App. of High-res. WorldView-3 Satellite Imagery to Distinguish Limber Pine in RMNP, L. Sindewald

Title: Application of High-resolution WorldView-3 Satellite Imagery to Distinguish Limber Pine in RMNP
Speaker: Laurel Sindewald
Date: December 20, 2022

Speaker Bio:

Laurel Sindewald is a PhD candidate in Diana Tomback’s Forest Ecology lab at CU Denver where she is completing several research projects on limber pine at treeline in Rocky Mountain National Park. Her first paper, published in Forests in 2020, described the structure and composition of communities in RMNP where limber pine is the dominant conifer. She is working on a follow up to that study comparing the structure and composition of randomly selected treeline communities in the park. She is also estimating limber pine seed viability across high elevations, comparing the number and proportion of viable seeds produced by krummholz limber pine to those produced by subalpine limber pine in RMNP. She received funding from RMNP in 2021 and 2022 to raise any seedlings resulting from this viability research for restoration planting in the Cameron Peak Fire burn area. Lastly, she is developing remote sensing applications for species classification at treeline—the subject of today’s webinar.

Abstract:

Limber pine (Pinus flexilis) is an ecologically important conifer in Rocky Mountain National Park (RMNP) and across much of the mountain west in the United States and Canada. While limber pine can occur across a broad elevational range, in RMNP it is commonly found at high elevations under windy, exposed, and droughty conditions where it may predominate over less hardy conifers. In RMNP, bioclimatic envelope models predict it will move upwards in elevation as the climate warms, but its treeline distribution is not fully known. We are using a combination of hyperspectral field spectroradiometer data and high-resolution satellite imagery to remotely distinguish limber pine from other major treeline vegetation. In this webinar, I will discuss (1) field methods for collecting a hyperspectral groundtruth (especially in regions where drones are not allowed), (2) regions of the electromagnetic spectrum (from 350 to 2450 nm) and spectral indices most useful for separating species in this system, (3) results of a convolutional neural network (CNN) classification of WorldView-3 imagery, and (4) recommendations for remote sensing applications for the conservation and management of limber pine and related species.