Figure 1.
Classified land cover map with overlaid elevation errors for the 407 field measurements collected with real time kinematic GPS. Representative land covers for the study area: (A) low-lying suburban housing, (B) low energy sandy beaches and (C) low-lying open spaces (D) to the south of the highway and wetlands (D) to the north.
Figure 2.
Representative land covers for the study area.
(A) Low-lying houses, (B) low-energy sandy beaches, (C) low-lying bare earth/open spaces and (D) mangroves forest (Source: Javier X. Leon).
Figure 3.
Flowchart of uncertainty propagation analysis.
Figure 4.
Spatial datasets used in regression analysis.
(A) Digital elevation model (DEM), (B) Normalized difference vegetation index (NDVI) and terrain variables including (C) slope and (D) terrain convexity. DEM was based on data provided by the State of Queensland’s former Department of Environment and Resource Management and NDVI was based on satellite image provided by DigitalGlobe.
Figure 5.
Histogram of elevation errors.
Figure 6.
Box-whisker plots of elevation errors grouped by land cover.
Table 1.
Regression coefficients for best linear model.
Figure 7.
Variograms of elevation errors.
Fitted Matérn variogram based on ordinary kriging model (blue dashed line and squares) and regression-kriging model (red line and circles).
Figure 8.
Interpolated elevation errors.
Inset showing interpolation elevation errors based on ordinary kriging model (left panel) and regression-kriging model (right panel). Location of (A) seawall and (B) park are shown for reference (see main text).
Figure 9.
Probability inundation map for a scenario combining a 2.9 m 100 ARI storm surge event over a 1 m SLR.
The fat solid black line represents the deterministic bathtub-derived inundation border. The thin solid line shows the area that has a probability greater than 1% to become inundated.