False Color Composite
This false color composite is possibly the best for displaying the features in and around the Endla Mire Complex at this time of the year.Bands 3, 4, and 5 displayed as blue, green, and red contain almost all of the imformation in a dataset. Band 3 is the red chlorophyll absorption band, useful to show plants versus bare soil. Band 4 is the reflective infra-red band, discriminating between types of vegetation and between land and water. Band 5 is sensituve to the turgidity of plant tissues, indicating plant vigor (Jensen, J. Remote Sensing of the Environment, Prentice-Hall, 2000). This image has overlays showing the relationship between vegetation and other features.
The diagonal mesh defines the land over 80 meters. In fact, the highest land rises only 15 meters above this contour and the lowest land lies only 10 meters below it. Transport facilities (roads and rail) mostly follow higher ground. This is not so important with today's technology, but Estonia has been settled for millenia. Modern transortation routes follow those laid down centuries ago when routes had to follow higher drier ground.
Settlements and cropland (pink) are also located at higher levels where cropland now occupies what was once deciduous forest. Not only are buildings cheaper to construct on the higher dry land, the habitations are further from the breeding grounds of insects, which can be unpleasant in spring and summer.
Drainage systems are evident in the scene. Some of the lowlands are suitable for agriculture, especially since the low-lying land is rich in organic material. While there are some drainage shemes within the Endla Nature Reserve, most are outside. The land occupied by the nature reserve may have been too low lying and thus too expensive to drain with methods available in the past. Fortunately, the land is now protected so that future improved drainage methods will not result in the loss of these wetlands.
The bright reddish-pink areas (upper-left and bottom-center) are sites of peat-mining that do not appear in composite images formed from other bands. Peat is mined from areas that are not so water-logged as the bogs themselves. This prompts the question whether or not these areas contain ombrotrophic peat formed during the Holocene in the same manner as the blanket mires elsewhere in Europe as a result of the interaction between climate and land use. This question may be raised by inspection of the images, but would have to be addressed by field and laboratory investigations.
Notes on Making the Image
The basic image was the 3-4-5 composite. The composite was formed using 2.5% saturation. This means that the contrast of some of the bands was found to be too high. By stretching the gray scales of the component images a better color balance was obtained as judged on the basis of color saturation.To make it possible for the thermal band to be used it was expanded by a factor of two to match its 60-meter resolution with the 30-meter resolution of the other bands.
Overlaying the roads, rail, boundary, and elevation features required that all layers be co-registered. It would be possible to use the plane system of the original dataset. However, the standard is now the Universal Transverse Mercator System (UTM) and this was used, first to geo-register the dataset, and then the features. The basis for determining control points was a TM image of the scene obtained free of charge from the University of Maryland.
The only "tricky" point in geo-registering the ETM+ dataset was to change the values in the metadata file as the first step. Kilometers were changed to meters and the mininum/maximum values for the expanded thermal band were changed to match the other bands.
The vector layer for roads is flawed: in places the roads from the map do not quite align with the roads from the satellite image. This arises from trying to convert all roads shown in the raster map into one vector layer using IDRISI. (This misalignment would not occur if the Clark Labs product Cartalinx were used to trace the road alignments.)
Using the IDRISI routine for converting raster maps to vector layers, more accurate placing of the line vectors for roads could be obtained by making several raster maps with one or two road routes per map and converting each to a vector layer. The roads in the raster images should follow one or two routes across an image from edge to edge to ensure that the IDRISI conversion routine will work correctly. While this approach will result in duplication of some road sectors in the vector layers, it will also provide a visual check on the accuracy of the process.
