Modeling and Mapping High Water Table for a Coastal Region in Florida using Lidar DEM Data

Area of Science:

• Geosciences
• Environmental Science
• Hydrology

Background:

• Accurate high water table elevation mapping is essential for coastal inundation risk analysis and engineering projects.
• Traditional geostatistical methods have limitations in predicting water tables in areas with sparse data or beyond observed spatial domains.
• Understanding the interplay of groundwater, tide, and surface water is vital in coastal landscapes.

Purpose of the Study:

• To evaluate multiple linear regression (MLR) and support vector machine (SVM) for high water table prediction and mapping.
• To develop an application protocol for MLR and SVM in coastal environments.
• To compare the performance of MLR and SVM against geostatistical methods.

Main Methods:

• Utilized fine spatial resolution lidar-derived Digital Elevation Model (DEM) data.
• Applied multiple linear regression (MLR) and support vector machine (SVM) techniques.
• Designed a protocol for applying MLR and SVM in coastal landscapes with complex hydrological interactions.

Main Results:

• Support vector machine (SVM) significantly improved high water table prediction, achieving a mean absolute error (MAE) of 1.22 feet and root mean square error (RMSE) of 2.22 feet.
• Multiple linear regression (MLR) also showed promise, with MAE around 2 feet and RMSE around 3 feet.
• Ordinary Kriging methods were unable to generate reasonable water table predictions.

Conclusions:

• Both MLR and SVM are valuable alternatives for estimating high water table elevation in coastal Florida.
• Fine-resolution lidar DEM data enhances the accuracy of high water table prediction and mapping.
• The developed protocol provides a robust framework for coastal water table assessment.