CHM

A Canopy Height Model (CHM) is a raster dataset where each pixel represents the height of vegetation or other above-ground objects above the ground surface. It is derived by subtracting a Digital Elevation Model (DEM, bare earth) from a Digital Surface Model (DSM, top of canopy): CHM = DSM − DEM. The result isolates the height of everything standing above the terrain — primarily trees, but also buildings, infrastructure, and any other elevated features.

CHMs are one of the most important derived products in forest remote sensing. They transform raw elevation data into ecologically meaningful information: tree height, canopy structure, gap distribution, and forest vertical stratification. Combined with allometric equations that relate tree height to biomass, CHMs enable large-area carbon stock estimation — a capability central to climate change monitoring, REDD+ verification, and national forest inventories. The advent of spaceborne LiDAR missions (ICESat-2, GEDI) and high-resolution stereo satellite imagery has made CHM production possible at continental and even global scales.

CHM creation requires two co-registered elevation surfaces:

The DEM is produced by filtering a LiDAR point cloud (or photogrammetric point cloud) to isolate ground returns, then interpolating these to a regular grid. The DSM uses first-return (highest) points from the same dataset. Subtracting the DEM from the DSM at each pixel yields the above-ground height.

From airborne LiDAR, CHMs achieve centimeter-level precision because both the DEM and DSM derive from the same point cloud with consistent georeferencing. Key quality factors include point density (higher density captures more canopy detail), pulse penetration (ability to reach the ground through canopy), and classification accuracy (correct separation of ground from non-ground points).

Spaceborne CHMs use different approaches. NASA's GEDI (2019–2023) fired full-waveform LiDAR pulses from the ISS, measuring the vertical distribution of energy returned from canopy and ground to derive canopy height at 25 m footprint resolution. ICESat-2's photon-counting approach detects individual reflected photons, reconstructing canopy top and ground elevations along orbital tracks. Meta and the World Resources Institute combined GEDI footprints with Sentinel-2 imagery and machine learning to produce the first wall-to-wall global canopy height map at 10 m resolution in 2024.

CHMs can also be derived from stereo satellite imagery (WorldView, Pléiades) by producing a photogrammetric DSM and subtracting an external DEM. However, optical stereo only captures the visible canopy surface — it cannot penetrate through to the ground — making the DEM source critical for accuracy.

CHM accuracy depends entirely on the quality of both input surfaces — errors in either the DEM or DSM propagate directly into canopy height estimates. Under dense tropical canopy, LiDAR pulses may not reach the ground at sufficient density, causing DEM errors and consequent CHM overestimation. Photogrammetric CHMs cannot penetrate canopy, so the DEM must come from an independent source (e.g., LiDAR), introducing potential misregistration between the DSM and DEM. Spaceborne GEDI measurements are sampled along orbital tracks with gaps between them, requiring interpolation or fusion with optical imagery for wall-to-wall coverage. Seasonal timing matters — deciduous forests yield very different CHMs in summer (full canopy) vs. winter (bare branches). CHMs represent a snapshot in time and can become outdated after disturbance events like logging, fire, or windstorms.

The concept of subtracting a ground model from a surface model to derive object heights has been practiced since the earliest days of LiDAR forestry research in the late 1990s and early 2000s. Researchers like Næsset (1997) and Means et al. (2000) demonstrated that LiDAR-derived canopy heights could predict forest inventory parameters with accuracies rivaling field measurements. NASA's ICESat mission (2003–2009) provided the first spaceborne canopy height measurements using laser altimetry, primarily over ice sheets but also used for forest applications. GEDI (launched 2018) was the first satellite mission specifically designed for measuring forest canopy height and structure. The Meta/WRI global canopy height map (2024) and ETH Zurich's Global Forest Canopy Height (2020) marked the transition from sample-based to wall-to-wall global mapping, enabled by combining spaceborne LiDAR training data with machine learning applied to Sentinel-2 imagery.