Airborne Thematic Mapper Research Articles

Potential Improvements in the Characterization of Forest Canopy Gaps Caused by Windthrow Using Fine Spatial Resolution Multispectral Data: Comparing Hard and Soft Classification Techniques

Abstract Gaps often form in forest canopies due to windthrow and have important management and ecological implications. Remote sensing has considerable potential for the provision of information on gap properties but this has not been fully realized. This is largely due to the use of conventional (hard, one-pixel one-class) image analysis techniques and imagery with a relatively coarse spatial resolution. This article investigates the potential to extract information on gap properties from fine spatial resolution airborne thematic mapper imagery using soft classification techniques that allow image pixels to have multiple and partial class membership. It is shown that a standard hard maximum likelihood classification may be used to derive an accurate map of the land cover of a forested site (95.1%) from which gaps in a canopy of Sitka spruce were accurately identified (94.5%). The maximum likelihood classification was also softened by outputting probabilities of class membership for each pixel. Softening the classification increased the information on gap properties that could be extracted from the data. In particular, the accuracy with which key gap properties, such as gap area, perimeter length and shape, were estimated was higher in the outputs of the softened than hard classification. Thus, while strong correlations between the remotely sensed and ground data estimates of gap area (r ≥ 0.96) and perimeter (r ≥ 0.87), based on a sample of 36 gaps, were derived from all classifications, the accuracy with which gap properties were estimated was generally highest when a soft classification was used. For example, the use of a soft rather than hard classification resulted in the root mean square error in estimating gap area declining from 144.90 to 132.87 m2. Furthermore, the soft classification allowed the sharpness of the gap boundary to be estimated, enabling further gap properties to be inferred. In particular, the soft classification output enabled the direction of the wind event causing the initial damage to be estimated, and it may aid the definition of sites with a future risk of windthrow. FOR. SCI. 49(3):444–454.

An evaluation of per-parcel land cover mapping using maximum likelihood class probabilities

The Centre for Ecology and Hydrology (CEH) has recently developed a per-parcel classification procedure which integrates remotely sensed imagery with digital cartography. The aim of the work reported here was to compare the per-parcel approach with a conventional per-pixel classification through examination of maximum likelihood class probabilities. An Airborne Thematic Mapper (ATM) image with 1.25 m spatial resolution was used as a source of training data. The ATM image was spatially degraded to 10 m resolution to form the remotely sensed input and the study area was subdivided into land parcels by manual digitizing. The per-parcel approach extracted raster data from a core area described by a shrunken version of a land parcel, thus eliminating mixed boundary pixels. Boundary pixels were shown to be more variable than the core pixels and their removal improved classification confidence. The per-parcel approach calculated the mean spectral response of the extracted core pixels. This was shown to remove within-parcel variation and improve classification confidence in comparison to per-pixel class allocations. A parcel-based representation was shown to be most appropriate for mapping agricultural land cover in comparison to semi-natural areas, because agricultural landscapes have an inherent parcel structure. When land cover is heterogeneous, as in many semi-natural areas, a per-pixel classification would appear to be more appropriate. A hybrid classifier, which could switch between per-parcel and per-pixel mapping, was suggested as a powerful land cover mapping tool.

The use of airborne remote sensing for extensive mapping of intertidal sediments and saltmarshes in eastern England

Airborne remote sensing with a Compact Airborne Spectrographic Imager (CASI) and a Daedalus Airborne Thematic Mapper (ATM) has been used to map sediment types and biotic associations for the intertidal zone along 270 km of coastline from the Humber Estuary to North Norfolk, UK within the LOIS BIOTA (Land-Ocean Interaction Study, Biological Influences On interTidal Areas) programme. This allows field-based biotic and sedimentary information to be extrapolated to provide information for modelling the coastal processes of erosion and accretion. The method adopted for image classification was unsupervised clustering which was used to map an extensive area (476 km 2 ) of the exposed intertidal zone at a spatial resolution of 5 2 5 m pixels. Ten broad classes of intertidal sediments and vegetation have been mapped with approximately 70% correspondence with ground data.

Multisource data classification with dependence trees

In order to apply a statistical approach to the classification of multisource remote-sensing data, one of the main problems to face lies in the estimation of probability distribution functions. This problem arises out of the difficulty of defining a common statistical model for such heterogeneous data. A possible solution is to adopt nonparametric approaches, which rely on the availability of training samples without any assumption about the related statistical distributions. The purpose of this paper is to investigate the suitability of the concept of dependence trees for the integration of multisource information through estimation of probability distributions. First, this concept, introduced by Chow and Liu (1968), is used to provide an approximation of a probability distribution defined in an N-dimensional space by a product of N-1 probability distributions defined in two-dimensional (2-D) spaces; this approximation corresponds, in terms of graph theoretical interpretation, to a tree of dependence. For each land cover class, a dependence tree is generated by minimizing an appropriate closeness measure. Then, a nonparametric estimation of the second-order probability distributions is carried out through the Parzen window approach, based on the implementation of 2-D Gaussian kernels. In this way, it is possible to reduce the complexity of the estimation, while capturing a significant part of the interdependence among variables. A comparison with other multisource data fusion methods, namely, the multilayer perceptron (MLP) method, the k-nearest neighbor (k-NN) method, and a Bayesian hierarchical classifier (BHC), is made. Experimental results obtained on multisensor [airborne thematic mapper (ATM) and synthetic aperture radar (SAR)] and multisource (experimental synthetic aperture radar (E-SAR) and a textural feature) data sets show that the proposed fusion method based on dependence trees is able to provide a classification accuracy similar to those of the other methods considered, but with the advantage of a reduced computational load.

Optimizing sampling strategies for estimating mean water quality in lakes using geostatistical techniques with remote sensing

AbstractIn planning a sampling regime, it is desirable that the sampling procedure should involve minimum estimation error for a given sample size or minimum sampling effort for a given accuracy. Two approaches for matching sampling effort to accuracy may be used: a classical approach, which ignores spatial dependence between observations, and uses a random scheme; and a geostatistical approach, which exploits spatial dependence, and uses a systematic scheme. Four Airborne Thematic Mapper images of two British lakes were processed to provide a chlorophyll index, reflecting variations in chlorophyll‐a concentration. Spatial structure was characterized using the variogram, and the modelled variogram was used in Kriging to plan sampling regimes for estimating the mean chlorophyll. For a given sample size, the systematic scheme incurred less error than the random scheme; and for a given error, the systematic scheme required smaller sample sizes than the random scheme. The relative advantage of the systematic approach over the random sampling approach increased with an increase in sample size and an increase in the proportion of variance in the data that was spatially dependent. This paper demonstrates that the sampling regime must be calibrated to the spatial dynamics of the lake under investigation, and suggests that remote sensing is the ideal means by which to determine such dynamics.

Estimation of velocity fields at the estuary-coastal interface through statistical analysis of successive airborne remotely sensed images

The study of surface water velocity fields through in situ sampling is intrinsically difficult because they are highly variable in time and space. With airborne remote sensing, however, it is possible to determine changes in velocity fields because spatially and temporally comprehensive data may be obtained. This letter shows how changes in the statistical properties of successive remotely sensed images may be used to estimate velocity vectors associated with chlorophyll-a and sea surface temperature (SST). The study area is Kirkcudbright Bay, a small estuary in south-west Scotland. Multi-temporal imagery of the study area were acquired by the NERC Daedalus ADDS-1268 Airborne Thematic Mapper (ATM) and processed to show chlorophyll and thermal indices as substitutes for chlorophyll-a and SST. Velocity fields were estimated by the Maximum Cross Correlation technique. Complex patterns were found, confirming that the comprehensive coverage provided by airborne remote sensing is required for their analysis. The chlorophyll-a velocity field differed from the SST velocity field, suggesting that these fields are relevant to the water quality variable in question, and not necessarily the water body itself.

Characterising windthrown gaps from fine spatial resolution remotely sensed data

This paper evaluates the potential of fine spatial resolution remotely sensed data for the identification and characterisation of windthrown gaps. The study site was Cwm Berwyn Forest, in central Wales, a planted forest of predominantly Sitka spruce ( Picea sitchensis (Bong.) Carr.) containing windthrown gaps ranging in size from 50 to 3000 m 2. The remotely sensed data used were acquired by an 11 waveband airborne thematic mapper (ATM) sensor with a spatial resolution of ≈4 m. This resolution is finer than the gaps on the site and comparable to that of future satellite sensors. A three class (gap, forest canopy and moorland) thematic land cover map was produced using a conventional maximum-likelihood classification of the data and provided an accurate representation of the classes (>90% of the pixels allocated correctly). This classification provided an accurate means of identifying windthrown gaps and was capable of identifying a greater number of gaps, than manual interpretation of temporally coincident aerial photographs. In addition, the magnitude and spatial distribution of class membership probabilities derived from the classification provided information on various gap properties, such as exposed soil and living, windthrown tree canopies. This may enable other properties such as windthrow direction to be inferred.

Remote sensing evidence for the episodic transport of phosphorus from the littoral zone of a thermally stratified lake

Summary 1 Esthwaite Water in Cumbria is a small, thermally stratified lake fringed with beds of reed in the shallow littoral. In this study, we used a combination of in situ measurements and airborne remote sensing to investigate some of the physical processes influencing the transport of water and nutrients from the littoral zone. 2 The analysis of water samples collected from the reed beds, the littoral zone and the open water showed that significantly higher concentrations of dissolved reactive phosphorus (DRP) were frequently recorded in the reed beds in early summer. 3 Experiments with surface and near-surface free-running drogues demonstrated that the movement of water from the littoral zone was strongly influenced by the development of a secondary thermocline. When there was no secondary stratification, the surface currents generated by light winds seldom exceeded 2 cm s-1. When a secondary thermocline was present, surface current speeds of 5–10 cm s-1 were recorded even when the wind speed was less than 200 cm s-1. 4 A series of thermal surveys using a Daedalus Airborne Thematic Mapper (ATM) demonstrated that plumes of warm water frequently developed in the littoral zone when the weather was calm. Some of these plumes covered several hundred square metres and persisted for several hours. Others were only a few metres in width and acted as episodic ‘pumps’ that appeared and dispersed in less than an hour.

Quantifying geomorphic and riparian land cover changes either side of a large flood event using airborne remote sensing: River Tay, Scotland

The potential of high resolution multi-spectral airborne remote sensing to detect and quantify changes in channel morphology and riparian land cover are illustrated. The River Tay is a partially embanked wandering gravel bed river, and Airborne Thematic Mapper data, which collect reflectance in the visible, near, mid and thermal infrared, were acquired in 1992 and 1994 either side of a 1:65 recurrence-interval flood event. Imagery was radiometrically, atmospherically and geometrically corrected in order to minimise atmospheric and geometric changes between the 1992 and 1994 scenes. A maximum likelihood classifier was then used on each image and change was quantitatively mapped using a classification comparison approach. Bathymetric mapping was undertaken by applying a Lyzenga algorithm to ATM bands 5, 6 and 8 to account for an exponential decrease in electromagnetic radiation penetration through the water column with depth. Despite the magnitude of the flood event, no major changes in channel position or form occurred but the change detection algorithms revealed subtle changes not observed in the field. Bar head accretion, bar tail formation and extension, bar dissection, localised bank erosion and the overriding of low level vegetated islands by gravel lobes were the main forms of change. On the floodplain, flood embankment failures resulted in fans of sands and gravels on agricultural land. More generally, the study reveals the potential for using airborne remote sensing to detect change in fluvial systems and as a mutually complementary tool to field survey.

Construction of a Digital Elevation Model of the Holderness Coast using the waterline method and Airborne Thematic Mapper data

This paper explores the potential for using the waterline method with Airborne Thematic Mapper (ATM) data to construct a Digital Elevation Model (DEM) of a steep and narrow beach. ATM images acquired at different stages of the tide were processed to delineate accurately the position of the shoreline. The sea surface is regarded as an altimeter, the altitude of which is determined using a hydrodynamic model. Geometrically registered shorelines were assigned heights using elevations predicted by the model. The heighted shorelines obtained were interpolated using universal block kriging to produce a DEM of the area. A DEM was also produced from transect data acquired for the beach. The shoreline DEM is compared with the transect DEM in order to assess the accuracy of the former. It is also shown that a more accurate DEM can be generated by cokriging the combined shoreline and transect data.

The significance of border training patterns in classification by a feedforward neural network using back propagation learning

Training patterns vary in their importance in image classification. Consequently, the selection and refinement of training sets can have a major impact on classification accuracy. For classification by a neural network, training patterns that lie close to the location of decision boundaries in feature space may aid the derivation of an accurate classification. The role of such border training patterns and their identification is discussed in relation to a series of crop classifications from airborne Thematic Mapper data. It is shown that a neural network trained with a set of border patterns may have a lower accuracy of learning but a significantly higher accuracy of generalization than one trained with a set of patterns drawn from the cores of the classes. Unfortunately, conventional training pattern selection and refinement procedures tend to favour core training patterns. For classification by a neural network, procedures which encourage the inclusion of border training patterns should be adopted as this may facilitate the production of an accurate classification.

Integrated use of Landsat TM and SPOT panchromatic imagery for landslide mapping: case histories from southeast Spain

Abstract This paper demonstrates some of the ways in which digital space imagery can be used for landslide mapping. Landslides ranging in size from a few metres to several hundred metres across have been studied at a test site in semi-arid terrain in Almeria Province, SE Spain. The study illustrates the capabilities and limitations imposed by spatial and spectral resolution in Landsat Thematic Mapper (TM) and SPOT panchromatic space imagery. The results demonstrate the scales at which, landsliding which can be studied, using SPOT P and Landsat TM imagery. Spectral and textural features of landslides were enhanced digitally. Textural discrimination was improved by linear contrast stretching combined with edge enhancement by adding back the results of Laplacian filtering. Spectral features of landslides, including soil moisture dependent vegetation and rock debris types, are well-displayed in supervised RGB colour composites and band ratio images. A data integration technique, the Brovey Transformation, was used to produce colour composites combining textural and spectral features from SPOT Pan and Landsat TM imagery. Landslides with a minimum width of 250 metres can be identified in space images. Interpretation of results revealed the presence of several previously unrecorded landslides. Daedalus 1268, Airborne Thematic Mapper (ATM) digital imagery has been used to demonstrate the value of improved spatial and spectral resolution imagery for landslide mapping on the scale of the Spanish test sites. The results illustrate the future system requirements for more detailed landslide mapping and demonstrate the improvements of scale which will occur when space imagery with pixel sizes of five metres becomes more readily available.

Digital image texture analysis for landslide hazard mapping

Abstract Textural enhancements using co-registered, multi-temporal, digital remotely sensed imagery have been used to study slope instability at three landslide localities. The remotely sensed imagery consists of Landsat TM, SPOT Pan and Daedalus 1268 Airborne Thematic Mapper (ATM); stereo, black and white aerial photographs have also been used. Image processing provides spatial and textural information concerning landform change, landslide location and morphology. The enhanced imagery can be integrated with digital elevation models (DEMs), as can other map and geotechnical data, into a GIS to analyse quantitatively the effect of slope geometries on the observed landsliding. The suitability of available imagery depends upon the type and scale of the study and the landslide dimensions. The case histories have shown that spatial resolution is the most important parameter for successful landslide identification. The optimum imagery is provided by a combination of textural and spectral enhancements using different types of imagery.

Leak detection from rural aqueducts using airborne remote sensing techniques

Water companies require quick methods of identifying leakage from rural subterranean aqueducts over large areas. This research assessed the possibility of locating water leaks using airborne remotely sensed data and thus considered if such leaks expressed unique and identifiable features. Analysis of soil moisture and vegetation biomass inferred using Airborne Thematic Mapper imagery enabled two known leaks to be highlighted as distinct from their immediate surroundings on the Vyrnwy Aqueduct, North West England.

Fully fuzzy supervised classification of land cover from remotely sensed imagery with an artificial neural network

Mixed pixels are a major problem in mapping land cover from remotely sensed imagery. Unfortunately, such imagery may be dominated by mixed pixels, and the conventional ‘hard’ image classification techniques used in mapping applications are unable to appropriately represent the land cover of mixed pixels. Fuzzy classification techniques can, however, accommodate the partial and multiple class membership of mixed pixels, and be used to derive an appropriate land cover representation. This is, however, only a partial solution to the mixed pixel problem in supervised image classification. It must be reognised that the land cover on the ground is fuzzy, at the scale of the pixel, and so it is inappropriate to use procedures designed for ‘hard’ data in the training and testing stages of the classification. Here an approach for land cover classification in which fuzziness is accommodated in all three stages of a supervised classification is presented. Attention focuses on the classification of airborne thematic mapper data with an artificial neural network. Mixed pixels could be accommodated in training the artificial neural network, since the desired output for each training pixel can be specified. A fuzzy land cover representation was derived by outputting the activation level of the network's output units. The activation level of each output unit was significantly correlated with the proportion of the area represented by a pixel which was covered with the class associated with the unit (r>0.88, significant at the 99% level of confidence). Finally, the distance between the fuzzy land cover classification derived from the artificial neural network and the fuzzy ground data was used to illustrate the accuracy of the land cover representation derived. The dangers of hardening the classification output and ground data sets to enable a conventional assessment of classification accuracy are also illustrated; the hardened data sets were over three times more distant from each other than the fuzzy data sets.

Airborne remote sensing of macrophytes in Cefni Reservoir, Anglesey, UK

This paper evaluates the use of Daedalus Airborne Thematic Mapper (ATM) remotely sensed imagery for monitoring the distribution of aquatic macrophyte species in Cefni Reservoir on the Isle of Anglesey. Images acquired by the ATM were compared to spectroradiometric measurements of the reflectance characteristics of the water and selected stands of submersed floating-leaved and emergent macrophytes. The information content of individual and combinations of ATM spectral bands were investigated for mapping macrophyte distributions. Discriminant analysis performed on both data sets indicated that good identification of macrophytes could be achieved by a combination of green, red and near infrared wavebands. A minimum distance supervised classifier using ATM bands 3, 7 and 8 showed separation of the species surveyed. The results indicate that airborne remotely sensed data have good potential for monitoring freshwater macrophyte species.

The observation of the surface roughness characteristics of the Rhine plume frontal boundaries by simultaneous airborne thematic mapper and multifrequency helicopter-borne radar scatterometer

In this paper, we describe how high spatial resolution (10 m) multisensor remote sensing techniques can be used to study the surface roughness characteristics of large scale frontal boundaries (in this case associated with the Rhine Plume). The instrumentation employed in the research consisted of a Daedalus AADS 1268 Airborne Thematic Mapper (ATM) operated by the UK National Environment Council, the HELISCAT helicopter-borne multifrequency microwave scatterometer of the University of Hamburg, and research vessels (R.V.s) from the University of Wales and the Dutch Rijkswaterstaat. The data we present were gathered on 24 April 1991 when calm wind conditions developed within the test area. A sequence of thermal infrared images gathered by the ATM provides a record of the motion of a frontal boundary through this experimental region which is then used to identify the frontal signature in the HELISCAT data. ATM sunglint images show that the front is characterized by a zone of reduced surface roughness, some 75m in width, which is detected on the 'upstream' side of the front (as defined relative to the tidal flow direction), where surface current convergence can be expected. Radar backscatter levels at X and C bands are reduced by 10 dB in this region but with increase in radar wavelength, the signature weakens and is rarely detected at L band. On crossing the front in the downstream direction, radar backscatter levels are rapidly restored. The available evidence indicates that the reduced backscatter signature is caused by a surface slick which is formed at the frontal interface rather than by short gravity wave damping from shear in local surface currents.

The airborne remote sensing of phytoplankton chlorophyll in the lakes and tarns of the English Lake District

The results of two Daedalus Airborne Thematic Mapper surveys of phytoplankton chlorophyll in six large lakes and ten small tarns are described. The water bodies surveyed covered a range of trophic states, but were all 'Case 1' waters whose optical properties were primarily determined by the concentration of phytoplankton in suspension. The surveys were supported by in situ measurements of phytoplankton chlorophyll and radiometric measurements of the upwelling radiance at selected sites. Statistical analyses showed that radiance measurements in ATM Band 3 (520-600 nm) provided the best estimates of phytoplankton chlorophyll in the more productive waters (correlation coefficient r = 0.97). In waters where the measured concentration of chlorophyll was less than 10 mu g l -1, the most effective pigment retrieval algorithm ( r = 0.91) was that based on the ratio of the radiances in ATM Band 3 and ATM Band 2 (450- 520nm). The two algorithms were then used to estimate the average concentration of phytoplankton in each water body and to produce maps of the spatial distribution of chlorophyll in the larger lakes. Chlorophyll concentrations could be predicted with an overall accuracy of +/- 3 mu g l -1 ( r = 0.98), and spatial distribution maps produced with contour intervals of 2-10 mu g l -1.

An evaluation of some factors affecting the accuracy of classification by an artificial neural network

Artificial neural networks are attractive for the classification of remotely sensed data. However, a wide range of factors influence the accuracy with which a data set may be classified. In this paper, the effect of four factors on the accuracy with which agricultural crops may be classified from airborne thematic mapper (ATM) data was investigated. These factors related to the dimensionality of the remotely sensed data, the neural network architecture, and the characteristics of the training and testing sets. A total of 288 classifications were performed and their accuracies evaluated. The artificial neural networks were able to classify the data to high accuracies, with kappa coefficients of up to 0.97 obtained, but the accuracy derived was highly dependent on the factors investigated. A log-linear modelling approach was used to evaluate the simultaneous effect of the factors on classification accuracy. Variations in the dimensionality of the data set, as well as the training and testing set characteristics had a significant effect on classification accuracy. The network architecture, specifically the number of hidden units and layers, did not, however, have a significant effect on classification accuracy in this investigation. This highlights the need to consider a broader set of issues than network architecture when using an artificial neural network for image classification.

Simulated reflectance technique for ATM image enhancement

The thermal band of the Airborne Thematic Mapper (ATM) has the same spatial resolution as that of the reflective spectral bands and therefore can be used together with other bands without degrading the resolution as happens with Landsat TM. Here we show that the total solar irradiance on the land surface can be simulated by a properly weighted summation of the thermal band and all the reflective spectral bands. The simulated reflectance as well as emittance can then be derived from the relation between irradiance, thermal emittance, spectral reflectance and albedo based on a simplified energy conservation model of solar radiation incident on a land surface. Compared with other methods for spectral enhancement and topography suppression, such as ratio and logarithmic residual, this method has the advantage of conserving surface albedo information and thus produces a reasonable simulation of reflectance.