We have investigated the value of several high spatial resolution satellite sensors for mapping of coral bottoms in Belize. The sensors were evaluated for the possibility to map cover type and changes of coral bottoms with respect to their spectral and spatial properties. Comparisons between different sensors with similar wavelengths have shown that almost all main structures (reef, coral sand, etc) show up remarkably well even in lower spatial resolution (30 m cf. 4 m), but the character within each object is much better in the higher resolution. Unsupervised classification was applied and higher spatial resolution gave better separation of classes within the reef, but what sensor to use was dependant upon the purpose of the classification.
The better quantisation of the Landsat and SPOT versus IRS showed up as sharper images. SPOT and IKONOS data verified our earlier detection of coral bleaching in Belize. The following sensors were used (Table 1-2)
If the purpose is only to get a coral map over a large area, Landsat or SPOT is less time consuming and less costly than IKONOS and often of sufficient quality. For small areas and details IKONOS is better. More narrow spectral bands are required in order to more accurately separate reef species and to distinguish between bleached corals and other bottoms or between dead corals invaded by algae and living corals.
From this study we can conclude that the IKONOS could be used for certain mapping tasks and monitoring of bleaching. We still need a better spectral resolution in order to define the type of change. The cost for IKONOS makes it difficult to use in global surveys. Landsat-7 could be used for coarser mapping and updating purposes, but in the future we need a more specific reef sensor with a few selected narrow bands and a good spatial and temporal resolution.
A follow up of the coral bleaching study including also SPOT and IKONOS indicates that many bleachings were in accordance with field observations having different spectral properties than unaffected areas and thus verified our earlier findings.
| Satellite | Sensor | Track/Frame | Date | Sun elev. | Resampl. method | Aq. quant. (bits) |
| Landsat | TM | 19/48 | 870221 | 44 | CC | 8 |
| Landsat | TM | 19/49 | 950518 | - | - | 8 |
| Landsat | TM | 19/48 | 991129 | - | NN | Best 8 of 9 |
| IRS 1C | LISS3 | 283/06100 | 980228 | 57 | NN | 7 |
| IRs 1D | LISS3 | 292/06100 | 980831 | 71 | CC | 7 |
| SPOT | HRV 2 | K612/J318 | 980910 | - | - | 8 |
| IKONOS | MSS | *)364625 | 010313 | - | - | 11 |
| IKONOS | PAN | *)1823470 | 010313 | - | - | 11 |
The LISS-III sensors on the IRS satellites have two channels in the visible part of the spectrum (Table 2). Unfortunately they lack a blue wavelength band, but the green channel covers approximately the same wavelength as channels in TM and SPOT.

Table2. Waweleghts and spatial resolutions of the sensors.

Map of the coverage of the Belize image.