Galveston Bay: Estuarine and marine habitat change analysis

Changes in wetland habitats between 1995 and 2002 were analyzed using digital photography in ArcGIS programs. The two time periods were chosen because of the availability of digital photography that was already properly georeferenced for overlay in a GIS system. ArcGIS 8.x software was chosen and then updated with ArcGIS 9.0 software because of widespread use of ArcGIS in Texas. The finished maps are in feature classes by 1:24,000 quadrangles contained in a geodatabase and are in UTM NAD83 coordinate system. The 1995 photography used was Digital ortho-quarter quad (DOQQ) color-infrared photos at 1 m resolution available in compressed MrSid format from the Texas Natural Resources Information System (TNRIS) website. The 2002 era photography included two types, a CIR mosaic in compressed ECW format of the lower Galveston bay ecosystem in UTM NAD83 north zone 15 coordinate system and true color photos at 1-foot resolution of the upper bay ecosystem, which was available from Houston-Galveston Area Council (H-GAC) in State Plane South Central- NAD83 coordinate system.

Habitats were classified using the National Wetland Inventory (NWI) classification system. For this project we classified only the estuarine and marine system habitats with emphasis on E2EM1N (regularly flooded intertidal emergent marsh), E2EM1P (irregularly flooded intertidal emergent marsh), E1AB (aquatic bed) habitats, and SS (scrub/shrub) habitats. Other habitats, including US (unconsolidated shore) and UB (unconsolidated bottom) were mapped. All work was done on-screen at greater than 1:4000 resolution to draw polygons around each habitat.

Work was done by quadrangles and maps were produced for each 1:24,000 quadrangle covering the Galveston Bay ecosystem. The 1995 vector maps were drawn by either modifying 1989 or 1993 NWI maps or drawing new maps using the 1989 and 1993 maps as guidelines for classification. For most maps the areas of change from 1995 to 2002 were drawn as a separate layer while the 1995 maps were overlaid on 2002 photography. The two maps were merged to produce one map containing three attribute labels (Habitat_1995, Habitat_2002, and Reason). The different attributes were used to calculate the change analysis and acres of each habitat in 1995 and 2002.

A previous analysis reported about 112,678 acres of wetlands in 1989 with most of the wetlands (109,909) being emergent type wetlands. Our maps indicated that there were 118,072 acres of wetlands in 1995 of which 116,563 acres were emergent wetlands. There were 45,071 acres of E2EM1N and 71,491 acres of E2EM1P type emergent-wetlands. There were 1,102 acres of shrub type wetlands and 407 acres of aquatic bed. In 2002, the total number of acres of wetlands was 116,890. There were 44,568 acres of E2EM1N and 70,872 acres of E2EM1P habitats, totaling 115,440 acres of emergents. There were 387 acres of E1AB and 1,063 acres of shrubs in 2002.

A gain of 938 acres of wetlands offset a loss of 2,162 acres of wetlands from 1995 to 2002, resulting in a net loss in wetlands from 1995 to 2002 of 1,181 acres (also calculated as 1,224 by second method). The loss was 1.0 percent of wetlands that were documented in 1995.

Losses of wetlands occurred primarily due to erosion, totaling 1,238 acres lost due to erosion. The greatest loss was in the E2EM1N habitat with a loss of 881 acres. Losses due to other causes were relatively small. Subsidence caused 69 acres of loss. The amount of direct man-induced loss (development, fill, excavation, man-made structures) was 830 acres; however, a large part of this loss (over 648 acres) was “fill” within dredged material containment areas, which are subject to periodic manipulations. It is unknown whether the other direct man-induced losses and changes were done through Corps of Engineers permits. Estuarine and marine shorelines and beaches showed considerable change due to erosion and wave-induced shifting of sediments.
Although fragmentation obviously occurred, we could not effectively detect the change due to fragmentation. The fragmentation analysis using raster data needs further refinement to be used accurately.

The technique of preparing 1995 maps and then modifying the maps to detect changes at a later period such as 2002 worked effectively once the 1995 maps were properly prepared. Considerable time was spent creating accurate polygon boundaries for the 1995 maps prior to doing the change analysis. Classifying of habitats will still be a problem for all persons making the maps, but once polygons are accurately drawn, classification and reclassification can be done by people knowledgeable of the regional habitats. One major problem in the work was in the software, which allowed only separate lines for overlapping polygons. The separate lines resulted in sliver problems due to the merging process. Some slivers remain in the maps. The use of an existing map as a base map eliminates problems associated with different interpretations of a habitat and differences in where boundaries are drawn by different technicians. The polygon boundaries can be adjusted as ground truth information or examination of other photos of different time periods create a need to reclassify or redraw boundaries of specific polygons. The existing digital maps can be a document that can be modified over time. Separate maps should be avoided for change type analyses. It is recommended that future work on changes be done based on existing maps.