The literature contains many examples of image acquisition and analysis which have been inappropriately applied and which have led to empirical results which may not be reproducible, or which are not conclusive. In this paper, we deal with eleven major assumptions which are implicit in the acquisition and in the analysis of passively sensed digital image data. It is hoped that an enumeration of such assumptions might lead to improved rules for image acquisition and analysis.

Assumptions implicit in remote sensing data acquisition and analysis

A thermal scanner in the 2-μm waveband is used for airborne studies of urban temperature patterns. The urban geometry is modelled by the Sky View-Factor (SVF). Comparisons of surface temperature and SVF show a good correlation. The largest difference in surface temperature (4°C) was recorded between dense canyons and open areas. Ground measurements correspond well with the temperature calculated from the infrared images. Different examples of inlra-urban thermal variations and the utility of the instrument and software for the determination of differences in temperature arc discussed.

Infrared thermography and urban temperature patterns

This study examines and improves upon the classification accuracy of specific land-use/land-cover categories by comparing and integrating spaceborne radar and radar-derived products with optical data. RADARSAT and Landsat Thematic Mapper (TM) multispectral data were assessed independently and in combination. The primary methodology was supervised spectral signature extraction and the application of a maximum-likelihood statistical decision rule to classify surface features in the Kathmandu Valley, Nepal. Relative accuracy of the resultant classifications was established by comparison to ground-truth information. Both radar post-classification smoothing and variance texture measures were improvements over the poor results achieved with the unfiltered, original radar data. Speckle filtering procedures were found to be particularly beneficial. Combinations of radar-derived products greatly improved results, achieving an overall accuracy of 87 percent. The best overall accuracy was achieved with the merger that included a texture image derived from despeckled radar and the despeckled original radar. The radar and radar-derived combination achieved much better results than did the TM and were comparable to a combined radar and TM data classification. These results suggest that radar and radar-derived measures produce the most accurate urban growth mapping.

Radar and Optical Data Comparison/Integration for Urban Delineation: A Case Study

Ecology of the urban forest: Introduction to part III

A significant and recent problem in the Kathmandu Valley of Nepal is the loss of very productive agricultural lands to uncontrolled urban growth. This problem is compounded by the lack of information available to the responsible organizations of the government of Nepal on where, and at what rate, this growth is occurring. A raster-based Geographic Information System (GIS) for the Kathmandu Valley was constructed with cooperating Nepali scientists to evaluate the location and extent of urban growth. The lack of current land cover information resulted in an evaluation of spaceborne data to assess urban expansion. Three dates of multispectral SPOT imagery were obtained for Kathmandu. Only one date, July, was useful (94 percent accuracy), because of the complexity of the crop calendar and confusions between cover types. Manipulation of spaceborne radar data, which has the advantage of cloud penetration, produced an urban location map with an accuracy of 87 percent. After adding the urban maps from remote sensing into the Kathmandu Valley GIS, a variety of spatial analyses were accomplished. Between 1954 and 1989, the urban areas expanded by greater than 3,500 hectares or over 300 percent. Another level of GIS analysis, and one that truly demonstrates the power of spatial data integration, is model building for the prediction of future growth, or for conducting suitability or capability analyses. A model on where future urban growth should be encouraged is presented.

Urban growth in kathmandu, nepal: mapping, analysis, and prediction

We describe a method of spatial filtering in the frequency domain which enhances edges and boundaries, thus making small urban features such as parks, tree-lined streets and new housing developments, visible on digital images with, for example, 30 m resolution. Hitherto, while satellite imagery has been useful because of its large area and repetitive coverage, the spatial resolution for multiband imagery has been such that it precluded the detailed studies which may now be possible. While spatial frequency filtering, edge enhancement, high-boost and directional filtering have been possible, this has generally involved the use of a convolution matrix whose elements have been defined from general empirical rules. Frequency domain operations offer the advantage of selectively tailoring the filter in order to enhance certain features.

R. A. Rowntree

Papers

The application of spatial filtering methods to urban feature analysis using digital image data