Accuracy Assessment of Vegetation Monitoring with High Spatial Resolution Satellite Imagery

For the years 1997 to 2000 it is expected that a number of new satellites will be launched into orbit by private companies which are specified to deliver panchromatic imagery of the earth surface with a spatial resolution as fine as 1m. In contrast to the panchromatic band, the spectrally resolved bands will have a four times coarser ground resolution. Therefore, image fusion algorithms will be employed in order to produce panchromatic-‘sharpened’ color imagery. The new satellites have the potential of stimulating and expanding the remote sensingmarket for image products at a resolution around one meter. In order to prepare for this era we have examined image fusion algorithms using already available airborne imagery. This paper describes tests of fusion algorithms on imagery which was simulated using multispectral images of an airborne scanner (DAEDALUSATM) with an average pixel size of 1m. The main advantage of this simulation of satellite images is the possibility to measure quantitatively the accuracy of the panchromaticsharpened multispectral imagery. By comparison with the original airborne imagery, we evaluate the accuracy of the sharpened imagery with respect to the spectral signature, NDVI, local variance, and multispectral land cover classification. 1 HIGH RESOLUTION SATELLITE IMAGERY FOR LOCAL ENVIRONMENTAL MONITORING For the years 1997 to 2000 it is expected that a number of new satellites will be launched into orbit by private companies (Doyle 1996, Fritz 1997) which are specified to deliver imagery of the earth surface of a spatial resolution as fine as 1m. This fine a resolution has so far been the privilege of airborne rather than spaceborne overhead imagery – at least as far as the civilian community and multispectral (in contrast to panchromatic) imagery is concerned. Airborne image flights have a longstanding importance for cadastre, local planning and local environmental monitoring (e.g. the health status of public trees in the city of Hamburg is monitored on aerial Color Infrared (CIR) photographs). So far the necessary image flights are conducted mostly by private enterprises on particular customer request. They are thus rather expensive. Multispectral spaceborne imagery on the other hand has been exploited for a number of environmental issues (such as deforestation, desertification, plant stress, water polution, climate warming etc.) but always on a global or regional scale – due to its limited spatial resolution (LANDSAT TM images have a pixel size of 30 30m). With the arrival of meter-range spaceborne imagery which can be purchased off-the-shelf by local authorities at the instance when the demand arises, overhead imagery may become a serious option even for purposes of local interest which up to now could not justify the higher cost of a dedicated image flight. 2 SIMULATION OF HIGH RESOLUTION SATELLITE IMAGERY FROM MULTISPECTRAL AIRBORNE SCANNER IMAGERY First image-recording of the announced satellites is expected for 1998. It can be assumed that the testing and calibration phase will last for the first year of operation. It has to be noted, however, that the schedule for all of the announced satellites already had to be delayed several times. In the meantime, we are in the position to simulate the high resolution satellite imagery from airborne scanner images of comparable spatial resolution. The imagery was recorded by a DAEDALUS ATM line scanner with 10 spectral bands on board a Dornier Do 228 aircraft during five campaigns from 1991 to 1997 in cooperation with the German Aerospace Center (DLR Weßling / München) at flight altitudes of 300m, 900m and 1800m. The 300m imagery has a nadir-looking ground resolution of 70 cm. Due to the panorama characteristic of soSPACE IMAGING EOSAT IKONOS DAEDALUS ATM