Using Advanced Technology to Reach Our Conservation Goals

December 26, 2019, Department, by Karl Schrass

2020 January Conservation Advanced Tech Conservation Goals 410

As 2020 begins, we prepare to celebrate the 50th anniversary of both Earth Day and the Environmental Protection Agency (EPA). This offers a time to look back and recognize the strides we have made in conservation, as well as to look to the future and determine what tools and tactics we need to address today’s threats.

During the past 50 years, the policies and institutions established during the early 1970s have profoundly benefited the health of our environment and our communities. Despite these successes, today we face even more complex crises, such as climate change, species extinction and the pervasive impacts of plastic pollution, just to name a few. In many ways, we find ourselves, 50 years after the first Earth Day, at a similar inflection point and must act with urgency to improve the sustainability, resilience and health of our parks and the communities that they serve.

While today’s challenges are daunting and complex, we also are equipped with new tools to help us identify and implement creative and scalable solutions.

Living in a Computerized World
Throughout these five decades, nothing has changed the way we live every aspect of our lives more than computers. A modern smartphone is more than 3,000 times more powerful than the state-of-the-art microprocessor that was first developed in 1971. This quantum leap forward in computing power has created powerful new tools for understanding our world. Specifically, the ability to collect big data and analyze it using artificial intelligence (AI) can provide park and recreation professionals with novel insights and inform their decision making.

AI Mapping: 200 Million Images in 10 Minutes
In December 2016, the nonprofit Chesapeake Conservancy released one of the largest, high-resolution land-cover maps made in the United States. The map covers 100,000 square miles, including the Chesapeake Bay watershed and the surrounding area. Stretching from the Adirondack Mountains to Virginia Beach, it covers more than 200 cities and counties across the mid-Atlantic and Northeast. This map has resolution to one meter, making it 900 times more detailed than most land-cover maps of the United States. Using the previous land-cover map, more than 1 in 10 of Baltimore’s parks would have been too small to show up at all, while now you can pick out individual trees.

The initial development of this map used a combination of semi-automated computer models and human classification and took 30 staff members 10 months to complete. While 10 months to develop a data layer of this size and scale is an impressive feat, it would need to be much less costly in both time and money to be scalable. To this end, the Chesapeake Conservancy has been partnering with Microsoft and its AI for Earth initiative to apply the methods developed for the Chesapeake Bay to national data, using AI and a highly specialized computer chip. As a result of this collaboration, Microsoft has been able to analyze 200 million satellite images and produce a beta land-cover map for the entire United States in only 10 minutes and for a total cost of $42.

As technology leaders like Microsoft release tools and resources that help us get highly detailed and accurate data of our world, it is incumbent that park professionals understand how to use these tools and apply them to make smarter decisions in the creation, design and management of their parks.

Machine Listening: Deciphering the Noise
Located on the north shore of Long Island, New York, the Avalon Park and Preserve is a private park that provides more than 140 acres of forests and meadows for wildlife habitat and for people to enjoy. As part of a master planning effort at the park, Andropogon, a landscape architecture firm, took an innovative approach and used recording devices to capture the soundscape of the park. Deploying teams of youth citizen scientists with microphones to different locations throughout the park, they captured recordings of the ambient sounds at each site during different times of the day and in different seasons. After collecting this data, the landscape architects and their partners used machine listening and an AI technique for audio data to classify the sources of the sound on the recording as either animal, physical (wind, water) or manmade. This data was then combined with the results of a Bioblitz to identify those specific areas of the park that had the highest biodiversity and the times of day when the wildlife, mostly birds, was most active and vocal.

Using this data, the park management could ensure that programming designed to connect kids with nature was located and scheduled to ensure the most contact with the wildlife in the park. In addition, they used this soundscape data to rearrange potential user conflicts within the park — for example, to ensure that noisy park maintenance activities were not scheduled at the same time as activities that require quiet, such as yoga or birdwatching.

The capabilities of machine listening are rapidly evolving. The Cornell Lab of Ornithology, creators of the renowned citizen science tool eBird, have recently released a prototype of the new app BirdNET that identifies nearly 1,000 of the most common bird species just by listening to an audio sample. As these and other tools continue to evolve, they can be used to quickly, cheaply and accurately gather data on the biodiversity present across our country.

These are just two examples of how the latest developments in computer and data science have been applied to issues relevant to conservation. Every day, more data is being created. Understanding how we, as park professionals and conservationists, can analyze that data for insights will help us create and manage great parks that can continue to thrive in the future.

Karl Schrass is Director of Conservation at NRPA.