If you would like more information, or you would like a particular area screened for late-successional/old growth forest using LiDAR, contact John Hagan at [email protected]

Late-successional/Old-growth (LSOG) Project

PROJECT GOAL: To map late-successional and old-growth forest in the unorganized territories of Maine so that timberland owners and conservation groups can work to conserve this special age-class of forest.

The 10-million acres of unorganized townships of Maine (Fig. 1) are unique in the entire eastern U.S.  The area is nearly 100% forested (except for logging roads and water bodies) and lacks either agricultural or ex-urban development characteristic of most of the eastern U.S.  At night, the area is an enormous dark “hole” on the U.S. map.  Although mostly managed for timber production, the area is a haven for wildlife species, from migratory birds to lynx and moose.  Its rivers still provide habitat for many aquatic species, such as cold-water brook trout and the endangered Atlantic salmon. In short, the unorganized townships of Maine are a rare gem of global biological significance.

However, because over 90% of these 10-million acres are privately owned and managed by timber companies, it can be financially difficult to retain late-successional forest in the mix of forest types and age classes.  In the commercial forest of Maine, generally stands older than 60 or 80 years of age are past their financial prime.  Yet late-successional forest (100-200 years old) and old-growth forest (>200 years old) develop ecological characteristics, such as such as larger-diameter trees suitable for raptor nesting, den sites, cavities for woodpeckers, and substrate for many late-successional lichens and mosses, that are not common in optimally managed commercial forest.  In old forests, large old trees eventually die and fall over, and can last for a half-century or more as a log on the forest floor simply because of their massive size, creating important habitat for other species.

Until now, there was no way to cost-effectively assess how much late-successional forest remains in the unorganized townships, or where it occurs.  Timber companies have a general sense based on the proprietary stand maps they use for tracking forest types and age classes.  But this information is not publicly available.  In the process of doing our “30-Year Bird Study” in 2021-2022, we made an unexpected discovery.  LiDAR (light detection and ranging) appears to be an almost foolproof tool for identifying stands and tracts of late-successional (LS) forest, with inordinate precision (to the acre).


How LiDAR Works

LiDAR stands for Light Detection and Ranging. It involves firing a laser beam, usually from an airplane, toward the ground (Fig. 2). The laser beam is reflected off of anything it hits, both the trees and the ground below. Because the laser beam is moving at the speed of light, the difference in the time it takes the first return to get back to the plane and the last return to get back to the plane tells us how tall the tree is. That is, the time difference is converted to a very accurate tree height (within 10 cm). Some of the laser beams bounce off of tree branches. The end result is a massive digital “point cloud” that basically paints a 3-dimensional view of the forest. The usable resolution of the publicly available LiDAR is a 1 square meter– extremely fine resolution compared to satellite imagery (30×30 m, or 10x10m).

It’s this LiDAR image that tells us where most of the old forest is in the unorganized territories.

Fig. 1. The Unorganized townships of Maine. They are called “unorganized” because too few people live here to form any local government. Most of this 10-million-acre area is managed for timber and forest products.

Fig. 2. LiDAR (Light Detection and Ranging). A laser mounted in an airplane fires thousands of laser pulses at the ground per second. The laser reflected back to the plane sensor results in a massive 3-dimensional digital “point cloud” that represents the forest it is “seeing.”

Fig. 3. The image below is a “Canopy Height Model” generated from LiDAR data. The different colors represent differnt heights of trees. The oranges and yellows are shorter forest, harvested in the last 20 years or so. The blue/magenta color represents tall forest. The magenta “flecks” represent trees over 22 meters (72′) tall. Our ground truthing of this “blue/magenta LiDAR signature” in 2023 showed us that over 95% of the time, this signature indicates late-successional or old-growth forest. We got the same result all across the unorganized townships of Maine. One limitation is that this method does not reveal short, stunted old forest that you might find at high elevations or in low, black spruce bogs. But those areas are not as much at risk of harvesting. So, the LiDAR signature works well where it needs to work most– in the majority of the landscape where harvesting is likely to occur. The large square grid in the diagram represents 100-hectare units (247 acres).

Fig. 4. This is the exact same scene as Fig. 3., except the “blue/magenta signature” for LSOG forest has been converted to a more user-friendly map. Below, we generated a simple polgyon “LSOG Yes/No” map below. Whereever there is a magenta polygon, there is a tree at least 22 m tall. Small magenta polygons represent a single tall tree. Larger polygons represent LSOG stands. As you can see, the large blue/magenta stand on the north side of the Golden Road (the black wavy line across the center of Fig. 3) shows up as solid magenta on this map. Our ground truthing of this stand verified that it is a late-successional stand, with some parts even being closer to old-growth. One spruce we cored was 253 years old– it was a seedling when the first shot of the Revolutionary War was fired. For scale, the large square grid in the diagram represents 100-hectare units (247 acres).

Fig. 5. Below is an “LSOG magenta polyon map” we generated for a 1.2-million-acre area of interest to the Rangeley Lakes Land Trust. The magenta polygons indicate likely LSOG stands at least 10 acres in extent. We can easily screen large landscapes like this for LSOG with incredible precision. Ground truthing is always recommended, but our own field work indiciates the map will be accurate at least 95% of the time. With further tweaking of our models, we expect to improve the accuracy even more.

Project Team

John Hagan, Ph.D., Our Climate Common
Ben Shamgochian, Our Climate Common
Molly Lynch, Tufts University
Michael Reed, Ph.D., Tufts University
Dave Sandilands, U. Maine Wheatland Geospatial Lab
Andy Whitman, Maine Forest Service

LSOG Working Group

Kyle Burdick, Baskahegan Co.
Shawn Fraver, University of Maine
Jake Metzler, Forest Society of Maine
Shelby Perry, Northeast Wildnerness Trust
Justin Schlawin, Maine Inland Fisheries and Wildlife

Forest Landowner Partners

Seven Islands Land Co.
Huber Forest Resources
Appalachian Mountain Club
The Nature Conservancy
Baskahegan Co.
Irving Woodlands
Acadian Timber Corp.


The Dorr Foundation
Cooperative Forest Research Unit, U. Maine
Maine Timberlands Charitable Trust
EJK Foundation
The Betterment Fund
The Arboretum Fund of the Maine Community Foundation
An anonymous donor