Mapping Seagrass from Space: Addressing the Complexity of Seagrass LAI Mapping

Abstract

Information of seagrass LAI is still lacking in most parts of the world due to the high cost of comprehensive mapping. In this paper, we described the use of remote sensing as the cost and time effective solution to perform continuous seagrass LAI mapping, and discussed the issues and difficulties encountered during the mapping. ASTER VNIR and ALOS AVNIR-2 were used to perform the mapping. We proposed at life-form seagrass classification scheme to accommodate the low accuracy of at species level mapping. We also developed sampling mapping unit consist of several factors affecting the distribution of seagrass LAI. The results showed that sensor, method, and environmental limitation contribute to the low accuracy of seagrass LAI mapping.

Improving the accuracy of Multispectral-based benthic habitats mapping using image rotations: the application of Principle Component Analysis and Independent Component Analysis

“…For example, seagrass [Phinn et al, 2008;Roelfsema et al, 2014] and macro algae [Kutser et al, 2006;Oppelt et al, 2012;Wicaksono, 2014] could be detailed into species level, but not possible for coral reefs class as the species variation of coral reefs could not be resolved with the satellite remote sensing image available today [Goodman et al, 2013], unless we used hyperspectral measurement on very high spatial precision Atkinson, 2000, 2003;Lucas and Goodman, 2015]. On the other hand, it might be possible to use the life-form as a working class descriptor for seagrass [Wicaksono and Hafizt, 2013] and coral reefs [Mumby and Harborne, 1999], but not for macro algae since their lifeform such as turf, calcareous and coralline is very difficult to be identified spectrally using multispectral image, especially when different life-forms contain similar characteristics of pigmentation [Wicaksono, 2014]. Ideally, since visible bands as the water penetration bands are sensitive to the variation of pigments composition and concentration [Penuelas et al, 1993], the detailed classification scheme should be constructed based on the variation of pigmentation characteristics of coral reefs, seagrass and macro algae.…”

“…3). The water column effect, which leads to the variation of benthic habitats reflectance at various depths, was compensated using the Depth Invariant Bottom Index (DII) [Lyzenga, 1978] and the Inversed Model (IM) [Wicaksono and Hafizt, 2013]. To perform water column correction using DII method, the reflectance of benthic habitat located at different depths is required, for which it is necessary to create a scatter plot between band pairs and to derive the ratio of water column attenuation coefficients for each band pair.…”

“…The detail and measurement procedure of the field bathymetry data used in this research is described in Wicaksono [2015]. The optically shallow water area of Kemujan Island was categorized into seven depth ranges: 0-1 m, 1-2 m, 2-3 m, 3-4 m, 4-7 m, and 7-18 m. Most benthic habitats in Kemujan Island are located in water shallower than 7 while 18 m is the maximum depth of penetration of visible bands in Kemujan Island water [Wicaksono and Hafizt, 2013]. In order to minimize error propagation in the water depth estimation, the regression analysis was carried out individually for each depth range instead of the whole depth range.…”

“…The water column attenuation coefficient for each band (k) was compiled from previous researches. The value for the blue, green, and red band is 0.018, 0.072, and 0.178 m -1 , respectively [Wicaksono and Hafizt, 2013]. These values were calculated using the modified Jupp [1988] bathymetry model.…”

“…Clear water reflectance for the cyan, blue, green, yellow, red and red-edge band is 0.002, 0.011, 0.015, 0.005, 0.003, and 0.003 respectively. From this information, the effect of water column in each deglint band was corrected using the IM formula as described in Wicaksono and Hafizt [2013]. …”