Stand Volume Estimation From Satellite Data Employing Two-Phased Sampling

LENCINAS J. D.

Buenos Aires

Simposio; 29th International Symposium on Remote Sensing of Environment.; 2002

ISRSE

To increase the efficiency, remote sensing data has been used in combination with terrestrialsamples for forest inventories in multiphase sampling methods. The present study deals with two-phase sampling in two large strata of Nothofagus forests in Patagonia, Argentina. The study area is almost inaccessible, mountainous and rugged.In the first sampling phase, the auxiliary variables were determined on satellite sampling plots. The reflexion values from a 5 ´ 5 pixel window of the SPOT Xi imagery were extracted and two vegetation indices were calculated. In the second phase, terrestrial subsamples were taken and the volume measured as accurately as possible. The field units are permanentclusters consisting of 5 circular sample plots, 500 m2 each. Corresponding data from the first and second phase were used to fit a regression model, which in turn was used to estimate stand volume. The accuracy and cost of the applied two-phase sampling werecompared with those of conventional one-phased terrestrial sampling. There was a clear relationships between vegetation indices and stand volume. Although the coefficients ofdetermination were not high (< r2 = 0,58), the regression was significant at the 95% level. Two-phase sampling was more accurate than one-phased sampling. Two-phase sampling represents an efficient alternative for estimating stand volume and monitoring large areas, especially in inaccessible forest regions. In order to increase accuracy some suggestionsfor improvement are discussed.