Our challenge and our solution

Photo Credit

USDA Forest Service photo by Wade Tinkham

A pre-treatment area on the Nez Perce National Forest as viewed from an Uncrewed Aerial System (UAS).

Traditionally, generating individual tree forest inventories from aerial point cloud data has required complex, multi-step processing routines that integrate various methodologies and require a high level of analytical expertise. Ultimately, this complexity raises the cost and limits the use of these data. To overcome these hurdles and enhance the integration of aerial point cloud data into forest and fuels management, a standardized platform for generating consistent single-tree inventory outputs is essential. The demand for such a solution is amplified by the increasing availability of open aerial point cloud data across large areas and affordable collection via uncrewed aerial systems (UAS) over smaller management units. 

What cloud2trees offers

Photo Credit

George Woolsey and Wade Tinkham

cloud2trees can ingest most aerial point clouds to produce near-census individual tree forest inventories

To improve aerial point cloud data integration in forest management, we present the cloud2trees R package, a streamlined data processing framework for generating consistent and scalable individual tree forest inventories. The cloud2trees package consolidates and builds upon a vast body of existing research about individual tree detection (ITD) and biophysical parameter extraction and modeling from aerial point clouds. The cloud2trees framework is intended to bridge the gap between disparate scientific disciplines and better connect researchers with each other and land management professionals. It recognizes the real-world challenge of acquiring the full range of technical skills needed to collect, analyze, and integrate the data and methodologies required for modern forest, fuels, and fire management. 

What is cloud2trees?

Photo Credit

Wade Tinkham

Data flow diagram for how raw point clouds are ingested and processed to create first order products including a digital terrain model, canopy height model, and individual trees as treetop point locations and crown polygons

The cloud2trees R package is designed as an end-to-end aerial point cloud processing tool that can take in raw point clouds and output spatial individual tree forest inventories with common forest and fuels inventory attributes. It is designed to ingest a variety of raw aerial point clouds, including manned aircraft laser scanning data (ALS or LiDAR) and UAS data derived from either structure-from-motion photogrammetry or laser scanning. Ultimately, the workflow produces a digital terrain model and a canopy height model at user-defined resolutions, a height-normalized point cloud for use in later analysis, and spatial datasets of individual trees including both tree-top point locations and crown polygons, both with each tree’s coordinates, height, and crown area.

cloud2trees in action

Photo Credit

Wade Tinkham

Extracted individual tree crowns from UAS data are pictured in the top left, superimposed on an aerial photograph of a forest

We are in the process of sharing cloud2trees with many different potential user groups who are putting it to work! To date, we have conducted several demonstration projects with multiple National Forests where we have used it to monitor forest treatments. We have also helped users process several thousands of acres of data for spatial forest treatment planning. The image at the left shows how we used cloud2trees to extract individual tree crown height from UAS data collected on the Santa Fe National Forest. We are continuing to share this tool broadly and plan to add more details and examples of use cases here in the future.