Konishii fir (Cunninghamia konishii Hayata) is an important conifer in Taiwan. The purpose of this study was to predict stand volume (V), aboveground biomass accumulation (AGB), and aboveground carbon storage (AGCST) for a Konishii fir plantation. This study was located at the Huisun Experimental Forest Station of Nantou County located in central Taiwan. Four sample plots, each with an area of 0.05 ha, were installed and surveyed from 29 June to 2 July 2020. Two models, the diameter distribution model (DDM) and allometric model (AM), were used to predict V, AGB, and AGCST. Each item predicted by these two models was compared by the paired sample t-test. We employed the Weibull function to quantify stand diameter distribution and this function can effectively quantify diameter distribution, because all plots passed the examination by the Kolmogorov–Smirnov test (non-significant). Therefore, the Weibull function was suitable for developing the DDM. The predicted V, AGB, and AGCST were 538.43 ± 140.52 m3 ha−1, 203.25 ± 52.79 Mg ha−1, and 100.85 ± 26.30 Mg ha−1 by DDM; and 555.90 ± 145.42 m3 ha−1, 209.10 ± 51.25 Mg ha−1, and 103.78 ± 25.51 Mg ha−1 by AM, respectively. Each item was insignificantly different between DDM and AM, indicating similarity in results for both predictions. Meanwhile, using DDM is advantageous, as it can provide more yield information in diameter classes; therefore, this approach was recommended for yield prediction of the Konishii fir plantation.
Accurate estimation of forest carbon storage is fundamental to climate change mitigation. However, the highly heterogeneous site conditions, together with the biological limitations of conventional linear and polynomial models, have posed considerable challenges for estimating carbon accumulation in voluntary carbon reduction projects. Based on data from 1190 permanent sample plots across Taiwan, this study developed a site-class-based carbon storage prediction model for coniferous and broad-leaved plantations in Taiwan by using the nonlinear Chapman–Richards growth function. The estimated model parameters indicated distinct carbon accumulation patterns for the two forest types. Coniferous plantations exhibited a higher model-estimated asymptotic carbon storage (1232.506 Mg CO2 ha−1), whereas broad-leaved plantations accumulated carbon more rapidly during the early stages of stand development. Because a single base model may produce substantial prediction errors when applied over large spatial scales, the stands were classified into high-, medium-, and low-site classes by using a reference age of 30 years. Site classification reduced the out-of-sample root mean square error by 56.8% for coniferous and 50.8% for broad-leaved plantations under nested cross-validation. The resulting site class models may provide useful quantitative tools for establishing baseline and project scenarios in voluntary forest carbon reduction projects.