Many scientists make discoveries by comparing or summarizing
information, or data. A collection of data is called a dataset. A dataset
is organized in rows and columns of data (table 1). Datasets represent
information using numbers, and numbers may be used to represent words.

Often, scientists collect data themselves and create their own datasets. Scientists may also use datasets created by other scientists. An example of a dataset created by other scientists is the collection of daily weather measurements, such as daily air temperatures, for communities across the Nation.

The National Oceanic and Atmospheric Administration (NOAA), for
example, collects and electronically stores daily weather measurements for U.S. communities. Scientists interested in studying the relationship of weather to other variables might use this dataset. Anyone with Internet access can view and use these weather measurements, which are stored on computers in databases.

Computers have enabled scientists to create and use large datasets. As
computers become more powerful, scientists are able to analyze larger and larger datasets and to combine large datasets. Sometimes, scientists combine existing datasets with their own datasets. As more datasets are created and are available to many scientists, scientists can learn even more about our world.

Over time, natural environments may become degraded through human activities or natural disturbances. Land managers often work to restore these environments to a healthier condition. In 2010, the White House Council on Environmental Quality addressed watershed restoration (figure 1). The Council recommended that degraded watersheds be identified. The Council was interested in watersheds where restoration actions could be taken easily and quickly.

The Great Lakes Restoration Initiative (GLRI) was started in 2010. The goal of the GRLI is to protect and restore the largest system of freshwater lakes in the world. The Great Lakes contain about 20 percent of the world’s freshwater (figure 2).

The watersheds surrounding the Great Lakes contribute either to each lake’s good health or to a degraded condition. In particular, water near the shorelines, also called nearshore water, is likely to be degraded if nearby watersheds are unhealthy. Nonpoint sources of pollution in these watersheds may affect nearshore water quality. Nonpoint sources
of pollution cannot be easily identified because they do not come from a single known source, or point. Examples of nonpoint pollution sources include cities and agriculture.

The scientists in this study were interested in helping land managers identify watersheds that might be sending pollutants into Great Lakes nearshore waters.

Water quality in lakes is influenced by the characteristics of the watersheds that surround them. Lakeside watersheds affect water quality because water that runs off and through them enters the lake water. Water quality affects the health of economies that depend on water and water bodies.

The Great Lakes contribute much to the economies of eight U.S. States, tribal communities, and one Canadian province that border the lakes. Ontario is the Canadian province that borders the Great Lakes. Identify Ontario in figure 2. What do you notice about Ontario, compared with the States south of the Great Lakes?

The nearshore region is important to Great Lakes communities. The nearshore region is defined as the water area extending from the shoreline to 20–30 meters of water depth (figure 3). The nearshore region is important because it is used as a drinking water source, is used for recreation, and is an important aquatic ecosystem.

The Great Lakes nearshore waters’ health is dependent upon having healthy watersheds (figure 4). These watersheds sometimes suffer from nonpoint source pollution.

Nonpoint source pollution is pollution coming from a wide area, such as from agriculture or cities (figure 5).

In contrast, point source pollution can be traced to one particular source, such as a factory (figure 6).

The scientists in this study were interested in figuring out a way to identify Great Lakes watersheds that need ecosystem restoration. Ecosystem restoration involves land management to restore the ecosystem to an earlier, healthier condition. The scientists wanted to provide watershed managers with mathematical models that would predict nearshore water quality based on land use changes occurring in the watersheds. Models are a representation of something. Models can be made from materials, mathematics, or images. The scientists also wanted to be able to identify watersheds for which restoration could be done quickly and at a reasonable cost.

The scientists studied watersheds around Lake Superior and Lake Michigan (figure 7).

The scientists identified ecoregions within these watersheds (figure 8).

Ecoregions are areas that contain similar ecosystems. Ecoregions are identified by a different mix of natural vegetation as compared with the other ecoregions. The scientists used several existing databases to describe the conditions in each watershed (table 2). To learn more about existing databases, read “Thinking About Science”.

The scientists used water quality information collected at various places
near and within the watersheds (figure 11). Notice that much of the water quality information was collected in the nearshore region, near the coastline.

Some of the water-quality information was collected in streams within the watershed. This information had already been collected by State, tribal, and Federal agencies. The scientists used measures of phosphorus and turbidity as indicators of water quality. Water quality is lowered when phosphorus and turbidity increase.

The scientists used existing phosphorus and turbidity data collected within and near Great Lakes watersheds. The scientists retrieved information about the watersheds from other databases. They used all of this information to develop mathematical models on a computer. The models explored relationships between the amount of phosphorus and turbidity in the water and the watershed conditions shown in table 2.

The models enabled the scientists to discover which of the landscape conditions in table 2 were most closely related to increased phosphorus and increased turbidity. Once the scientists understood which landscape conditions were most closely related to water quality, they were able to identify watersheds that had the greatest likelihood of contributing to water quality problems. This finding included watersheds from which the scientists had no water quality data, but for which the landscape conditions were known.

The scientists used several different variables to help them understand water quality. The Vegetation Change Tracker (VCT) database enabled the scientists to explore the relationship between changing landscapes and water quality. This exploration was possible because the VCT database includes information about how forests change over time. The relationship between forest change and water quality varied across the watersheds studied.

The scientists found that when urban development and other landscape
conditions such as agriculture increased, water quality decreased (table 3).

If a forest had been disturbed in the past but was not recently disturbed, turbidity and phosphorus levels were lower than in areas with recent forest disturbance. In watersheds with more open water or wetlands, water quality was higher (figure 13).

The scientists’ models identified differences between water quality within
and near Lake Michigan and Lake Superior watersheds, between ecoregions, and between watersheds within ecoregions. The scientists, therefore, had confidence in their ability to identify which watersheds needed immediate restoration.

The computer models enabled the scientists to identify which watersheds needed immediate restoration. The models also enabled the scientists to estimate water quality within and near watersheds where they did not have water quality data. The scientists were also able to predict water quality within and near watersheds where the landscape was expected to change. With this information, watershed managers could respond to watersheds needing the most attention. Managers could also identify which watersheds would be most easily restored, and they could plan for the future as they expected the landscape to change.