Analysis and Summary of Water-table Maps for the Delaware Coastal Plain PDF Download

Author: Matthew J. Martin
Publisher:
ISBN:
Category : Coasts
Languages : en
Pages : 20

Get Book Here

Book Description
A multiple linear regression method was used to estimate water-table elevations under dry, normal, and wet conditions for the Coastal Plain of Delaware. The variables used in the regression are elevation of an initial water table and depth to the initial water table from land surface. The initial water table is computed from a local polynomial regression of elevations of surface-water features. Correlation coefficients from the multiple linear regression estimation account for more than 90 percent of the variability observed in ground-water level data. The estimated water table is presented in raster format as GISready grids with 30-m horizontal (-98 ft) and 0.305-m (1 ft) vertical resolutions. Water-table elevation and depth are key facets in many engineering, hydrogeologic, and environmental management and regulatory decisions. Depth to water is an important factor in risk assessments, site assessments, evaluation of permit compliance data, registration of pesticides, and determining acceptable pesticide application rates.Water-table elevations are used to compute ground-water flow directions and, along with information about aquifer properties (e.g., hydraulic conductivity and porosity), are used to compute ground-water flow velocities. Therefore, obtaining an accurate representation of the water table is also crucial to the success of many hydrologic modeling efforts. Water-table elevations can also be estimated from simple linear regression on elevations of either land surface or initial water table. The goodness-of-fits of elevations estimated from these surfaces are similar to that of multiple linear regression. Visual analysis of the distributions of the differences between observed and estimated water elevations (residuals) shows that the multiple linear regression-derived surfaces better fit observations than do surfaces estimated by simple linear regression.