A spatial statistical operator applied to multidate satellite imagery for identification of coral reef stress

Abstract

We examine the potential for a spatial operator to highlight the change in the spatial homogeneity of a coral reef over time from satellite imagery. The Getis statistic may be used to quantify the degree of spatial homogeneity or heterogeneity of a spatial data set. We test the application for sequential SPOT imagery that includes a coral reef in the Savusavu region of Fiji that underwent extensive damage due to an industrial accident between the image dates. The ability of the statistic to capture this damage is evaluated through comparison against a null case of open water with no change. A decadal time sequence of SPOT imagery for the Bunaken region of North Sulawesi is also examined to determine whether the statistic may be useful for image understanding where there is a lack of independent knowledge of the degree of coral stress in the area. For the time series of data analyzed for the Bunaken region, we believe that a change in the spatial statistic applied to SPOT data represents a change in reef heterogeneity. Our premise is that heterogeneity is, at this scale, a surrogate for changes in ‘reef health’ with the assumption that a healthy reef illustrates heterogeneity at the scale of the statistical operator. With additional data, this change detection procedure can be used to identify potential ‘hot spots’ that warrant field examination for potential reef damage. This alternative to per-pixel-based approaches has the advantage that it is relatively insensitive to changes in depth, water constituents and atmospheric variability within the area of the spatial operator.