Improving the accuracy of timber volume and basal area prediction in heterogeneously structured and mixed forests by automated co-registration of forest inventory plots and remote sensing data

Accurate georeferencing is essential if forest indicators, such as timber volume, are to be modelled and predicted area-wide by establishing a linkage between local inventories remote sensing data. Nevertheless, due inaccuracies in global navigation satellite system (GNSS) measurements under closed canopy, determining the exact position of sample-plots inventory major challenge. In this study different methods were evaluated, each which designed improve co-registration field data inventory. The evaluated two areas Switzerland (Bremgarten Zurich), all have heterogeneously structured mixed forests. A simple algorithm that searches for best match tree-tops, detected data, an tree-top point-cloud led unsatisfactory results. failure was primarily related lack accurate single-tree-identification (STI) airborne laser scanning (ALS) (10 points m−2) on deciduous stands. These hampered successful sources. To omit step, rely comparison artificial canopy height model (CHM) calculated from CHM generated ALS tested. Correlating CHMs made it possible identify plausible positions within search range. quality assessed leave-one-out cross-validated root-mean-squared error (RMSE) volume estimate subsequently calibrated regression model. results achieved with method tree-crowns spheres applied correlation metric named SQDIFF_NORMED. Co-registration increase accuracy estimates Bremgarten 31 RMSE-%, i.e. 134.4 m3 ha−1 (without co-registration) 92.6 co-registered positions. Additional testing identified superior larger dataset canton Zurich, confirmed these There, RMSE basal-area improved 10 13.56 m2 12.18 ha−1. For CHM-based methods, integrating information deciduous–evergreen (DecEv) raster positional but not overall predictive power models.