Airborne Lidar Bathymetry (ALB) surveys are traditionally used for measuring depths in shallow nearshore and back-bay areas. In this paper, we present a novel ALB waveform processing procedure, namely bottom return residual analysis, for bottom characterization. Waveform features obtained from the bottom return residual analysis are used in a supervised classification approach, i.e. Support Vector Machine, to differentiate between: 1) sand and rock bottoms and subsequently, 2) fine and coarse sand bottoms. The classification procedure was tested on ALB survey data collected with an Optech SHOALS-1000T ALB system that covers a ~7 km2 area within 1 km from shore in the western Gulf of Maine, USA. The bottom classification results, when compared to ground-truth measurements, indicate a 96% overall accuracy for sand and rock classification and 86% overall accuracy for fine and coarse sand classification. Results of ALB-based bottom classification are compared with interpretations of a multibeam echosounder acoustic backscatter mosaic collected from the survey area.