A new approach to DTM error estimation basing on Laplacian probability distribution function

A Digital Terrain Model (DTM) derived from Airborne Laser Scanning (ALS) was the subject of our research. In this paper, the vertical accuracy of the DTM was analyzed on the basis of the commonly used statistics, i.e. mean error and standard deviation, assuming a normal (Gauss) error distribution. The further approach, the so-called robust method (Höhle, Höhle 2009), was also tested, where the median was a substitute for the mean error and the Normalized Median Absolute Deviation (NMAD) for the standard deviation. An alternative method based on the Laplace function is proposed in the paper for describing the probability density function, where the parameters of the Laplace function are proposed for DTM error estimation. The test area was near the Joint Research Centre in Ispra, Italy; raw ALS data covering the test area were collected in 2005 and processed for DTM generation. Accuracy analysis was performed based on the comparison of DTM with the raw ALS data and with in-situ height measurements. The distribution of DTM errors calculated from ALS data was significantly non-normal, confirming other results reported in the literature. The Gauss distribution function considerably overestimated the vertical DTM errors; however, the robust method underestimated them. The Laplace function matched the error histograms the best, and accuracy parameters derived from this function could be considered as an alternative method for DTM accuracy evaluation.JRC.H.4-Monitoring Agricultural Resource