Geostatistics applied to growth estimates in continuous forest inventories

Abstract

This study addresses the use of geostatistics to ensure sampling representativeness in a continuous forest inventory (CFI). A database of 89 permanent plots was used. Dominant height was employed for stratification by ordinary kriging. The correlation between the values estimated by kriging was calculated for all measurement occasions to define the earliest age for stable stratification. Growth estimates were obtained by simple random sampling (SRS) and poststratification. Mean volume and volume growth values were computed along with their sampling errors for the four growth intervals. The impact of decreased sampling intensity on volume growth precision was based on the poststratification. Site index modeling considered the reduced and full databases. The earliest age for reliable stratification was 3.1 years. Poststratification resulted in greater precision in volume growth estimates. A 40% decrease in sampling intensity did not result in significant losses in precision. Site index modeling with the reduced database had the same precision when the full database was used. Geostatistics improved the precision and reliability of the CFI statistics, because it allows the segregation of different forest sites and ensures CFI representativeness while improving the precision of the estimates and allowing decreased sampling intensity.