Predicting tree crown defoliation using color-infrared orthophoto maps

Abstract

Orthophoto maps based on color-infrared aerial photography have been used by the Lithuanian forest inventory since 2001. This study aimed to investigate the opportunities for using these orthophoto maps to predict tree crown defoliation at the single tree and sample plot levels. The test area was located in the Aukstaitija National Park, eastern Lithuania, and it was photographed in the summer of 2008 using a Vexcel UltraCam D digital frame aerial camera to produce digital orthophoto maps with a 0.5 x 0.5 m ground sampling density. Some 1721 tree crowns (mainly pine, spruce and birch), located in 166 permanent sample plots, were identified and delineated on the orthophoto maps. Crown defoliation and other dendrometric characteristics were field-estimated for all of these trees in summer 2008. Judgments on the suitability of using color-infrared aerial photography based orthophotos to estimate tree crown defoliation were based on the accuracy of the defoliation prediction. Defoliation for each crown was predicted using the non-parametric k-Nearest Neighbor (k-NN) method and characteristics extracted from the digital orthophoto maps as the auxiliary variables for prediction. Prediction accuracies were validated using the “Leave One Out” technique by comparing the predicted data with data from field-assessed crown defoliations. The lowest root mean square errors for the predicted tree crown defoliation values were 7.564 for pine trees, 9.166 for spruce and 7.712 for birch and the highest coefficients of correlation between field-estimated and predicted crown defoliations were 0.576, 0.600 and 0.386, respectively. However, there was no best performing solution for using the k-NN prediction found, as the best results were achieved using different approaches. Next, predicted and field estimated tree crown defoliation values were aggregated up to the sample plot level by taking an averaging of trees in the same sample plot. The root mean square error at the sample plot level was around 3.7 %, the bias was statistically not significant and the correlation coefficients between plot-wise average values of field-estimated and predicted defoliations were around 0.8. The achieved results suggested that color-infrared orthophoto maps could be a potential data source of forest health characteristics for use in stand-wise forest inventories.