The basic idea is simple. Each person uses a certain area to live in and employs a certain amount of light to illuminate it. The more people there are in a city, the larger the area of the city and the brighter the lights. Thus, measuring the area or total brightness of a city's lights will give an estimate of the population of the city. Add up all the city lights in a nation, and you have an estimate of the population of that nation. The area and intensity of city lights are also a measure of the industrial capacity of a nation. In general, the more industrialized a nation is, the more energy it uses per person and the greater the intensity and number of lights it shines into the night. Imagine the brightness seen from space of a typical American town complete with street lights, auto lights, mall parking lights, industrial plant lights, and other light sources, compared with the brightness of an African community of the same population that depends on only a few camp fires for light. By measuring the light output of a city or a nation you can get an estimate of the population, energy use, and economic activity.

You may wonder why you should bother to analyze satellite images for information you can easily obtain from a library. The answer is that the information libraries have about other nations may not be correct. Do you suppose that the military and industrial statistics supplied by one nation to another are always accurate? Not necessarily, especially if the nation supplying the information has reason to safeguard information about its industrial or military capacity. Too often we must rely on a single source, one that may be inaccurate because it is mistaken, outdated, or purposely misrepresented. On the other hand, information compiled from multiple sources has a greater degree of reliability. It is common practice for industrialized nations to use multiple sources of data-collecting for surveillance purposes.

How to Use Imaging Data
In this activity, you will measure

1. the lit area, and/or
2. the total light output ("integrated density") of several individual cities, states, or nations.

However, this is only half of the information you need, because your measurements will be in square pixels or total brightness, not in number of people or dollars. To make the conversion, you have to use other sources such as the Internet or your school or local library to find the population, energy output, or economic activity of each of the cities, states, or nations whose light output you measured. A common measure of economic activity is the Gross Domestic Product, or GDP. Information on population, energy use, and GDP can be found in almanacs or encyclopedias. Another good source is the CIA World Fact Book, updated each year. Detailed population and other demographic information can be found for the United States in printed or online publications of the U.S. Census Bureau. Detailed global population information can be obtained at CIESIN. Details of energy use around the world can be found at the Energy Information Administration.

Realizing that the information you have obtained from these sources may not be completely reliable, you must nevertheless start with it to find patterns in the relationships between it and the city lights seen from space. Then, armed with both the image measurements and the library data for the cities or nations of a given image, you can graph the sets of data in pairs. Your axes may be any pairing of measured quantities and library data such as population versus lit area; energy output versus lit area; or, GDP versus integrated density. The points on the graph will be cities, states, or nations from a single image for which you have both measurements and library data. The relationships you find will most likely be linear (the graph will be a straight line) because, on the average, each person in a given society uses roughly the same amount of lighting; on average, the same percentage of energy output in a given society is used for lighting; and, on average, the GDP of a given society is directly proportional to its lighting output.

If you made your measurements accurately, then any points falling far from the linear relation indicate that either the population, energy use, or GDP that you looked up is inaccurate or that something very unusual is happening in the cities or nations represented by those anomalous points. In either case, you will want to investigate further to see what is really going on.

Once you have made a graph for cities or nations of known population, etc., in a particular image, you can use the graph to estimate the population, etc., of cities or nations in the same image for which you have only the lit area or integrated density. For example, if you have plotted lit area of cities on the horizontal axis and population on the vertical, simply mark the measured area of a city for which you do not have the population on the horizontal axis and move vertically up to the graphed points (i.e., up to those points that represent cities with known population and lit area that you have already plotted). Then move horizontally to the population axis to get an estimated population for the city.

In this activity, we will use images from the U.S. Defense Meteorological Satellite Program (DMSP). These images often show features unavailable on any other type of satellite image because of the very high sensitivity of the instruments and because one of the spectral bands used in the DMSP extends from the visible all the way into the near infrared (0.40-1.10 microns). This combination of sensitivity and spectral range enables DMSP images to record faint light sources such as moonlit clouds, forest or prairie fires, lightning, aurorae--and city lights.

Using NIH Image The basic procedures for measuring lit areas or integrated densities are the same for any image or area within an image. (If you are uncertain of the meaning of any of the following terms, go to the ETE Remote Sensing Activities or look them up in an NIH Image Users Guide, which you can download from the NIH Image Web site.)

1. Download the TIFF version of the image you want to work with.
2. In NIH Image, select Edit/Invert to make a "negative" of the image. Doing so insures that the Integrated Density will be calculated correctly.
3. Select Analyze/Options to get the Options dialog box. Select Area, Mean Density, Integrated Density, and anything else you think might be useful by clicking in the adjacent box (an"X" in the box means it is selected). "Area" simply measures the area of pixels selected in square pixels. "Mean Density" gives the average DN (Digital Number) in each of the selected pixels. "Integrated Density" = Area x (Mean Density - Background). The Background is the DN value of the "noise" in the image. The noise is computed from the lowest DN values of the pixels you selected. This is not easy to get right with DMSP images. Try a few different selections, and plot the results to see which gives the straightest graphed line.
4. Select Density Slice and use the LUT tool to highlight the lit areas of interest in red.
5. Use the Freehand Selection tool to carefully outline the geographical area of interest such as South Korea.
6. Select Analyze/Measure and then Analyze/Show Results to get the Results Table, which shows the area of the highlighted pixels within the selected area.

Repeat steps (5) and (6) to find the lit area or integrated density for several other cities, states, or nations in the same image. Then graph your results. Avoid using data from two different images on one graph since each DMSP image has a unique sensitivity setting, or "gain."

Images To Try Out Making accurate measurements on DMSP images can be challenging. As with any real data, you need to be aware of several factors that may affect the image and which you should take into account to make the most accurate measurements possible. You can practice on the following set of images to perfect your technique before you tackle the Korea image. The practice set illustrates different factors you should be aware of. In addition, each image shows an interesting part of the world that you may want to investigate further.

Eastern U.S. (download USEast.tif). This image shows city lights and clouds in a section of the eastern United States. This is a good practice image to compare population data from individual states or cities with measured lit areas or integrated densities. Note the effect of the clouds on city brightness and definition. You can also work with a cloudless image of the entire U.S. if you feel adventurous. You can download a half- size version of the image (US Lights, 0.6 test) or go to the DMSP site to get the full-size image. To measure state populations, you will need to download an overlay (USoverlay.pic) of state boundary outlines. Scale and then Paste the boundary image onto the DMSP image. Unfortunately, the projection of the DMSP image is not exactly the same as the boundary image, so different sections of the country will have to be fit separately. Use the lights along different coastlines as reference points.

South Eastern Europe (download SEEurope.tif). This image shows an interesting mix of industrialized and developing nations including Italy, Austria, Bosnia, Albania, and Romania. Use this image to see differences in light output between industrialized, agricultural, and war-torn countries. (Note the slight mismatch between the locations of the coastlines and national borders and the city lights. The process of adding the image's coastlines is automated and not always accurate.)

The Near East (download Near East.tif). Here is another image with an interesting group of developing nations. The image also has some very bright lights associated with the extensive petroleum industry around the Persian Gulf. You can compare the distribution of lights with a population map of the area (download Near East Pop.tif).

Korean Images Now you are ready to take on the real thing! To measure the light output of the nations of North and South Korea, you can use the image on the Situation page of this module (download TIFF), which includes coastlines and national boundaries. However, you will achieve greater accuracy for coastal cities by using a version of the same image which does not have the coastline and boundary overlay (download TIFF, download PICT). This will be a challenge because you will need to Scale and Rotate this image and overlay it onto a Korean City Locator Map (download TIFF, download PICT) to identify different cities in the image. After a few try's, you will soon recognize how the pattern of lights fits the pattern of the cities. Once you have identified the lights of important cities and measured their light output, you can compare your measurements to the populations provided in the tables of Korean cities in the Korea Today/People section. (If you do not know how to do an overlay or use the Rotate and Scale commands in NIH Image, you might try looking at the Overlay Activity in the Remote Sensing Activities.)

[ Activity One ] [ Activity Two ] [ Activity Three ]
[ References ] [ PBL Model ]

[ Home ] [ Teacher Pages ] [ Modules & Activities ]