Conifer Seedling Mapping in Postfire Landscapes using UAS Data

Screenshot%202026-03-06%20093448.png

We map conifer seedlings together with other co-occuring vegetation in a post-wildfire landscape in Colorado Front Range using UAS image derived parameters. This repository shares the actionable visuals, a concise briefs, and shareable codes used in the process.

Project summary

This repository provides a step by step workflow to derive canopy height models, vegetation indices, texture indices, and a random forest based classification workflow to identify conifer seedlings from the co-occuring vegetation classes such as shrubs, deciduos trees, mature conifer trees, and standing dead trees in a postfire environment.

Our product

A random forest based classification using vegetation indices, canopy height and rumple to identify vegetation as small as 30 cm in height. This highlights the use of high resolution UAS based red-green-and Blue only images for forest restoration, recovery, and resiliency needs.

Screenshot%202026-03-06%20102237.png

Why this matters

In the western US and most of the global dry forest landscapes postfire recovery is studied in local scales and are studying site specific questions. We believe that with current high frequency, high intensity wildfires in the dry forest systems, there need a standard way of understanding how these postfire landscapes evolves and how intentional managemnet needed to be supplied. To make these descitions, it is importamt to know the mechanisms on postfire seedling regeneration and the drivers impacting. This study provides a low cost, highly efficient workflow to identify conifer seedlings. Stay tuned for the study about seedling establishment probability and drivers.

Team