Data Management Plan (Deprecated)

This data set consists of salt marsh habitats for the state of Rhode Island derived from high resolution (0.5m) imagery collecting during June 2012. This data set utilized 4 Band multispectral aerial orthoimagery contributed to RIGIS by the Rhode Island Eelgrass Mapping Taskforce [(c) 2012 RIGIS] and LiDAR data collected through the LiDAR for the NorthEast project (2011).

Process Steps:

• 2017-03-17 00:00:00 - This data set was created by NOAA Office of Coastal Management (NOAA OCM). This version of the classification was created by the Narragansett Bay National Estuarine Research Reserve (NBNERR) and partner agencies to include multiple vegetation communities that individually and collectively indicate aspects of marsh health. This section outlines the classification procedure for the Rhode Island Salt Marsh Habitats maps. This product was developed through a partnership between NBNERR and the NOAA OCM. The product was derived from high resolution 4 band digital imagery in combination with derivatives from LiDAR data. The mapping methodology employed an Object Based Image Analysis (image segmentation) approach. Data Preparation: The aerial imagery was supplied to NOAA OCM by NBNERR. Orthoimagery from the spring of 2012 was pre-processed into spectral indices and band derivatives. LiDAR data collected through the LiDAR for the NorthEast project (2011) were processed into elevation and surface models. Field Data Collection: Field samples were collected prior to (August 2014) and during (June 2015) the mapping effort. Salt Marsh Extent Classification: Image segmentation was performed in Trimble's eCognition Developer software package. The resulting image objects or segments were classified using modeling routines which leveraged existing wetland datasets acquired from RIGIS. Boundaries were refined using spectral and elevation information. The final extent was used to focus the supervised habitat classification process. Salt Marsh Habitat Classification: A new set of image objects were derived within the salt marsh extent boundary. The resulting image objects classified into habitat classes using a Random Forest classifier. Logical rule sets and contextual modeling were used to introduce classes such as Dieoff Depressions and Natural Pools and to improve, and refine the habitat classification. Anomalies in the classification that were not addressed through the automated processes were corrected through manual editing. Review and Map Finalization: Once various stages of the salt marsh extent and habitat classification processing were completed, the resulting maps were provided to the Narragansett Bay NERR and other local experts for comments. The majority of the resulting feedback was incorporated to improve the final habitat dataset. Attributes for this product are as follows: 0 Background 1 Unclassified 2 Invasive Phragmites australis 3 High Marsh; S.patens-D.spicata 4 Mudflat /Bare 5 Open Water 6 Dieoff Depression 7 Natural Pool 8 Salt Shrub 9 High Marsh; S.alterniflora 10 High Marsh; Mix 11 Low Marsh 12 Brackish Marsh, Native 13 High Marsh; J.gerardii 14 Wrack