results are discussed in the paper.     

Temporal and spatial soil moisture change pattern detection in an agricultural area using multi‐temporal Radarsat ScanSAR data

Monitoring the characteristics of spatially and temporally distributed soil moisture is important to the study of hydrology and climatology for understanding and calculating the surface water balance. The major difficulties in retrieving soil moisture with Synthetic Aperture Radar (SAR) measurements are due to the effects of surface roughness and vegetation cover. In this study we demonstrate a technique to estimate the relative soil moisture change by using multi‐temporal C band HH polarized Radarsat ScanSAR data. This technique includes two components. The first is to minimize the effects of surface roughness by using two microwave radar measurements with different incidence angles for estimation of the relative soil moisture change defined as the ratio between two soil volumetric moistures. This was done by the development of a semi‐empirical backscattering model using a database that simulated the Advanced Integral Equation Model for a wide range of soil moisture and surface roughness conditions to characterize the surface roughness effects at different incidence angles. The second is to reduce the effects of vegetation cover on radar measurements by using a semi‐empirical vegetation model and the measurements obtained from the optical sensors (Landsat TM and AVHRR). The vegetation correction was performed based on a first‐order semi‐empirical backscattering vegetation model with the vegetation water content information obtained from the optical sensors as the input. For the validation of this newly developed technique, we compared experimental data obtained from the Southern Great Plain Soil Moisture Experiment in 1997 (SGP97) with our estimations. Comparison with the ground soil moisture measurements showed a good agreement for predication of the relative soil moisture change, in terms of ratio, with a Root Mean Square Error (RMSE) of 1.14. The spatially distributed maps of the relative soil moisture change derived from Radarsat data were also compared with those derived from the airborne passive microwave radiometer ESTAR. The maps of the spatial characteristics of the relative soil moisture change showed comparable results.     

Detecting infestation of take‐all disease in wheat using Landsat Thematic Mapper imagery

Take‐all is an important root disease of wheat and has caused considerable decreases in crop yield in China. Although take‐all is the most studied disease of any crop, there is little research on detecting the infestation of the disease using remote sensing imagery. In this Letter, the Normalized Difference Vegetation Index and Masking, Principal‐Components Transformation, and Analysis techniques were used to detect the severe infestation of the take‐all disease in wheat using Landsat Thematic Mapper imageries on 30 April and 16 May 2006. Results showed that the technique was simple and effective in the study areas, and could be applied to other wheat‐planting regions with similar growth situations.     

Assimilating satellite imagery and visible–near infrared spectroscopy to model and map soil loss by water erosion in Australia

Soil loss causes environmental degradation and reduces agricultural productivity over large areas of the world. Here, we use the latest earth observation data and soil visible–near infrared (vis–NIR) spectroscopy to estimate the factors of the Revised Universal Soil Loss Equation (RUSLE) and to model soil loss by water erosion in Australia. We estimate rainfall erosivity (R) using the Tropical Rainfall Measuring Mission (TRMM); slope length and steepness (L and S) using a 3-arcsec Shuttle Radar Topography Mission (SRTM) digital elevation model; cover management (C) and control practice (P) using the national dynamic land cover dataset (DLCD) of Australia derived from the moderate-resolution imaging spectroradiometer (MODIS); and soil erodibility (K) using vis–NIR estimates of the contents of sand, silt, clay and organic carbon in Australian soil. We model K using a machine-learning algorithm with environmental predictors selected to best capture the factors that influence erodibility and produced a digital map of K. We use the derived RUSLE factors to estimate soil loss at 1-km resolution across the whole of Australia. We found that the potential gross average soil loss by water erosion in Australian is 1.86 t ha⁻¹ y⁻¹ (95% confidence intervals of 1.78 and 1.93 t ha⁻¹ y⁻¹), equivalent to a total of 1242 × 10⁶ tonnes of soil lost annually (95% confidence intervals of 1195 and 1293 t × 10⁶ y⁻¹). Our estimates of erosion are generally smaller than previous continental estimates using the RUSLE, but particularly in croplands, which might indicate that soil conservation practices effectively reduced erosion in Australia. However we also identify localized regions with large erosion in northern Australia and northeastern Queensland. Erosion in these areas carries sediments laden with nitrogen, phosphorus and pollutants from agricultural production into the sea, negatively affecting marine ecosystems. We used the best available data and our results provide better estimates compared to previous assessments. Our approach will be valuable for other large, sparsely sampled areas of the world where assessments of soil erosion are needed.     

The recognition of road network from high‐resolution satellite remotely sensed data using image morphological characteristics

With the development of remote sensors and satellite technologies, high‐resolution satellite data such as IKONOS images have been available recently. By these new high‐resolution satellite data, remote sensing technologies can be successfully applied to more application areas such as extracting road network from high‐resolution satellite images. This paper proposes a newly developed approach to extract a road network from high‐resolution satellite images. The approach is based on the binary and greyscale mathematical morphology and a line segment match method. First, the outline of road network is detected based on the grey morphological characteristics. Then, the basic road network is detected by the line segment match method. Next, the detected basic road network is processed based on the knowledge about the roads and binary mathematical morphological methods. Finally, visual analysis and three indicators are used to evaluate the accuracy of the extracted road networks. The results of the accuracy evaluation demonstrate that the developed road network extraction approach can provide both good visual effect and high positional accuracy.     

Discrimination of invaded and native species sites in a semi‐desert grassland using MODIS multi‐temporal data

Over the past several decades, one of the most significant changes in semi‐desert grasslands of the southwestern US has been the invasion of South African grass Eragrostis lehmanniana. The objective of this study was to characterize the phenology of systems occupied by E. lehmanniana and/or native grasses using time‐series of field observations and the Moderate Resolution Imaging Spectroradiometer Normalized Difference Vegetation Index (MODIS NDVI) and brightness (red and near‐infrared reflectance) data. Results demonstrated that it was possible to use NDVI and/or spectral reflectance data to discern the phenological differences across a gradient of E. lehmanniana infested grasslands due to variations in plant biodiversity, morphology and seasonal productivity. This work establishes the feasibility of integrating field and MODIS vegetation and spectral time‐series data to characterise landscapes dominated by different herbaceous species, which in turn provides opportunities to monitor E. lehmanniana in semi‐arid environments at a large spatial scale.     

Use of high‐resolution satellite data for the structural and agricultural inventory of tank irrigation systems

Tanks are small storage reservoirs impounding the runoff from monsoon rains to regulate the supply of water mainly for irrigated command areas that are typically less than 200 ha. They account for one‐third of the irrigated areas in Tamil Nadu, Karnataka and Andhra Pradesh. Years of neglect and indifference in tank maintenance and management have eroded their functional efficiency and jeopardized their multifarious benefits. In Tamil Nadu, this has resulted in a decline in their contribution to irrigation from 40% in 1995 to 25% in 2000. The modernization of these tanks requires prioritization and investment. Remote sensing technology, with its unique advantages and the latest high‐resolution sensors, can provide the information on the agricultural, hydrological and structural conditions of the tank irrigation systems necessary for prioritization. The National Remote Sensing Centre (NRSC) has carried out a study of the Nanjur tank cascade in Tamil Nadu using high‐resolution data from the IKONOS satellite during the crop season of 2003–2004. This study demonstrated the use of high‐resolution satellite images to obtain an inventory of the different components of a tank irrigation system such as tank bunds, surplus weirs, supply channels and distribution networks. It was also found useful in updating the road–rail network at village level. The 1‐m merged satellite data were useful in mapping open wells and minor roads in a tank cascade. The cropping pattern in a tank system can be mapped at cadastral level using these images, which will be useful for micro‐level water and agricultural management. The 4‐m multispectral image was found to be sufficient for mapping different crops at field level. The high‐resolution image also provided information on intrafield variability in crop condition. The reliability and cost‐effectiveness of high‐resolution images from Indian satellites provide scope for the generation of information for tank system studies as well as for micro‐level natural resource management.     

Comparing farmer‐based and satellite‐derived deforestation estimates in the Amazon basin using a hybrid classifier

The Amazon basin remains a major hotspot of tropical deforestation, presenting a clear need for timely, accurate and consistent data on forest cover change. We assessed the utility of a hybrid classification technique, iterative guided spectral class rejection (IGSCR), for accurately mapping Amazonian deforestation using annual imagery from the Landsat Thematic Mapper (TM) and Enhanced Thematic Mapper Plus (ETM+) from 1992 to 2002. The mean overall accuracy of the 11 annual classifications was 95% with a standard deviation of 1.4%, and z‐score analysis revealed that all classifications were significant at the 0.05 level. The IGSCR thus seems inherently suitable for monitoring forest cover in the Amazon. The resulting classifications were sufficiently accurate to assess preliminarily the magnitude and causes of discrepancies between farmer‐reported and satellite‐based estimates of deforestation at the household level using a sample of 220 farms in Rôndonia mapped in the field in 1992 and 2002. The field‐ and satellite‐derived estimates were significantly different only at the 0.10 level for the 220 farms studied, with the satellite‐derived deforestation estimates 8.9% higher than estimates derived from in situ survey methods. Some of this difference was due to a tendency of farmers to overestimate the amount of forest within their property in our survey. Given the objectivity and reduced expense of satellite‐based deforestation monitoring, we recommend that it be an integral part of household‐level analysis of the causes, patterns and processes of deforestation.     

Assessing long-term variations in sagebrush habitat – characterization of spatial extents and distribution patterns using multi-temporal satellite remote-sensing data

An approach that can generate sagebrush habitat change estimates for monitoring large-area sagebrush ecosystems has been developed and tested in southwestern Wyoming, USA. This prototype method uses a satellite-based image change detection algorithm and regression models to estimate sub-pixel percentage cover for five sagebrush habitat components: bare ground, herbaceous, litter, sagebrush and shrub. Landsat images from three different months in 1988, 1996 and 2006 were selected to identify potential landscape change during these time periods using change vector (CV) analysis incorporated with an image normalization algorithm. Regression tree (RT) models were used to estimate percentage cover for five components on all change areas identified in 1988 and 1996, using unchanged 2006 baseline data as training for both estimates. Over the entire study area (24 950 km²), a net increase of 98.83 km², or 0.7%, for bare ground was measured between 1988 and 2006. Over the same period, the other four components had net losses of 20.17 km², or 0.6%, for herbaceous vegetation; 30.16 km², or 0.7%, for litter; 32.81 km², or 1.5%, for sagebrush; and 33.34 km², or 1.2%, for shrubs. The overall accuracy for shrub vegetation change between 1988 and 2006 was 89.56%. Change patterns within sagebrush habitat components differ spatially and quantitatively from each other, potentially indicating unique responses by these components to disturbances imposed upon them.     

Integration of Landsat imagery interpretation and geomagentic data on verification of deep‐seated transverse fault lineaments in SE Zagros,Iran

In this paper, Landsat images are used in the mapping of transverse fault lineaments in Zagros, Iran, and the origin of the fault lineaments as the result of basement reactivation faults is examined using geomagnetic maps. The area under investigation is located in the SE part of the Zagros Fold Belt (ZFB) and was affected by influence of concealed faults, or fault lineaments, during the late Alpine Zagros orogeny. Image interpretation of geological structures such as curvilinear geometry of fold hinges, en echelon pattern of surficial lineaments and younger folds, and breached salt plugs, has been used to map the fault lineaments. Two sets of strike‐lateral transverse fault lineaments have been detected. The first set, NW‐trending, was identified by right lateral curved geometry of the main fold hinges, generation of younger folds, and en echelon pattern of surficial lineaments. The second set, NE‐trending, was verified based on left lateral displacement of the structures and alignment of salt plugs pierced anticline hinges. The placement of Upper Proterozoic salt plugs on the surface along the fault lineaments, together with their correlation with the magnetic‐driven lineaments, implies that the fault lineaments have a basement origin. The trend of these fault lineaments can be correlated with the trend of basement faults in the northern margin of the Arabian plate. The key result of this study is that the fault lineaments identified by integration of remotely sensed data and geomagnetic maps are generated as first order wrench faults by reactivation of the N–S‐trending basement faults. The thick Hormoz Salt formations overlaying the basement do not allow the fault lineaments to emerge at the surface. Therefore, they can be mapped through regional remote sensing and their basement origin can only be recognized by their correlation with magnetic faults. It is also concluded that satellite imagery can greatly contribute to the structural mapping of the ZFB and is therefore a valuable aid for oil and gas exploration in analogous tectonic environments.     

Spatio‐temporal modelling and analysis of urban heat islands by using Landsat TM and ETM+ imagery

In this paper we propose a method for characterizing UHIs by using a non‐parametric model. Landsat 5 and 7 images obtained over the city of Indianapolis were used for analysis of UHI over space and time. The images were subjected to kernel convolution implementing Gaussian bi‐variate function for non‐parametric characterization of the UHI. The convoluted images were then analyzed, for pattern over space and over time. It was found that the spread of the UHI in Indianapolis increased from 13.90 km to 21.63 km in the west‐east direction and from 10.98 km to 15.90 km in the north‐south direction from 1985 to 2000. The changes in the UHI from 1995 to 2000 were evident in all cardinal directions. The model aided in successfully characterizing UHI in terms of its location, spread, and the rate of increase, facilitating a clearer image of UHI through space and time.