How Can New Remote Sensing Technologies Help Evaluate the Effectiveness of Resource Conservation Measures?

By Amanda Stahl and Alexander Fremier, Washington State University

Grassy area with GPS equipment on a tripod in the foreground and trees along a riparian corridor in the background.
Conserving riparian areas means a small footprint can contribute to protecting a county’s Critical Areas and mitigate the effects of climate change. Photo: Amanda Stahl.

Washington State is taking steps to foster environmental stewardship in agriculture using an alternative approach to direct regulatory oversight. Twenty-seven counties in Washington have opted into the Voluntary Stewardship Program (VSP), which requires them to self-assess (with state oversight) whether voluntary management actions are maintaining or enhancing Critical Areas. Critical Areas include wetlands, fish and wildlife habitat conservation areas, critical aquifer recharge areas, frequently flooded areas, and geologically hazardous areas. Most counties cite riparian conservation measures as a strategy to maintain or enhance at least one type of Critical Area. Riparian conservation measures, like planting or allowing natural vegetation to grow, can also address the impacts of climate change, providing shade to cool water in the stream, improving habitat for species stressed by climate change, and possibly helping moderate extremes in moisture availability year-round. Conserving small land areas can thus have a large impact for mitigating the effects of climate change. The question is, how can we quickly determine if these measures are working, and meeting the goals of the VSP?

Through monitoring, of course. Conservation Districts assist counties with monitoring, commonly involving site visits for each project. Due to limited budgets and logistical challenges, a ground-based approach cannot realistically provide the landscape-level monitoring needed for the VSP. Monitoring plans thus refer to aerial and satellite imagery as important tools for detecting change (or no change) over time. Yet, many local governments and Conservation Districts have minimal capacity to perform image analysis in-house. And existing products that detect land cover change from imagery are typically either too coarse-grained (particularly for narrow riparian corridors) or do not apply to all regions of the state, so their use for reporting at the landscape scale remains limited.

Our research team at Washington State University partnered with the Palouse Conservation District to investigate potential options for leveraging advances in UAV (drone) technology, freely available satellite images and cloud computing to add remote sensing techniques to the conservation practitioners’ toolbox. We completed a pilot study in drylands in Whitman County and vicinity, where riparian restoration is increasingly adopted as an off-farm-field strategy to protect critical areas, reduce climate change impacts and improve water quality conditions. We aimed to develop steps that could be easily repeated and adapted to fit the wide variety of riparian ecosystems across Washington State or beyond to other ecosystems and locations, including agricultural lands.

First, we flew a drone mounted with cameras at nine sites with different types of streamside settings, thanks to a group of producers who volunteered to participate in the study. By taking 100 or more aerial photographs from different angles with substantial overlap, we were able to use photogrammetry to construct georeferenced 3-D renderings of the plant canopy and stream, as well as use different parts of the reflected light spectrum (Figure 1) to gauge plant health. We are sharing what we are learning with the Palouse and other Conservation Districts as well as conservation practitioners across Washington to help identify opportunities to incorporate drone-based data collection into their monitoring programs.

Aerial image showing a riparian corridor and a road along it, viewed from above. The image is repeated, with false colors highlighting the riparian corridor.
Figure 1. This image mosaic illustrates our ability to identify and distinguish among types of green riparian vegetation from above. It was generated from a drone-mounted camera flown at about 50-meter altitude (with 10-20-cm ground resolution) over a riparian area in the Washington State University Hudson Biological Reserve. Images collected by Amanda Stahl.

Second, we are using the Google Earth Engine, a cloud-computing platform that catalogues a wide variety of remotely sensed imagery and makes it easily accessible to the trained user. Once the code (in JavaScript) has been programmed, the user can analyze the images and extract data from them without having to find, download, and painstakingly process large files. In other words, what would have once taken days to weeks takes minutes to produce, requiring no computational power beyond a standard computer with an internet browser. The example map (Figure 2) shows how we can look at patterns through time and automatically map categories onto the land. For example, areas mapped in green are “always green”, meaning the vegetation in that spot is green in the late summer of all years studied or “sometimes green, sometimes brown”, meaning that we detect change in that spot that could be due to natural differences in the pattern of precipitation from year to year, or there was a change in practices like a restoration project. The maps we generated have already contributed to VSP reporting for Whitman, Garfield andWalla Walla counties. In the future, these maps could help inform where it would be most worthwhile to invest in on-the-ground monitoring.

Map with upland vegetation in yellow, and stringers of green and purple showing other vegetation categories, such as riparian.
Figure 2. Example output map for Whitman County from an analysis of Sentinel-2 satellite imagery for discriminating vegetation types based on season phenology in Google Earth Engine. Image sources accessed from the European Space Agency and Google Earth, prepared by Amanda Stahl.

An added benefit of using the cloud is sharing data and analysis. We can then train others on the programming, sharing our work, or we can share the outputs through a user-friendly Earth Engine App. The App can be made publicly available and can enable any user to interact with the data. For example, personnel preparing monitoring reports at Conservation Districts or County offices will be able to change the dates or location of the analysis. This means that anyone can access big data, whether to track changes before, during, and after a restoration project or to anonymously report riparian improvements for the VSP.

We found an untapped potential to use emerging remote sensing technologies to enhance the efficiency and effectiveness of natural vegetation monitoring in agricultural settings. We are continuing to collaborate with Conservation Districts across the state to tailor these tools to their needs and share the code we co-develop. Harnessing these capabilities may ultimately enable researchers and practitioners to build stronger links between conservation measures and the broader outcomes of mitigating climate change impacts, maximize the benefit of limited on-the-ground monitoring, and contribute to ensuring agricultural sustainability.


2 comments on "How Can New Remote Sensing Technologies Help Evaluate the Effectiveness of Resource Conservation Measures?"
  1. Thanks for the article. Quick correction: shade does not cool water. It inhibits warming.

Leave a Reply

Your email address will not be published. Required fields are marked *