Sources of materials: aerial photographs

As mentioned above, the initial source of material came from archives at Everglades National Park. As the project expanded beyond ENP to the Greater Everglades we began the task of contacting numerous governmental agencies, both federal and State of Florida, that we felt may contain useful materials. The former U.S. Soil Conservation Service (now the Natural Resources Conservation Service [NRCS]) had conducted soil surveys throughout Florida in the 1940-50s and used 1:40,000 scale aerial photographs for their base maps. The numerous offices of the NRCS throughout the state were contacted and many still retained hard copies of the original images in their files. Offices were visited and photographs scanned using portable computer equipment. We subsequently discovered that a state agency, the South Florida Water Management District, had an almost complete set of the 1940s photographs for areas south of Lake Okeechobee. These were in emulsion, rather than print, format. We borrowed these in sets and scanned them at our laboratory in Gainesville, Florida.

The USGS maintains extensive collections of aerial photographs and satellite imagery at its Earth Resources and Observations Data Center (7). These archives were searched and photographs that were missing from the ENP archives were obtained. At present, relatively complete coverage is available for four time periods: 1928, 1940, 1952 and 1964. Recent aerial photography (1970s on) has been routinely collected by Florida's Water Management Districts and Department of Environmental Protection and is accessible via numerous web-sites (e.g. the Land Boundary Information System of FDEP [8]). These photos, called Digital Orthographic Quarter Quadrangles (DOQQs) are available in several map projections. We use the North American Datum of 1983 (NAD83) and the Universal Transverse Mercator (UTM) Zone 17 North projection.

Sources of materials: historical charts and maps

Cartographers have been producing maps and charts for centuries, but accurate mapping is much more recent. Comprehensive surveys of ports and harbors were being made in colonial times in the United States (1700s). Soon after independence, the Survey of the Coast (now the National Geodetic Survey, NGS) was established and given the mission of surveying the entirety of the country's harbors, ports and all coastlines. During the second and third Seminole Wars (1835-1842 and 1855-1858 respectively), surveying of Florida's coastlines became a priority. Many expeditions were mounted (e.g. Bache 1857) and numerous maps and charts, of varying scales and detail, were produced. They are archived by the CGS and scanned copies are provided to other U.S. government agencies upon request.

Data georeferencing

Figure 16.1: An example of georeferencing using the "rubber sheet" approach is shown here. Map (A) is the scanned ungeoreferenced historic T-Sheet # 649 from 1857. Known points (e.g. #1-5) of latitude (Lat) and longitude (Lon) are visible on this map. Using the points of latitude and longitude (Table), this map is georeferenced resulting in a new rectified map (B). Areas of the rectified map B can now be compared with other rectified and georeferenced images. The box in (B) has been enlarged and compared with the same area from a 1928 chart (C vs D). Areas of change, or no change, can then be identified. [larger image]

Once scanned into digital format, brightness numbers, as well as row and column positions are assigned to the lines, shades, and text on historic maps and photos. Column and row positions of each digital number implicitly provide information on relative spatial location of boundaries and other features. Before these scanned images can be effectively combined with other digital spatial data and used in computer aided analyses (geographic information system or GIS), the implicit "X" and "Y" values for each point must be transformed to well-described geographic ones (Figure 16.1). This process of georeferencing can be simple or complex, approximate or very accurate, depending on the information available and the technique used to assign the geographic coordinates. The georeference coordinates can be longitude and latitude values or other units such as meters or feet from a designated reference. These coordinates are assigned through a system (a "map projection") that mathematically transforms 3- dimensional earth surface features to 2-dimensional ones (Robinson et al. 1995). When a map is made, its projection is often documented. If the map's projection is known and identifiable points are labeled with known coordinates, georeferencing is a simple process of selecting points with known coordinates in the scanned image and defining the map projection for the computer. However, for many historic maps the projection is poorly described or unknown. Moreover, for historic aerial photographs, information that can be used to systematically project the imagery, such as aerial camera characteristics (e.g. pitch and roll), is often unavailable. In such cases, it is common to use a "rectification" or "rubber sheet" approach (Jensen 2004). With this method, numerous points for which coordinates can be found on another map or photomap with a well-described projection are selected in the scanned map. Through statistical regression, an equation that minimizes differences between known coordinates for those points and those predicted by the equation is created. The equation can then be used to transform all points within the scanned photograph or map to projected coordinate values in the output image. A hypothetical example using a portion of an Everglades historical topographic map ("T-Sheet") is provided (Figure 16.1). For our T-sheets, a mixed case exists in which lines of latitude and longitude are shown on maps but the map projection is poorly described. These maps were georeferenced by using four steps:

1) converting a minimum of 4 shown longitude and latitude pairs to a commonly used coordinate system named the Universal Transverse Mercator (UTM) system (Snyder 1987);
2) determining the scanned image column/row coordinates for those points;
3) calculating a regression equation to transform the column/row designations for those points to UTM values; and,
4) applying that equation to every column/row pair in the image to transform them to UTM coordinates (Smith et al. 2002).