Forest stand characteristics estimation using a most similar neighbor approach and image spatial structure information

In this study, information on variograms is integrated with image interpret ation of standwise volume of tree stock by using a nonparametric method bas ed on a distance-weighted mean of most similar neighbors (MSN). The usabili ty of the various indicator attributes, including image pixel characteristi cs and features derived from a variogram curve, in the interpretation is st udied by analyzing the accuracy of the results at the compartment level. Th e accuracy of the volume estimation at the stand level was improved when th e empirical variogram values were included in the set of indicator attribut es in the MSN analysis. With this kind of MSN analysis, the chosen k neares t neighbors have a similar spatial variation structure in the image materia l as in the target stand. It was found that increasing the number of simila r neighbors beyond 3 did not improve the accuracy. When the variogram infor mation was included in the indicator attribute set, the root mean square er ror (RMSE) of volume estimate was at its lowest at 18% and the bias was the n - 0.6%. When the variogram information was not used, the RMSE was at its lowest at 24-27% and the bias was 0.2-1.8%, depending on the number of indi cator variables used. (C) 2001 Elsevier Science Inc. All rights reserved.