Wetland inundation mapping and change monitoring using Landsat and airborne LiDAR data

Abstract

This paper presents a new approach for mapping wetland inundation change using Landsat and LiDAR intensity data. In this approach, LiDAR data were used to derive highly accurate reference subpixel inundation percentage (SIP) maps at the 30-m resolution. The reference SIP maps were then used to establish statistical relationships between SIP and Landsat data. Inundation changes were mapped by applying the derived relationships to Landsat images acquired in different years. This approach was applied to the upper Choptank River sub-watershed to map wetland inundation for average (2005 and 2007), dry (2009), and wet (2010) years. The derived SIP maps revealed large changes in wetland inundation among dry, average, and wet years. Total areas of near complete inundation (SIP>75%) and high inundation (SIP between 50% and 75%) in the wet year of 2010 were about five and three times of those in the dry year of 2009, respectively. The wet year also had more medium inundated areas (SIP between 25% and 50%) than the average and dry years, but low inundated areas (SIP<25%) did not have any particular trend. The mapped inundation changes were found correlated with local drought conditions and stream flow, with the near complete inundated and highly inundated areas having the highest correlations. Given the fact that Landsat are globally available and LiDAR data are becoming increasingly more affordable and available, the approach developed in this study has potential for deriving historical inundation changes over the past decades and for monitoring ongoing changes over much larger areas than demonstrated in this study.