Landsat-based Monitoring of Landscape Dynamics in the National Parks of the Great Lakes Inventory and Monitoring Network (Version 1.0)

Kennedy, Robert; Kirschbaum, Alan; Gafvert, Ulf; Nelson, Peder; Yang, Zhiqiang; Cohen, Warren; Pfaff, Eric; Gholson, Brett

This protocol details the objectives, methods, and data analysis for monitoring landscape changes over time using remote sensing data and technology in the western Great Lakes region of the United States. Detailed methods are in standard operating procedures (SOPs). The protocol involves compiling and processing a time series of Landsat imagery (since 1984) to automate change detection on the landscape using algorithms developed by research scientists at Oregon State University. Collectively, these algorithms comprise a trajectory-based approach known as LandTrendr (Landsat-based detection of trends in disturbance and recovery). The monitoring will take place at Apostle Islands National Lakeshore (APIS), Voyageurs National Park (VOYA), Isle Royale National Park (ISRO), Sleeping Bear Dunes National Lakeshore (SLBE), Pictured Rocks National Lakeshore (PIRO), St. Croix National Scenic Riverway (SACN), Mississippi National River and Recreation Area, and Indiana Dunes National Lakeshore (INDU). This monitoring is funded and conducted by the NPS Great Lakes Inventory and Monitoring Network (GLKN). The primary justification for monitoring the change in landscape dynamics over time is to document the change where and when it occurs. In addition to where and when, we will also describe the agent of change (forest harvests, blowdown, fire, beaver, development, etc.) and the starting/ending vegetation class for the disturbed area. This will allow managers to better prepare for and then manage for ecosystem changes that are likely to affect processes, systems, and individual species. This protocol will be repeated every year, with a full analysis resample time of 6 years for each park and a midterm, less intensive analysis, occurring every 3 years. Depending on feedback from park personnel, we may revisit a park every 3 years with a less intensive implementation of the protocol. If parks find the revisit unnecessary, we will pursue developing and implementing new protocols for monitoring ice/snow cover and/or phenology using remote sensing data. Specifically, this monitoring protocol is designed to answer the following questions: 1. Where and when are disturbances (≥ 1ha) occurring inside and outside the park? 2. What is causing these disturbances? 3. What is the mean size of disturbances? 4. Has forest harvesting occurred inside park boundaries? 5. As a result of blowdowns, beaver activity and possibly fire, where are large areas of recently downed woody debris located? At the beginning of each year, Landsat imagery will downloaded and processed for inclusion into LandTrendr analysis. Landsat images will be acquired which meet certain parameters for inclusion into the process. The most important parameter for inclusion is date of image acquisition. Each image chosen should have the same phenology as the remainder of images in the time series. Generally speaking, images taken in July and/or August meet the criteria, although these dates can be adjusted slightly depending on the park of interest and its specific phenology. In addition to acquisition of Landsat imagery, additional airphotos or higher resolution satellite imagery (IKONOS, Quickbird, SPOT) should be acquired when funds exist. Additional preprocessing steps to remove clouds and radiometrically normalize the images to other images in the stack need to be taken before running the LandTrendr algorithms.

Landsat-based Monitoring of Landscape Dynamics in the National Parks of the Great Lakes Inventory and Monitoring Network (Version 1.0)

Series