10. Limitations and Restrictions of the Data

Because of the spatial extent of this data base, the period of record, sampling frequency, and scale of the source documents varied. The use of long-term averages and the use of the 0.25° grid cell as the spatial scale for these data has minimized the error that may have been introduced when these data sources were integrated into a single data base with uniform formats and scales.

The geologic data were classified in terms of an ordinal scale based on the relative hardness of minerals comprising the rock, and derived from state geologic maps. Since these characteristics cannot be deduced from the geologic maps alone, field checking would be required to obtain a more detailed classification than that used in this data set.

The sea-level trend variables (derived from long-term tide-gauge records) may have significant error due to the interpolation methods used. The tide-gauge records used for calculating the sea-level trends on the West Coast were obtained from the records of the Permanent Service for Mean Sea Level (Pugh et al. 1987). These records have been examined and contain no identifiable errors, are of very high quality, and have been used in several sea-level-rise studies (Douglas 1991). However, the sparse station network has made it necessary to calculate the sea-level trend variables for intervening grid cells by calculating a slope line between the two closest adjacent stations. Confidence in the accuracy of the local subsidence variable and the relative and corrected sea-level trend variables estimated with this method decreases as the distance between grid cells that are missing data and adjacent tide-gauge stations increases. For the U.S. East and West coasts, it was found that if the distance from a grid cell with no data to the nearest two long-term gauge stations (i.e., that are east and west or north and south of the no-data grid cell) exceeds ~350 km (i.e., at that distance the r2 of adjacent stations is 0.717), then the sea-level trend variable derived for the no-data grid cell may be erroneous. However, the highly variable topography, geology, and geomorphology of the West Coast, together with the active tectonism suggest that interpolations of the sort proposed here should be used with caution. Whenever feasible, local subsidence (or uplift) data should be used. Some longer-term geologic trends are listed in Appendix D.

Should the user choose to apply the methods and data illustrated in Appendix D to calculate a revised local subsidence trend variable, it will be necessary to reassign a risk value as well, before calculating a coastal vulnerability index.

The statistical summations given within this NDP reflect 0.25° latitude by 0.25° longitude gridded data values and may vary slightly from those given in publications by the contributors. Another discrepancy is that the tide tables used in this document are for 1992 (NOS 1992) as were those used in NDP-043B, while the East Coast, NDP-043A tide tables were for 1988 (NOS 1988).

The coastal hazards data base presented here for the U.S. West Coast omits several factors that may be important when determining the risk of a given area to inundation or erosion. Other variables that may be useful in the risk assessment process are: storm surge, storm frequencies, storm intensities, presence of exposed infrastructure, coastal population density, the role of sediment transport, and the risk of saltwater intrusion (Titus et al. 1991, Snedaker and Sylva 1987). Several studies have been done that consider several of these factors. Gornitz et al. (1994) conducted a pilot study with anexpanded CVI based on the seven relative risk variables in this NDP and six climatic factors derived from Birdwell and Daniels (1991). A copy of the results of this study is reprinted in Appendix D.