Search

[ABSTRACT] Satellite mapping of large-area forest lands and timberlands is also accomplished in the Philippines with the use of a combination of visible, near-infrared and 3.7μm. thermal-infrared AVHRR data of NOAA-12 morning pass (0730-LST) satellite. This technique was performed on a series of NOAA-12 morning pass images acquired from 1992 to 1994. The linear contrast stretch of the 95% (2 standard deviations from the mean) of the data of the 3 AVHRR channels combined with the use of the 1988 Land Use Map of the Philippines, the 1989 Land Classification Map of the Philippines and the 1:250,000 topographic maps as reference data enabled the detection of a spectral pattern for forest and timberlands and checking of its suitability for delineation. The ranges of the digital values defining this spectral pattern were determined and used in the classification. The separability of the data values of forest/timberlands from those of the surrounding cover types was evaluated using means, standard deviations and data values profiles. A normalized difference vegetation index image was generated and compared with the classification output. Accuracy assessment performed by pixel-to-pixel comparison with the digital 1988 Land Use Map yielded 85% classification accuracy.