ABSTRACT:Terrestrial laser scanning (TLS) is increasingly used for forestry applications. Besides the three dimensional point coordinates, the 'intensity' of the reflected signal plays an important role in forestry and vegetation studies. The benefit of the signal intensity is caused by the wavelength of the laser that is within the near infrared (NIR) for most scanners. The NIR is highly indicative for various vegetation characteristics. However, the intensity as recorded by most terrestrial scanners is distorted by both external and scanner specific factors. Since details about system internal alteration of the signal are often unknown to the user, model driven approaches are impractical. On the other hand, existing data driven calibration procedures require laborious acquisition of separate reference datasets or areas of homogenous reflection characteristics from the field data. In order to fill this gap, the present study introduces an approach to correct unwanted intensity variations directly from the point cloud of the field data. The focus is on the variation over range and sensor specific distortions. Instead of an absolute calibration of the values, a relative correction within the dataset is sufficient for most forestry applications. Finally, a method similar to time series detrending is presented with the only pre-condition of a relative equal distribution of forest objects and materials over range. Our test data covers 50 terrestrial scans captured with a FARO Focus 3D S120 scanner using a laser wavelength of 905 nm. Practical tests demonstrate that our correction method removes range and scanner based alterations of the intensity.