GCE IV - Key Finding in 2021
Tools for Tidal Filtering of Remote Sensing Imagery
One of the challenges in using aerial imagery to evaluate intertidal wetlands is the fact that they are regularly covered by water. Aerial images are most useful when the area is exposed, and airplanes and drone flights to assess wetlands are often strategically timed for low tide. However, most satellites that orbit the earth capture images regardless of flooding conditions. This provides a challenge for researchers interested in studying wetlands, as flooding reduces spectral reflectance, making vegetation indices and other characteristics inaccurate and highly variable. In 2017, GCE researcher J. O’Connell developed a tool called the Tidal Marsh Inundation Index (TMII) that automatically filters out inundated pixels in MODIS imagery. The TMII, which was validated based on imagery from the GCESapelo PhenoCam, has proven broadly useful for producing time series of wetland vegetation. It has been used extensively by GCE researchers (e.g. Tao et al. 2018; Alber and O’Connell 2019; Hawman et al. 2021; O’Connell et al. 2021) as well as by researchers at the Plum Island LTER (e.g. Forbrich et al. 2018), and has been cited more than 30 times – including several review papers that have pointed out that a tool like TMII would be useful for other satellite platforms. Building on this work, we have now created a similar application for Landsat-8 and the recently launched Lansat-9 satellites (Narron et al., subm.). The tool, called Flooding in Landsat Across Tidal Systems (FLATS), was again ground-truthed with data from the GCESapelo PhenoCam. It shows how flooding patterns vary over both space and time at a much higher resolution than MODIS, and can also be used to examine changes in flood frequency and flooding patterns (Fig. 1). The strong seasonal pattern in marsh flooding revealed by this analysis may have important implications for marsh ecology. We anticipate that both of these tools will have widespread use for landscape scale analyses of wetland dynamics.
Fig. 1 Evaluation of flooding frequency in the Duplin River based on analysis of Landsat8 imagery using FLATS. The overall average (A) and seasonal pattern (B) showcase the spatiotemporal distribution of flooding across the study area. The pink polygon in panel A shows the field of view of the GCESapelo PhenoCam camera along with an example image, which was used for groundtruthing. Source: Narron et al., subm.
