The IESRE/IPST Sky Photography Project

This project is supported by the Institute for Earth Science Research and Education, Worcester, Pennsylvania, USA, and the Institute for the Promotion of Teaching of Science and Technology, Bangkok, Thailand.

Introduction

Whenever we want to find out about the weather, the first thing we do, even when we are indoors, is look at the sky. Is it blue and clear? Does it look "dirty" from air pollution? Is it gray and overcast? Are there cumulus or cumulonimbus clouds? Are clouds moving fast across the sky or hardly at all? It is clear from our experiences with sky-watching that the sky tells us a lot about the weather and our local environment. Is it possible to learn more by making quantitative measurements that complement our qualitative obserations? If we can do that, then we can turn the simple observations we use to plan our outdoor activities into real atmospheric science.

Sky photography with a digital camera allows us to build a permanent record of the state of the atmosphere. Any photograph of the sky provides useful information about the atmosphere when the photo was taken. But, if we photograph the sky over long periods of time, following an established protocol, these photographs can also provide quantitative information about the changing state of the atmosphere.

What do you need?

If you already have a digital camera, participating in this project will not cost anything! Here is all you need:

How much time will this project take?

It takes only a few minutes to step outside and photograph the sky. It may take a little longer if the scene you choose to photograph is not close to your home or school. This activity should be done only on days when you can see most of the sky above the horizon at the location you have chosen—it is OK for a few clouds to be present. Analyzing a photograph with ImageJ software (generating a density plot along a line drawn up from the horizon) and saving the results of that analysis will take no more than 10-15 minutes. Additional study of your results, such as using a spreadsheet to compare all data collected during a month, can take more time.

What are appropriate student ages for this project?

This project is suitable for any students mature enough to use a digital camera correctly—older elementary and young middle school students will certainly have no difficulty. The software is easy to use for this application, and the required processing can be done by any student with basic computer skills—using a mouse, naming and saving files, etc.

What will I learn from this project?

Sky photography is an excellent way to become more aware of what is happening in Earth's atmosphere, and how it reflects weather, climate, and the quality of your local environment. Especially if you are already collecting basic weather data, this project provides a visual record that supplements the measurements you are already making. If you are collecting pyranometer data (at one-minute intervals), sky photographs will help to explain the fluctuations in insolation you will usually see in those data. Like many projects that involve observations and recording data, this project will teach you how to organize and conduct an experiment in a way that makes it as simple as possible for others to understand and interpret what you have done. You will learn a lot about digital photography, too, especially if you are using a camera with manual settings.

This project serves as an introduction to the scientific applications of digital photography. It is also a great way to collaborate with students in other parts of your country and in other countries. Because the experiment protocol is very simple and requires only a digital camera and some free software, there are no cost constraints that will prevent schools around the world from participating.

Protocol for sky photography

Sky photography is easy to do with a digital camera, and the photographs are easy to analyze using the software referenced above. But, several important steps are required to convert sky photographs into a consistent long-term record that has both educational and scientific value. These steps include:
  1. Always photograph the same scene, and include a little land or water below the horizon, to track seasonal changes on the ground.
            In order to provide a consistent record of atmospheric conditions, it is important to choose a specific scene and always photograph that same scene.

  2. Photograph the scene at the same time of day, for example, sunset or solar noon.
            The sky looks different depending on the time of day, so it is important to photograph your scene always at the same time of day. During the day, local solar noon is a good choice. Another choice might be looking west at sunset. If your photographs include distant objects such as mountains or buildings, you can use the photos to provide a record of horizontal visibility through the atmosphere. But, in order to do that, you must consider the lighting conditions on that object. If the distant object, such as a distant mountain range, is sometimes lit by the sun and sometimes in shadow, you should choose one of these conditions and use it consistently so you can compare visibility from one day to another. You may want to photograph a scene twice during the day -- once near local solar noon, for analyzing sky conditions, and again when you are analyzing horizontal visibility. (See below for more information on measuring horizontal visibility through the atmosphere.)

  3. Collect images regularly over long periods of time.
            The most important step required to convert photographs into atmospheric science is to create a consistent record of the same scene over a wide range of conditions. This almost always means that the same scene should be photographed over long periods of time. Over time, and with practice, you will learn to interpret even small differences that indicate atmospheric conditions have changed.

  4. Keep careful records about scenes, dates, times, and camera settings.
            In science, it is always important to document your work, both so you will remember what you did in the past and so that others will be able to understand and interpret what you have done. The first step is to record information about the scene itself. This information should include the latitude, longitude, and elevation from which the photo is taken, the time and direction in which the photograph is taken, and camera settings. Of course, if you are consistently photographing the same scene, some of these values need to be recorded only once.

  5. Always use the same camera and, if possible, use the same manual settings.
            Some digital cameras do not allow you to manually select settings for ISO ("film speed"), focus, exposure time, and f-stop. settings. This means that whenever you take a photo, the camera will automatically try to "fix" its settings to give the best possible photo. This makes digital photography easy, but it is not the best situation for scientific sky photography. Ideally, photographs of the same scene over a long period of time should be taken always with the same camera settings. Fortunately, there are many inexpensive cameras that do allow manual settings. For your camera, you will need to consult its instruction manual. Then, see below to learn how to select appropriate manual settings for your camera.
    If your camera does not allow manual settings:
            You can still participate in this project. Check the "automatic settings" box on the project reporting worksheet.

  6. Use the highest resolution that your camera supports.
            Some cameras provide a range of resolution settings. You should always use the highest resolution settings to get the most detailed photographs. But, older cameras with image resolutions of 5 megapixels, or even less, are OK for this work as long as you follow the rest of these guidelines. (All the photos appearing on this web page were taken with a 5 megapixel camera.)

  7. Do not apply digital enhancements or resize or compress the image.
            The ability to manipulate images is what makes digital photography so much fun. However, for this purpose, it is important to save pictures in their original condition, without any manipulation of the image such as resizing or "sharpening." or attempting to adjust the brightness, contrast, or color saturation.

  8. Report your data.
            Here is a worksheet for reporting your sky photographs.

    Site Information
    School NameAddressContact Information
    (name, email)
    Longitude
    (decimal degrees)
    Latitude
    (decimal degrees)
    Elevation
    (meters)
               

    Camera Information
        Camera Brand        Camera Model    Camera Settings
        Automatic ◊
    ISO ("film speed") =        
    f-stop =
    exposure =
    focus =

    Scene Information
    Photo taken from...    Facing...
    (compass direction)   
    Comments and important scene features   
         

    Record of photographs
    Date      Time (UT)    File Name        Weather Conditions                                Additional Comments                               
             
             
             
             
             
             
             
             
             
             
             
             

Interpreting sky photographs

Below is an example of a sky photograph. It was submitted by Srisaang Panyangam from Chumchon Banmaekhaowtomluang School in Chiang Rai—the first photograph submitted following a workshop at IPST, Bangkok, in early January, 2009. This photograph was taken using automatic camera settings. The "line" box is checked on the ImageJ toolbar and the cursor has been used to draw a straight line from the bottom of the photo to the top. The graph shows the result from choosing ImageJ's "Plot Profile" from the "Analyze" option on the toolbar. ImageJ converts the photograph to "grayscale," with brightness levels from 0 (black) to 255 (white) and plots the values as a function of distance above the horizon, represented as pixels. The data represented in this graph can be saved as a comma-delimited (.csv) file than can be opened directly in Excel, or "imported" into an existing worksheet.

This analysis represents quantitatively what is qualitatively obvious from looking at the photo. The sky is mostly clear, but there is a layer of haze close to the ground that causes the sky to be a little darker near the horizon. The decrease in brightness, and the location of the maximum brightness above the horizon, is related to aerosols—particulates suspended in the atmosphere. The shape of the graph provides a quantitative representation of what is observable as haziness and decreased horizontal visibility through the atmosphere. Often, the haze evident in this photo is caused by urban pollution from vehicles, for example. Windblown dust can also decrease atmospheric visibility. The ability to quantify the information contained in these photographs is what converts "just" an image into real atmospheric science.

The next photo shows a scene taken a little after local clock noon, facing toward the northeast along a road near my home. I chose this scene because it gives the best view down to the horizon that is available near my home. Although the power lines running across the road might at first might seem just to get in the way, they actually serve as convenient markers that help to relate the density plot to a particular location in the sky above the road. This photo, too, is taken with automatic camera settings. The sky doesn't appear much different in this photo from the sky over the scene from Thailand. But, the shape of the density curve, with its maximum a little closer to the horizon, might indicate that the air over this scene is a little cleaner than the air over the Thailand scene. Indeed, this scene, looking toward the northeast, is facing away from the large metropolitan area around Philadelphia. However, note that it is not easy to compare these two photos and their density plots directly (especially the absolute values of the densities), because they were taken with different cameras. Comparisons are further complicated because the time of day and direction of the photo from Thailand is not known.

Using photography to measure horizontal visibility through the atmosphere

When you include objects on the ground in your sky photographs, it is possible to relate the brightness of those objects to changes in horizontal visibility through the atmosphere. Horizontal visibility is strongly affected by water vapor suspended in the air and by air pollution generated near the ground—urban air pollution from vehicle emissions, for example.

Imagine looking at a distant tree-covered mountain ridge. On a very clear day, the ridge will be distinct and green. On a hazy day, when visibility is poor, the ridge will be grayish or bluish, and less detail will be visible. The same kinds of changes will be visible in the face of a distant large building. On very hazy days, it may even be the case that a building or other distance scene will not be visible at all through the atmosphere. With ImageJ software, it is possible to quantify changes in the appearance of a distant object by comparing characteristics of a small portion of that image either against the same portion from a different day, or against the sky background just above the object.

The basic requirement for interpreting horizontal visibility is that the lighting conditions on the scene being viewed must always be as nearly the same as possible. If the object you are photographing is sometimes sunlight and sometimes shadowed, then, of course, the brightness of that object will be much different under those two conditions. It is probably best to photograph a distant object when it is sunlit, and not in shadow.

If you are facing north in the northern hemisphere, outside the tropics (north of the Tropic of Cancer), the sun is always at your back at noon. So, an object to the north will always be sunlit at noon. If you are close to the Tropic of Cancer, the summer noontime sun will be very close to directly overhead. The reverse conditions apply if you are in the southern hemisphere, south of the Tropic of Capricorn. If you are inside the tropics, sunlighting conditions on a distant object become more complicated. In that case, it may be better to choose an object to the east or west and photograph it consistently in the afternoon (if it is to the east) and in the morning (if it is to the west), to ensure that it is always sunlit. But, in any case, sunlighting conditions on a distant object will always vary with the seasons. As a result, you may need to photograph the same object during several seasons in order to separate seasonal effects from changes in horizontal visibility through the atmosphere.

The first photo below is the skyline over the city of Philadelphia, PA, USA, taken at about 12:30EST on 24 February 2009. The next two photos show histograms of grayscale pixel counts (one of the choices under ImageJ's "Analyze" tool) generated first by zooming in on one of the tall buildings on the skyline and then on the sky just above the building. The histograms show the mean grayscale value of all pixels in each scene. The ratio of the mean pixel value for the building to the mean pixel value for the sky is a measure of the horizontal visibility. This building is about 29 km southeast of where the photo was taken. Similar histograms could be generated by using an area just below the building. In the summer, when skies usually hold more water vapor and when there is more air pollution over Philadelphia, it is not uncommon for these tall buildings on the Philadelphia skyline to be invisible—that is,
(mean pixel value for building)/(mean pixel value for sky) = 1.

The effects of manual versus automatic camera settings

The three photos below are taken at the same place, date, and time, as the photo near my home shown previously. The first photo is the same photo as shown above, taken with automatic camera settings. The next is at a manually set exposure of 1/500 s and the third (in the second row) is at 1/400 s. Depending on your computer monitor, you may not be able to detect any differences among these three photos. As shown in the accompanying graph, the manual exposure at 1/400 s produces a density plot essentially identical to the auto exposure. The manual exposure at 1/500 s produces a slightly darker image, but the shape of the density plot is the same. Under other conditions, there may be more significant differences in the absolute values of the grayscale densities at these three exposure settings (auto, 1/500 s, 1/400 s). Using automatic exposure means that the shape of the density plots can be compared from different days, but not the magnitude of the density plot values. So, some information is lost if manual camera settings are not available, but quite a bit of information still remains. With automatic camera settings, it would probably be a good idea to "normalize" density plots taken on different days. This involves selecting one day as a "reference" and then adjusting other density values so that the maximum values agree with the reference day. Then, it will be possible to examine differences in shapes among several density plots.