Laboratory Exercise 1

Transcription

1 Page 1 Laboratory Exercise 1 GEOG*2420 The Earth From Space University of Guelph, Department of Geography Prof. John Lindsay Fall 2013 Total of 32 marks Learning objectives The intention of this lab exercise is to familiarize students with common data sources for Earth observation including aerial photography and satellite imagery. In addition, students will gain familiarity with the Whitebox Geospatial Analysis Tools software package for analyzing and visualizing these data, learning skills that they will use frequently in later lab exercises. This first lab will require you to do quite a lot of careful reading. Before you begin You will need to download the latest version of Whitebox from the following web page: The Whitebox download site can also be accessed from the Whitebox GAT homepage, which you can find by searching for 'Whitebox' in Google. Whitebox GAT is free and open-source software (i.e. it will not cost you anything). If you are working on a computer in the Geography undergraduate computing lab, I would suggest downloading the Whitebox zip file onto the computer's desktop. There is no installation process for the software. The Whitebox download is a compressed zip file, which you will need to unzip (decompress) before running the program. To launch the program, simply open the unzipped Whitebox folder, locate the WhiteboxGIS.jar, and double-click this jar file. (A jar file is Java's equivalent to the MS Windows '.exe' file.) If you are running Whitebox on your own machine, you will need to ensure that you have an updated (Java 7 or higher) version of Java installed and enabled. You can download the latest version of the Java Runtime Environment (JRE) from the Oracle website. Whitebox will not run when you try to launch it if you don't have Java installed and configured properly on your machine. So, why don't we simply make sure that Whitebox is on every machine in the undergraduate computing labs so that you don't have to bother with downloading the file? Good question! The answer is that I'm likely going to be adding new functionality to the program as we proceed with the lab exercises throughout the semester. So you'll always need to be running the most up-to-date version of Whitebox to complete the current lab. In addition to the software, you will need to download the data associated with this lab exercise from the CourseLink page under the Lab 1 directory. Note that the data for this course will be quite large and

2 Page 2 you may want to purchase a flash drive to dedicate to your lab exercises. Also, if you are downloading the data over the wireless network, or at home, you may have to be patient. The data are also compressed and will need to be unzipped into a data folder before you can proceed with the exercise. Note that you should back-up this folder frequently and that all of the data that you place on the hard drive of the workstation in the Undergraduate computing lab will be cleared every time you log on. YOU SHOULD NOT LOG OUT WITHOUT FIRST BACKING UP YOUR DATA FILES. What you need to hand in You will hand in a printed report summarizing the answer to each of the questions in the following exercise along with the necessary colour images. Notice that you will need to have paid your lab fee to have printing privileges in the Hutt building computer labs. Part 1: Earth observation data Earth observation data is largely imagery and can be categorized based on the platform that these data are acquired. The two main types of images that we'll use in this course are captured using aerial platforms (airplanes and helicopters) and satellites. You will see both types of data in this first lab exercise. Many types of images can be captured using both platforms including film or digital photographs, multispectral images, hyperspectral data, thermal images, passive microwave and radar (i.e. active microwave) images, and laser altimetry (LiDAR) data. In this course we'll focus more on the first two types of images because they are the most common and widely used. If you go on to take the advanced 3 rd year remote sensing course, you'll likely encounter some of the more 'exotic' Earth observation data types. Air photo interpretation is the field that has historically been associated with the acquisition and application of aerial photographs. Remote sensing is the broader field that is concerned with detection and classification of features of Earth's surface using the properties of electromagnetic radiation measured using some kind of sensor. Although remote sensing is more commonly associated with satellite-based imaging, it also includes aerial platforms as well. Photogrammetry is the field that focusses on measurement of geometric properties from photographs (not necessarily images of the Earth's surface, but for our application, we will constrain this). Each of these overlapping fields of study are a part of the overall discipline known as geomatics (note: geomatics also includes GIS). Each of the sub-disciplines of geomatics are similar in that they require the acquisition and application of geospatial data. Earth observation data can be acquired for use by 'end-users' like us from national aerial photograph archives (e.g. the National Air Photo Library in Ottawa) and various government and commercial providers of satellite imagery. The United States Geological Survey (USGS) provides free access to much of the data captured by the various satellites operated by NASA. Other Earth observation data sets, e.g. the French operated SPOT satellite data, are sold to end-users by commercial vendors. New aerial photography can also be purchased for a specific project by one of many smaller companies that specialize in the acquisition of aerial photography. Obviously it cost more to get new air photos or satellite images than it does to purchase archived data. Part 2: Using Whitebox with image type data When you get data it may come in hard-copy (i.e. a contact print of the original film photograph) or

3 Page 3 digital format. Since the majority of Earth observation applications these days rely on digital image processing, hard-copy data are usually digitized using scanners as a first step. Digital data is usually provided in a generic image format such as a TIF, GIF, PNG, or BMP file. These are the same types of files that your digital camera or desktop scanner create. One of our first tasks, therefore, is to import these image files into the software that we'll use for Earth observation work. If you haven't already done so, launch Whitebox GAT and familiarize yourself with the basic layout of the program. Like most desktop applications, Whitebox possesses a menu, toolbar, tabbed side pane, and a large area used to display data, in our case mainly imagery and vector data. The toolbar contains the common functions that are needed to add data layers and to navigate (i.e. zoom in and out and to move around the image). The tabbed pane at the lefthand side contains a 'Tools' tab, which has a treebased themed toolbox structure that can be used to access the various image processing tools that we'll use, and a 'Layers' tab that shows information about the displayed raster (image) and vector (line) data that are displayed in the current map area. Select the 'Tools' tab if it is not already active. Click on the 'Data Import/Export' toolbox. There are many different geospatial data formats that Whitebox is able to work with, however, we are mainly concerned with the Import GeoTIFF (a geotiff file is a common image format used for exchange geo-registered satellite image) and the Import Image tools. A fair amount of imagery data may also come in a format handled by ESRI's ArcGIS. Whitebox can exchange raster data with Arc using the Import ArcGIS Binary Grid tool and its equivalent export tool. For now, open the Import GeoTIFF tool. This tool will convert TIFF files into the raster format that is used by Whitebox (*.dep and *.tas files). Select the Guelph_A tif file for import from whichever directory you have saved the lab exercise data in. After pressing 'Run' the imported tiff image should be automatically displayed. This image is an example of an aerial photograph that was originally captured using photographic film and later digitized using a high-resolution flatbed scanner. If you click the Layers tab, you will see that the imported file appears listed as a layer within the active MapArea on the active Map. In Whitebox, a Map is a page that contains one or more map elements, i.e. things like map areas, titles, legends, scales, etc. These elements can be added and manipulated from the 'Cartographic' menu. Maps can be saved, open, closed, printed, and exported as images that you can embed into your report document. As a basic cartographic element, a MapArea is the space in which data layers (e.g. raster images and vector files) are displayed. By default, when you open Whitebox, there is one map that contains one map area, which occupies the entire map page (notice that you can resize the map area if you click on it). In fact, multiple Maps can be open at any one time, and each Map can contain more than one MapArea (e.g. if you needed to make an inset map). Later in the course we'll need to use multiple open Maps to handle the simultaneous display of multiple images of different areas. Within a MapArea, data layers such as our displayed aerial photograph can be added, removed, reordered (i.e. moved up or down the display stack) using the 'Data Layers' menu or the tools in the second grouping of icon buttons in the toolbar. At the moment, however, we only have one displayed data layer in the MapArea. The third grouping of tools in the toolbar, and the tools within the 'View' menu, can be used to navigate within a Map or MapArea. By default, the 'Select Map Element' tool is selected when you first open the program. When a map element is selected, it will appear with a dashed outline, and you will be able to move and resize the element. We will frequently use the Pan (hand icon), Zoom In (magnifying glass icon), Zoom Out, and Zoom to Full Extent (globe icon) tools to navigate around a displayed image within a MapArea. Practice using these tool by zooming into the displayed aerial photograph and moving about (i.e. panning). You can also shortcut these tools by using

4 Page 4 the arrow keys and plus or minus keys on your keyboard. One thing to be mindful of is that if you try to zoom into an image, but your cursor is not over the MapArea as you trace the zoom box, you will zoom into the Map (i.e. the page) rather than the MapArea (i.e. the data layers). In this case, you can zoom back out of the page using the Zoom To Page (icon of magnifying glass with page inside) tool. Q1. The process of scanning the hard-copy aerial photograph will naturally result in a degradation of image quality when compared to the original. Are there any artifacts of the scanning that are apparent within the image and, if so, describe them. It may help to compare the image to the original hard-copy photo, which your TA will have in lab. (2 marks) Part 3: Digitizing features The aerial photograph is of the south size of the City of Guelph and surrounding areas taken many years ago. Quite a lot has changed in this area of the city in the passing decades since the picture was taken. Nonetheless, some familiar features are apparent, including the campus of the University of Guelph. Johnston Green, the football field (Alumni stadium), and even the Hutt Building are present. Try to locate these three features. Be mindful however, that the imported file was a raw image and did not contain any information about georeferencing. The northing and easting data displayed at the bottom lefthand corner are not useful and even the orientation of the image may not be what you expect (in a later lab, we're going to work through the process of applying georeferencing data to raw imagery like this). It may be helpful if you consult what the modern area looks like from above by looking at the imagery available on Google Maps. An image analyst will use any available resource to help them identify features of interest. Once you've located Johnston Green, the Guelph University football field, and the Hutt Building in the image will need to mark their locations in some way. To do this, you will create a vector shapefile and add point features to the three locations. You will first have to read and carefully review the instructions in the 'How to digitize new vectors' tutorial within Whitebox's help. We're going to use this functionality quite often in this course, so you'll need to get comfortable with this process. Play around with digitizing features into various shapefiles. You can always delete your practice shapefiles later using the Delete Files tool. Once you are comfortable with the digitizing process, create a shapefile to enter the locations of your three sites of interest. The shapefile will be of the Point shape type and you should create a record called 'NAME' within its attribute table to store the specific name of the point feature (i.e. Johnston Green, Alumni Stadium, and Hutt Building). You should name your file with your Guelph username (e.g. jlindsay) and 'campus points'. So, for example, my points shapefile would be called 'jlindsay campus points.shp' (note that Whitebox allows spaces in file names). Position your points as near the centre of the features as you can. Your TA will provide instructions for how you can submit this digital file for grading (6 marks). Also embed an image (using the 'Export Map As Image' function from the 'File' menu) zoomed into the campus and your three points in your submitted report (2 marks). Q2. Describe the process that you used to locate and identify the three features of interest. That is, what was involved in the process of recognizing the features. Hint: in answering this question, read about the Elements of Image Interpretation (also known as the factors of recognition, which we'll talk about further in lecture later in the semester) in Jensen Chapter 5 pp (4 marks)

5 Page 5 Part 4: Satellite imagery Add a new map using either by selecting 'New Map' from the 'File' menu or by pressing the 'New Map' icon on the toolbar (first icon). The new map should be displayed with a single full-extent MapArea and no displayed data layers. Being the active map, it will appear in the 'Layers' tab with bolded text. Note that you can switch between the two maps (i.e. your new map and the previous map containing the displayed aerial photograph) simply by clicking them in the 'Layers' tab. The active map will always appear bolded. Making sure that your new map is activated, select 'Add Layers to Map' from the 'Data Layers' menu or press the 'Add Layer' icon from the toolbar. Add the files 'Landsat 8 OLI band1' through 'Landsat 8 OLI band7' to the MapArea. Note that you can add all seven images at one time. Each of the seven images will be displayed, however, because they are completely overlapping, you will only see the layer at the top of the stack until you either change the stack order, or, more conveniently, toggle the layer's visibility by deselecting the checkbox that appears beside its entry in the layers tab. The displayed images are the first seven bands of a multispectral satellite dataset acquired by the recently launched Landsat 8. The data were imaged July 15, A multispectral dataset captures the intensity of light reflected or emitted by the ground surface within a specific regions of the electromagnetic spectrum (i.e. a relatively narrow range of wavelengths) and store these data in several images, or bands. For each scene of Landsat 8 data, there are 11 images captured and each band, when displayed, appears as a greyscale image. The brightness of each pixel in the image is directly related to the intensity of reflected light within the part of the spectrum recorded by the satellite sensor. We'll talk about the electromagnetic spectrum in detail in lecture but if you should read Chapter 2 of Jensen now. Q3. Landsat 8 is the latest in a series of Landsat Earth observation satellites, having been launched in February of It has a sensor suite (Operational Land Imager, or OLI) that is different from each of the previous Landsat satellites. Identify the range of wavelengths, in micrometers, that each of the eleven Landsat 8 bands are sensitive to. Also identify the region of the electromagnetic spectrum in which each band is positioned. (4 marks) I've only given you a small section of the first seven bands because the original dataset is very large. The Landsat scene depicts an area of Southwestern Ontario including parts of Lake Ontario and the Greater Toronto Area, and the cities of Guelph, Cambridge, and Kitchener-Waterloo, and Hamilton. Unlike the raw aerial photograph that we examined in the previous section, this image has been properly geo-registered, i.e. it has proper map coordinates. The USGS provide Landsat data already geo-registered, which is very convenient for end-users. Select the image in the 'Layers' tab and right-click then select 'Layer Display Properties' from the popup menu. The Properties window presents information about the data layer and its display characteristics. Q4. How many rows and columns are in the Landsat image? What is the grid resolution, i.e. the length of each grid cell? (3 marks)

6 Page 6 On a hard-copy colour aerial photograph the intensity of each of the three primary colours (e.g. Red, Green, and Blue) are recorded in different layers on the film. Clearly this information is largely permanent and difficult to separate, i.e. you can't easily see the red image, the green image, and the blue image. One of the great benefits of multispectral data is that the information about scene brightness in a specific range of the spectrum (e.g. red) is separated. This allows the image analyst to modify individual bands (e.g. to improve the contrast of the blue image) and to combine data in ways that allows for the visualization and analysis of data in a very effective manner, namely through the creation of a colour composite. For example, you can create a colour image that closely approximates an image acquired using a true-colour, or natural-colour, film. Use the 'Create Colour Composite' tool found under the 'Image Processing' tool folder to create a natural-colour composite by specifying Band 4 as the red image, Band 3 as the green image, and Band 2 as the blue image. Do not specify an alpha channel file, which would be used to incorporate image transparency. You can improve the colour balance by reducing bias, or overrepresentation of particular colours, by applying the 'Balance Colour Enhancement' tool found under the 'Image Enhancement' tool folder. Call the resulting image 'jlindsay Landsat OLI 432 composite BCE', replacing my username with your own. Include a colour print-out of this image with your written hand-in for the exercise and submit an electronic copy of the.dep file (2 mark). (Note that rasters in Whitebox consist of two files, a.dep header file and a.tas data file. In this case, we only want a copy of the header file.) One of the greatest advantages of multi-spectral data is in the ability to visualize information about a landscape's reflectance in regions of the spectrum far outside of the narrow portion that is visible to human eyes. False-colour composites provide a means of visualizing these data in a very effective way. The key to interpreting a false-colour composite is to realize that the image has been produced by mapping the reflectance of the landscape in some distant portion of the spectrum (e.g. the short-wave infrared) onto one of the three additive primary colours (red, green, and blue) such that we can see the image in 'false colour'. False-colour composites are often referred to by the three band RGB designation. For example, a Landsat OLI composite is the image that results from placing Band 7 into the Red 'channel', Band 5 into the Green channel, and Band 3 into the Blue channel. Create a Landsat OLI 543 composite image, also performing colour balance enhancement, and include a printout of the image with your hand-in and submit an electronic copy of the.dep file (2 mark). Be sure to use the same naming convention as above. Q5. What do more intense red colours in this image indicate about the land cover in these areas and about the brightness within the three bands (5, 4, and 3)? What colours do 1) water surfaces, 2) bare ground or soil, 3) clouds, and 4) urban areas appear as in this image? (5 marks) One of the advantages of using imagery that has been properly geo-registered is that you can measure distances and areas. Q6. There is an application to increase the length of the runways at the Billy Bishop Toronto City Airport, which has caused quite a bit of discussion among the residents of the Toronto islands and waterfront. Using the distance tool (the ruler icon), measure the length of the longest runway at the island airport as of the time the image was acquired (July 2013). (2 marks)