Remote Sensing for Restoration and Resilience

By Abby Smith on October 5, 2021

Upstream Tech’s pilot with the USDA Forest Service shows how remote sensing can help government agencies more efficiently track restoration outcomes

Tonto national forest Tonto National Forest. (photo credit: Sam Goodgame, Unsplash)

When ecosystems are degraded or damaged, human intervention is sometimes necessary to restore them to a resilient and productive state. As technological advances in Earth observation transform the scale and pace of monitoring, they also bring new understanding to the long-term effects of environmental restoration projects.

The efficiencies offered by remote sensing technologies will be critical for federal agencies as they implement the Biden administration’s “America the Beautiful” initiative, which seeks to support restoration efforts and conserve 30% of U.S. land over the next decade. The accessible and timely insights offered by remotely sensed data can enhance oversight of public lands over vast areas and better illustrate restoration outcomes, ensuring that limited conservation budgets are spent most effectively.

The USDA Forest Service saw a clear opportunity to leverage remotely sensed data to provide insights to inform stewardship and streamline monitoring. This past year, we conducted a pilot project with them using Lens to assess long-term restoration outcomes. Upstream Tech developed Lens to support conservation monitoring by providing remotely sensed data and analytical tools in an easy-to-use web interface.

We found that the frequent, cost-effective snapshots of changing landscapes provided by publicly available truecolor and vegetation remote data in Lens can augment on-the-ground data collection. Lens’ online platform also enabled teams to collaboratively view and analyze imagery. This points to a significant opportunity for the Forest Service, as well as other government agencies and conservation organizations, to use remotely sensed data to efficiently conduct long-term observation of restoration and management work outcomes across vast landscapes.

Read on for a summary of our project and findings, and check out our full Remote Sensing Analysis & Lens Pilot Study report here.

A Landscape-Scale Need

The Forest Service established the Watershed Condition Framework (WCF) in 2011 to improve the health of watersheds on national forests and grasslands. This initiative draws attention to the importance of protecting these special places and provides a protocol to prioritize and evaluate the success of restoration interventions such as tree planting, regeneration, or removal. As WCF’s program managers work across large regions, they saw remote sensing as an opportunity to more efficiently assess whether their management efforts were having their intended outcomes.

WCF region 3 map The Forest Service’s 2011 Watershed Condition Classification map of Region 3 highlights that a significant proportion of watersheds within national forests in the Southwest are impaired or functioning at risk.

We worked in collaboration with the Region 3 WCF team in New Mexico and Arizona to evaluate the opportunities and limitations of using publicly-available remotely sensed data for restoration evaluation and monitoring. The Forest Service piloted Lens to visualize ecological data and public domain imagery to highlight pre- and post-restoration patterns across the region.

To do this, we partnered with WCF program managers and hydrologists to identify study and control sites on national forests and grasslands. We uploaded boundaries for these sites into the Forest Service’s online Lens portfolio, where our team members and Forest Service staff could collaboratively view, analyze, and make notes on the sites. Lens users are also able to add their own layers to Lens for further analysis, such as flow lines, roads, or fire perimeters.

Restoring the Range

Working closely with the WCF program manager, we selected the Woolhouse Grassland in eastern Arizona to highlight rangeland restoration. In this area, the encroachment of juniper trees and shrubs had been having negative impacts on the grassland ecosystem’s ability to sustain forage for grazing animals. The Forest Service used bulldozers to mechanically remove juniper trees and shrubs across a 915-acre study site over several months in 2015.

Woolhouse site Lens displays USDA National Agriculture Inventory Program (NAIP) aerial flyover data of the Woolhouse study site at 1-meter spatial resolution. This truecolor image displays ground conditions in a natural color palette, similar to what we can observe with our own eyes. It was captured on June 14, 2015, when the western side of the site had already been cleared of juniper trees. (Includes copyrighted material 2015 Upstream Tech and USDA NAIP. All rights reserved)

Their aim was to restore the grazing capacity of the site by removing the roots of unwanted plants, ideally preventing them from reestablishing. However, the challenge remained for the Forest Service to effectively monitor the results of their work across this large area over the following years. WCF program managers saw remote sensing as an opportunity to efficiently study the sites over time and ensure that their desired management outcome would persist long-term.

Woolhouse site infrared comparison A closer look at the tree and shrub removal on the Woolhouse study site, using Lens’ near-infrared layer. Plants reflect near-infrared and green light, while absorbing red through the process of photosynthesis. Trees and shrubs in this color infrared layer are easy to distinguish by their reddish color. (Includes copyrighted material 2013 & 2015 Upstream Tech and USDA NAIP. All rights reserved)

Hyperspectral sensors detect information about how plants are growing, which can be used to complement the changes visible in truecolor imagery. Lens’ vegetation layer measures vegetation vigor using the Normalized Difference Vegetation Index (NDVI), which indicates photosynthetic activity based on red and near-infrared wavelengths that satellite-based sensors are able to detect. NDVI is a number with higher values indicating greener and denser vegetation and low values signalling low or little vegetation growth. This data provides a reliable way to evaluate vegetation health and changes over time, which makes it an ideal tool for the Forest Service to track the outcomes of their vegetation management work.

The series below shows USDA NAIP aerial images of the Woolhouse study site in the Forest Service’s Lens portfolio, with truecolor imagery on the left and Lens’ vegetation layer on the right. The top row is from June 2, 2013, before any restoration activity; the second row is from June 14, 2015, as mechanical tree and shrub removal was underway; and the third row is from June 22, 2017. Cattle grazed in this area consistently pre- and post-vegetation removal.

Woolhouse site NDVI (Includes copyrighted material 2013, 2015, & 2017 Upstream Tech and USDA NAIP. All rights reserved)

The large areas of white in the 2015 and 2017 NDVI imagery indicate significantly lower vegetation vigor, as compared to the uniform yellow and green visible in the pre-removal 2013 imagery. This clear indication that the removed vegetation did not reestablish helped the Forest Service confirm that their intervention continues to have their desired effect.

Digging Deeper into the Data

To further understand vegetation growth trends over the past several years, we also analyzed time series data on vegetation vigor. A control site in the Woolhouse Grassland was selected where encroachment of trees and shrubs was visible but they had not been removed. The control site was in close proximity to and shared similar soil and vegetation characteristics with the study site. We used NDVI values from the European Space Agency’s Sentinel-2 satellites, NASA’s Landsat-8 satellite, and USDA NAIP aerial sensors within Lens to compare the study and control sites and determine the timing and extent of any differences which might show impacts from the restoration efforts.

The graph below shows average NDVI, or vegetation vigor, values derived from the Sentinel-2 satellite sensors, with data for the study site—where tree and shrub removal occurred in 2015—in red and control site in blue. These data provide granular detail into the effects of restoration. We can see that NDVI at the study site is clearly lower than at the control site, likely due to the lack of trees on the study site after removal.

Woolhouse NDVI comparison chart

Data from this project further confirm the enduring impact of the Forest Service’s intervention to limit juniper tree and shrub regrowth, as vegetation indices have been consistently lower in the study site compared to the control site over the past six years.

A New Way to See Change

Human intervention is sometimes needed to advance the recovery or restore the function of ecosystems that wildlife and human communities depend on. Upstream Tech’s pilot with the Forest Service highlights the promise of remote sensing technology to aid stewardship and inform future actions by providing more timely and cost-effective information about how these important landscapes are changing. We’re excited to explore how these kinds of insights can further benefit the work of government agencies, enhance policy-making, and ultimately ensure that these lands continue to thrive into the future.

Interested in learning more about how Lens might work for your organization? We’d like to hear from you! Please send any questions or thoughts to lens@upstream.tech.