Measuring our collective imprint on Earth

Google Earth
Google Earth and Earth Engine
5 min readFeb 8, 2019

By David M. Theobald, PhD, Senior Scientist at Conservation Science Partners

The “Earth at Night”, NASA.

Visual information — art, photos, and imagery — encourages deep reflection, fuels our science, and empowers social change. One such seminal image is a 1968 photo of the Earth rising above the moon, which has inspired much of our modern environmental conscience: “The vast loneliness is awe-inspiring and it makes you realize just what you have back there on Earth.” — Jim Lowell, NASA Astronaut. Another iconic image, the so-called “Earth at Night” (shown above), was created by synthesizing satellite images that captured light emitted at night across the globe. With the recent explosion of remotely-sensed (e.g., satellite) data and powerful computational platforms becoming available, one might wonder — what other images would evoke and further stoke action towards the conservation of all Earth’s inhabitants?

Central to recent discussions, policies, and global targets that underlie efforts to conserve biodiversity and related ecosystem goods and services are global biodiversity indicators. These important indicators are designed to monitor the pulse of the natural world. One of the core indicators focuses on quantifying patterns and changes in land use — to better understand where lands have been protected as parks and reserves, where lands are highly modified by humans, and where (and which) protective or restorative actions would make the most meaningful contribution to a resilient conservation network. These biodiversity indicators are key to many global initiatives, such as the Convention on Biological Diversity, Aichi Biodiversity Targets, UN Sustainable Development Goals and Half-Earth project.

To address our need for basic information about our effects on global ecosystems, I recently co-authored an article in Global Conservation Biology with colleagues from The Nature Conservancy, where we argued that better understanding of the gradient of human activities provides deeper insight into the nature, extent, pattern, and impact of human activities on our lands. A key product of our work is a map of global human modification. Building on an established approach applied to the United States, we quantified the degree of human modification (HM), which is a measure of both the spatial extent (i.e. the footprint) and the intensity of land uses. The foundation of this formulation is a simple, direct, and meaningful measure that ranges from 0–1, which provides a comprehensive, consistent, and current map of the proportion of land modified by humans. HM integrates a variety of anthropogenic stressors such as human settlement, agriculture, transportation, mining, energy production, and electrical infrastructure.

A map of the human modification that largely reflects the year 2016 (black=low, yellow/white=high), from Kennedy et al. 2019.

Compelling as this global HM map is, conservation action more often occurs at national, regional, or even landscape-scales. To this end, we have also provided similar information and analyses on land modification, for example, to better understand how to achieve climate connectivity in a fragmented US, to evaluate the patterns and trends of natural land loss in the western US (Disappearing West), and to inform conservation planning for the Green River Basin that straddles the states of Colorado, Utah, and Wyoming, USA. Moreover, providing trans-boundary information, especially across national borders, is useful to fill critical knowledge gaps, because the natural world — including plants and animals, and natural processes such as fire and climate — doesn’t recognize political boundaries.

A map of the human modification for North America (black=low, white=high).

The HM dataset will be invaluable for a variety of analyses, such as those that have leveraged similar maps to understand the extinction risk of terrestrial mammals due to habitat fragmentation or to complement existing efforts to protect the last of the “wild”.

Here I provide just a few examples of ways to leverage the HM dataset within the Google Earth Engine platform. And, to unleash even more of its potential, the HM dataset is freely and publicly available directly through the Data Catalog. Explore the HM dataset in Earth Engine.

To quantify the differences in human modification within a protected area itself, as expressed by the categories of protection levels by the IUCN World Data on Protected Areas (WDPA), I generated the graph below that compares the distribution of human modification for different categories of land protection. As expected, the more stringent management categories, such as strict nature reserves (Ia), wilderness areas (Ib), and national parks (II), have lower levels of human modification relative to other management categories such as natural monuments (III) and habitat/species management areas (IV). Generating the distribution of values within each management category helps to quantify these differences — both between protected and unprotected areas across the terrestrial world in general.

How do estimates of human modification vary by management categories of the world’s protected areas? This cumulative distribution function graph shows the distribution of human modification values from 0 (low) to 1 (high) for each of six IUCN management categories: such as strict nature reserves (IUCN_Ia), wilderness areas(IUCN_Ib), national parks (IUCN_II), natural monuments (IUCN_III), habitat/species management areas(IUCN_IV), and All (both formally and not formally protected lands).

Another example use of the HM is to update a measure used to identify which of the dozen or so global biomes are most at risk, called the Conservation Risk Index (CRI). For each ecoregion globally, I calculated the average proportion modified (i.e. the mean HM value) and the proportion of protected lands, and then calculated the CRI as the modified to protected ratio.

The Conservation Risk Index (CRI) values mapped for a portion of Europe-Asia-Africa. The CRI is calculated for each ecoregion as the average human modification divided by the proportion of protected lands from the WDPA (2017). Low CRI values are shown in blue (e.g., highlands of Turkey) and high CRI values in yellow (e.g., lowlands of India).

I then summarized the ecoregions to calculate a biome-level CRI value, and ranked them by their highest to lowest values (below). These preliminary results show that tropical & subtropical coniferous forests and temperate grasslands, savannas & shrublands are among the biomes most-at-risk, while boreal forests & taiga and tundra are among the biomes least-at-risk. These biome-level and ecoregional-level findings can help direct strategic conservation initiatives, though needs to be complemented with higher resolution and locally-relevant data and interpretation.

Biomes are ordered by their Conservation Risk Index, calculated as the ratio of the average percent of human modification to the percent protected as an index of relative risk of biome-wide biodiversity loss.

In this short communique, I’ve offered just a few ways to harness the computational platform and vast catalog of remotely-sensed data provided by Google Earth Engine, to better understand and communicate our effects on Earth.

What beautiful and powerful images will you create, to propel deeper understanding and stronger connections within our world?

--

--