PE&RS July 2001

Highlight Article

Introduction
Landsat Thematic Mapper (TM) and Landsat 7 Enhanced Thematic Mapper Plus (ETM+) sensors provide multiple spectral bands containing information on reflected energy ranging from the visible through the near to mid-infrared wavelengths. Landsat imagery has been well documented for its utility in terrestrial applications such as vegetation studies and land use/land cover mapping; applications which rely primarily on the infrared bands (bands 4, 5 and 7) to discriminate between cover types. Landsat imagery has not been used as extensively for coastal zone applications, primarily due to its spatial resolution (30 meters), but also due to the difficulty of imaging aquatic environments.

For aquatic applications, the blue, green, and red wavelengths provide most of the information, because longer wavelengths (near-infrared and beyond) are absorbed by water. The transmittance and reflectance of shorter wavelengths can be affected by many factors. The primary factor is the atmosphere itself, where transmittance and reflectance are decreased due to Rayleigh and Mie scattering (Schowengerdt, 1997). In aquatic environments turbidity, depth, sun angle, and surface winds also affect the transfer of energy back to the sensor.

Many of these effects can be minimized by acquiring imagery during certain times of the year. In the coastal zone of Florida, conditions are best in the winter when biological productivity is low, bio-organic turbidity is minimal, and the climate is drier. If timed just right, when there are clear skies and water, Landsat imagery may display very little atmospheric or environmental degradation, and the data contained in the blue, green, and red bands (bands 1, 2, and 3 respectively) may be of high quality.

Creating the Cover Image
At the Florida Fish & Wildlife Conservation Commission’s Florida Marine Research Institute (FMRI), scientists were involved in a comprehensive study of Charlotte Harbor, Florida. Charlotte Harbor is located along the Gulf Coast of Florida south of Tampa (Figure 1). The Peace and Myakka Rivers flow into the Harbor, creating a dynamic estuary where salinities fluctuate with river flows and tidal forces. The shoreline is fringed with mangroves and marsh along natural stretches, and seawalls or rip-rap in areas of urban development. The Harbor is fairly shallow ( 0.25 - 6 meters), but there are a series of deeper man-made and natural channels near the river mouths and passes (7 - 20 meters).

A number of data sets were available to the interdisciplinary science team, including data from the Marine Resources Geographic Information System (MRGIS): a collection of spatial data maintained at FMRI. To help the team visualize the study area, two Landsat TM scenes were extracted from the MRGIS; both images were acquired on January 26, 1997. The images were cloud free, and had no apparent haze or smoke. The two scenes were rectified to sub-pixel accuracy using ground control points obtained from digital ortho-photo quadrangles (DOQs). They were then joined to form an image mosaic (base image) covering the entire Charlotte Harbor area. Because both scenes were acquired during the same satellite pass, no color balancing was necessary.

Typically a Landsat image is portrayed as a color infrared composite, displaying bands 4, 3, and 2 with red, green, and blue (RGB) respectively. This is not an optimal band combination for aquatic environments. Rather, a natural color composite using bands 3, 2, and 1 (RGB) or a false color composite using bands 4, 2, and 1 provide more detail. To increase contrast, land areas can be removed from the image (masked), allowing the full dynamic range to be applied to the features of interest.

Although the Landsat TM mosaic contained some very good information after the land mask was applied, it was still difficult to discern differences in water depth. Although deeper channels within the estuary were identifiable by their darker tones and sinusoidal patterns, submerged aquatic vegetation also appeared dark, and could be confused with deeper waters. Narrow channels were difficult to distinguish due to their smaller size, and the 30 meter pixel size of Landsat TM data. Slopes, drainage patterns, and shallows were also difficult to perceive.

To help with visualizing the depth of the Harbor, bathymetry data were obtained primarily from the National Oceanic and Atmospheric Administration National Ocean Service’s (NOAA/NOS) National Hydrographic Survey data (National Geophysical Data Center, 1998). Additional bathymetry data were compiled as necessary from NOAA nautical charts and other databases available from FMRI’s MRGIS. The shoreline and islands were included defining the areas of zero depth. The combined data, consisting of points, lines and polygons, were converted to a triangulated irregular network (TIN). A TIN represents a surface using contiguous non-overlapping triangular facets. The resolution of a TIN can vary internally in relation to the surface complexity and data availability. The TIN can be used as a master source for the creation of products such as grids or contour maps. For this project, a 10 meter bathymetry grid was produced using a linear interpolation method.

The bathymetry grid was co-registered to the base image using shoreline and island features. Displaying the bathymetry data alongside the Landsat TM mosaic was helpful, but truly useful information could be obtained by integrating them into a single image. To add the bathymetry information to the Landsat TM mosaic, hill-shading was applied to the bathymetry grid using a solar azimuth of 225°, a solar elevation of 45°, and a scaling factor of 75x. The solar angles and scaling factor are user defined, and the values were chosen to yield the most visually pleasing view of the Harbor. The Landsat TM mosaic was combined with the hill-shaded view using an overlay function, specifying a natural color band combination of 3, 2, 1 (RGB). The cover image was created by displaying the base image in the background as a false color composite, and overlaying the hill-shaded view in natural color.

The enhanced Landsat image provided a wealth of information not readily apparent in the Landsat imagery or in the bathymetry data alone. By combining bathymetry data and the spectral information in the visible TM bands, a synergistic picture of Charlotte Harbor emerged: a picture which was truly worth a thousand words!

References
National Geophysical Data Center. 1998. GEOphysical Data System for Hydrographic Survey Data (GEODAS). Volume 1, version 4.0.
Schowengerdt, R. A. 1997. Remote Sensing: Models and Methods for Image Processing. Academic Press, Chestnut Hill, MA, 522 p.

Acknowledgments
This work was supported in part under funding from the Department of the Interior, U.S. Fish and Wildlife Service, Federal Aid for Sport Fish Restoration Grant Number F-66. The author wishes to thank Bill Sargent (FMRI) for compiling the bathymetry data used in this project.

Author
Paula Houhoulis Smit
Currently the Landsat 7 Science Data Specialist with Emergent Information Technologies, Incorporated at the U.S. Geological Survey’s (USGS) EROS Data Center; working under indirect contract to NASA.

Formerly with
Florida Fish and Wildlife Conservation Commission, Florida Marine Research Institute, 100 8th Ave. S.E., St. Petersburg, FL 33701-5095, smit@usgs.gov.

Top

Home