RELATIONSHIPS BETWEEN TOTAL TREE HEIGHT AND DIAMETER AT BREAST HEIGHT FOR TROPICAL PEAT SWAMP FOREST TREE SPECIES IN ROKAN HILIR DISTRICT, RIAU PROVINCE

Reliable information on total tree height (H) is fundamental in forest resource management and forest ecological studies, including in forest biomass assessment. Adding an H variable can improve the performance of the biomass allometric equations by reducing the average deviation significantly. However, measuring H is relatively complex, less accurate, time consuming, and expensive. Thus, H is only measured for sampled trees within the plots, whilst diameter at breast height (DBH) is commonly measured for each tree during the forest inventory. The missing H information is usually estimated based on a stand-specific allometric relationship between H and DBH (H-D model) constructed from sampled trees. Despite extensive studies on H-D model for boreal forests and for single-species/plantation forests, few studies have focused on tropical forests. Furthermore, relationships for peat swamp forest tree species, and especially those in Indonesia, have not been widely published. Thus, the objective of this study was to develop site-specific H-D models for tropical peat swamp forests using linearized and non-linear regression functions. The results indicated that the non-linear models outperformed the linearized models based on the statistical parameters and the biological criteria. The modified logistic function (Model 7) is recommended for estimating H in the study area as it has comparable model performances to the exponential function (Model 6) and passed the point diameter-height of (0, 1.3). However, all five non-linear models performed equally well and the differences between them were trivial. Further improvements are needed to improve the accuracy, the predictive ability and the geographical applicability of the models by grouping the species, adding stand variables and (or) using advanced techniques of mixed-effect modelling. In addition, model validation should be carried out prior to their application by collecting a new dataset from the forest being studied