You are here

Landscape Change Monitoring System (LCMS)

January, 2016

The Landscape Change Monitoring System (LCMS) is an emerging remote sensing-based system for mapping and monitoring land cover and land use change across the US. Envisioned as a framework for

Map of United States showing locations of LCMS studies in Maine, Pennsylvania/New Jersey, South Carolina, Minnesota, Colorado, and Oregon
Map of pilot study locations
integrating Landsat-based products and other datasets, LCMS  is producing spatially, temporally, and thematically comprehensive data and information from which landscape change can be consistently assessed, documented, and analyzed at the national scale. 

LCMS is a direct outgrowth of two separate efforts: (1) the Monitoring Trends in Burn Severity (MTBS) program co-lead by the US Geological Survey (USGS) and the US Forest Service (USFS); and (2) the NASA-funded North American Forest Dynamics (NAFD) project which had significant involvement of the USFS, both in terms of science development and forest inventory applications. As an interagency USGS-USFS program, LCMS supports the change detection needs of a range of federal and non-federal land managers, and specifically targets the information needs of initiatives such as LANDFIRE, a joint USGS-USFS program for mapping vegetation, fire, and fuel characteristic for the US, and the National Land Cover Database (NLCD). When successfully implemented, LCMS could become the dominant provider of national-level Landsat-based change information for all lands of the US, with agency partners and specific programs augmenting the change products to suit their needs. LCMS is currently in development, with commitments for partial funding from the USFS to conduct pilot studies.

The LCMS Science Team (LST) includes developers of diverse Landsat-based change detection algorithms that address different types and rates of change over a variety of cover types in the context of two pilot studies.


One pilot study focuses on six sites around the US where we are comparing and evaluating Landsat change map products from these algorithms, where each site is defined by a single Landsat scene (by WRS2 path/row, and State): 12/28, ME; 14/32, PA/NJ; 16/37 SC; 27/27, MN; 35/32, CO; and 45/30,OR (Map of sites). Validation for this assessment is being conducted with the Landsat visualization software called TimeSync and relevant airphoto and field-based datasets. The goal is to highlight the unique and complementary capabilities of various algorithms for describing forest change in terms of year of detection, sensitivity to change magnitude, ability to relate spectral trends as well as anomalies to disturbance occurrences, and disturbance causal agent. Because no algorithm is expected to perform best in all situations, an important research task is to develop and test an ensemble modeling approach that would integrate alternative base learner disturbance maps in a way that maximizes accuracy across the range of ecosystems and disturbance processes. Ensemble modeling combines map outputs from each of the base learners to produce an integrated “best available” map covering multiple disturbance processes and cover types. Every pixel is represented by a probability of membership for different landscape changes. This approach provides a framework for intelligently assimilating change maps with varying strengths for specific situations into a comprehensive change product. Identification of change in terms of a probability value, instead of a single discrete class, also allows easy adjustment of the map to match independently collected change estimates.

A second pilot study focuses on implementation of the ensemble modeling framework into a prototype production stream. The focus is on the Northwest Forest Plan (NWFP) area of Washington, Oregon, and California. The objective is to harmonize the diverse running environments of the different base learner algorithms into a common framework leading to operational processing for the United States.

Project Contact: 

Principal Investigators:
Warren B. Cohen - USFS Pacific Northwest Research Station
Zhiqiang Yang - Oregon State University

Tom Loveland - USGS EROS
Dan Steinwand - USGS EROS
Jim Vogelmann - USGS EROS
Steve Stehman - SUNY ESF
Chengquan Huang - University of Maryland
Matthew Hansen - University of Maryland
Robert E Kennedy - Oregon State University
Curtis Woodcock - Boston University
Zhe Zhu - Boston University
Brian Schwind - USFS RSAC
Kevin Megown - USFS RSAC