Mapping marine environments with IKONOS imagery: Enhanced spatial resolution can deliver greater thematic accuracy

Mumby, Peter J. and Edwards, Alasdair J. (2002) Mapping marine environments with IKONOS imagery: Enhanced spatial resolution can deliver greater thematic accuracy. Remote Sensing of Environment, 82 2-3: 248-257. doi:10.1016/S0034-4257(02)00041-X


Author Mumby, Peter J.
Edwards, Alasdair J.
Title Mapping marine environments with IKONOS imagery: Enhanced spatial resolution can deliver greater thematic accuracy
Journal name Remote Sensing of Environment   Check publisher's open access policy
ISSN 0034-4257
1879-0704
Publication date 2002-10
Sub-type Article (original research)
DOI 10.1016/S0034-4257(02)00041-X
Volume 82
Issue 2-3
Start page 248
End page 257
Total pages 10
Place of publication New York, United States
Publisher Elsevier
Language eng
Formatted abstract
The IKONOS 2 satellite was launched in late 1999 and carries the first commercial multispectral instrument to achieve 4 m spatial resolution. The cost and accuracy of using IKONOS imagery to map shallow-water marine environments is evaluated and compared directly to that using a suite of satellite and airborne instruments including Landsat Multispectral Scanner (MSS) and Thematic Mapper (TM), Satellite Pour l'Observation de la Terre (SPOT) High-Resolution Visible (HRV) multispectral and panchromatic, and Compact Airborne Spectrographic Imager (CASI). Evaluations were conducted in the Turks and Caicos Islands which are ideally suited to aquatic remote sensing because of the large areas of clear, shallow water containing a range of different habitats. Over 600 field sites were surveyed and used to define habitat categories, supervise image classification, and make an independent assessment of thematic map accuracy. For the high-resolution IKONOS imagery, pixel sizes were small enough to allow within-habitat textural information to be added to the classification. Making full use of this textural information in supervised classifications significantly improved (P<.01) thematic map accuracy for fine-level habitat discrimination (13 classes). Although significantly (P<.001) and almost 20% more accurate than Landsat TM, like other satellite-borne sensors, IKONOS data were unable adequately to discriminate 13 categories of coral, algal, and seagrass habitats (overall accuracy 50%). Comparison with CASI in a restricted area where only nine of these habitats were represented indicated that CASI (81% user accuracy) was significantly (P<.01) more accurate than IKONOS (64% user accuracy). Comparisons of IKONOS, Landsat TM, and CASI along a transect in the maximally penetrating blue part of the spectrum suggest that the poor descriptive resolution of satellite sensors is partly due to loss of radiance contrast, presumably as a result of atmospheric Rayleigh scattering. Problems of scattering, coupled with the relatively poor spectral resolution of satellite sensors such as IKONOS, currently constrain their use for fine-level marine habitat mapping. IKONOS data enabled significantly more accurate mapping at a geomorphological scale (e.g., coral vs. seagrass) than other satellite sensors except Landsat TM. While reasonably high accuracies (75%) were obtained for such coarse-level habitat mapping, Landsat TM was more cost-effective than IKONOS, even for small (50 km2) areas. IKONOS may only be a cost-effective option if (i) independent field data are available to identify habitat patches, (ii) the area to be mapped is fairly small (<500 km2), and (iii) small-scale (<10 m) habitat dynamics are to be monitored. IKONOS data are unlikely to identify deterioration in coral reefs directly because changes in community structure cannot be resolved spectrally from the satellite.
Q-Index Code C1
Q-Index Status Provisional Code
Institutional Status Non-UQ

Document type: Journal Article
Sub-type: Article (original research)
Collection: School of Biological Sciences Publications
 
Versions
Version Filter Type
Citation counts: TR Web of Science Citation Count  Cited 168 times in Thomson Reuters Web of Science Article | Citations
Scopus Citation Count Cited 208 times in Scopus Article | Citations
Google Scholar Search Google Scholar
Created: Tue, 23 Nov 2010, 14:37:11 EST