Identifying and analyzing closed depressions in karst areas is important for sinkhole hazard evaluation and land management. We created a sinkhole inventory in the karst region of northwest Puerto Rico using a lidar-derived elevation model acquired in 2018 approximately eleven months after Hurricane Maria. The goal of this project is to develop a geodatabase of sinkhole feature classes (polygons and points), relevant geometric attributes of each feature, and a density raster to portray areas of greater clustering of sinkholes as an input for future sinkhole susceptibility assessment. We used ArcGIS Pro® v3.0 to create closed depression polygons using two semi-automated extraction methods. A fill-difference method was used to capture depressions nine square meters and larger, and a contour tree method was used to capture nested depressions larger than one hundred square meters. Quality checks were conducted to eliminate non-karst depressions, such as human-made depressions and those resulting as artifacts from the automated methods. Geospatial data of land cover, soils, and geology helped to refine the results and improve quality control. The most challenging aspect of this effort was determining a true karst sinkhole from other depressions extracted from the lidar-derived elevation model. Limitations of this semi-automated method include false-positive depressions in the automated results and the exclusion of sinkholes in conducting large-scale eliminations based on landscape attributes. We approached this challenge by combining layers of other geospatial information to evaluate the type of process that could result in a closed depression. This project will help develop an efficient method to visualize karst hazards utilizing lidar-derived elevation models and sinkhole geomorphic expressions. The resulting geodatabase can be used to efficiently identify sinkhole susceptibility and support land management decision-making in karst areas.