A MULTISPECTRAL APPLICATION FOR TRANSFORMATION OF LIDAR BASED DIGITAL SURFACE MODELS INTO DIGITAL ELEVATION MODELS

A filtering mask was constructed for conversion of LIDAR based digital surface models into digital elevation models and an accuracy assessment of the results was completed. Light Detection and Ranging (LIDAR) elevation data offers highly accurate topographic information. Filtering of LIDAR datasets enable analysis of terrain, devoid of vegetation and urbanization features. Infrared and red spectral bands of Digital Orthophoto Quarter-Quadrangles (DOQQ) are employed to develop a vegetation classification filtering-mask in ERDAS, Imagine. A final mask created with the DOQQ is overlain on a LIDAR based digital surface model, allowing for manual removal of all vegetation and manmade structures, thus producing a digital elevation model. The study area, which includes the city of Boca Raton, is located in southeastern Florida and is equivalent in size to one DOQQ, approximately 4 miles by 4.5 miles. Twenty-five LIDAR tiles are mosaicked to correlate with the dimensions of the DOQQ. The LIDAR dataset was downloaded via Florida International University's International Hurricane Center's website, which offers the datasets in filtered and unfiltered versions. A comparative analysis was completed by ground-truthing the study area along with digitized 1940 aerial photographs. Results depict benefits and disadvantages for filtering the raw LIDAR datasets by means of the multispectral methodology verses the filtering method utilized by FIU.