Generating Continuous Surface Probability Maps from Airborne Video Using Two Sampling Intensities Along the Video Transect
Airborne videography can be an effective tool for assessing the effects of catastrophic events on forest conditions. However, there is some question about the appropriate sampling intensity to use, especially when trying to develop correlations with probabilistic data sets such as are assembled through the Forest Inventory and Analysis (FIA) surveys. We used airborne videography to assess damage to forest resources by Hurricane Andrew and the catastrophic February 8-l 1, 1994, ice storm; but, those efforts were limited by the absence of a direct link between video imagery and FIA field plots. In this study, we used the 1994 ice storm in northern Mississippi to test bias and accuracy at two sampling intensities-14.5 by 14.5-km spacing (square) and a 14.5- by 1.6-km spacing (rectangular). Results showed that the square pattern resulted in less biased estimates of damage. We suggest that this bias would be lessened further if sampling was made on the 4.8~km sample grid of FIA data points. We assert that bias could be eliminated if imagery was acquired directly over FIA plots.