Two large-scale forest scenario modelling approaches for reporting CO2 removal: a comparison for the Romanian forests.

BACKGROUND: Forest carbon models are recognized as suitable tools for the reporting and verification of forest carbon stock and stock change, as well as for evaluating the forest management options to enhance the carbon sink provided by sustainable forestry. However, given their increased complexity and data availability, different models may simulate different estimates. Here, we compare carbon estimates for Romanian forests as simulated by two models (CBM and EFISCEN) that are often used for evaluating the mitigation options given the forest-management choices. RESULTS: The models, calibrated and parameterized with identical or harmonized data, derived from two successive national forest inventories, produced similar estimates of carbon accumulation in tree biomass. According to CBM simulations of carbon stocks in Romanian forests, by 2060, the merchantable standing stock volume will reach an average of 377 m3 ha-1, while the carbon stock in tree biomass will reach 76.5 tC ha-1. The EFISCEN simulations produced estimates that are about 5% and 10%, respectively, lower. In addition, 10% stronger biomass sink was simulated by CBM, whereby the difference reduced over time, amounting to only 3% toward 2060. CONCLUSIONS: This model comparison provided valuable insights on both the conceptual and modelling algorithms, as well as how the quality of the input data may affect calibration and projections of the stock and stock change in the living biomass pool. In our judgement, both models performed well, providing internally consistent results. Therefore, we underline the importance of the input data quality and the need for further data sampling and model improvements, while the preference for one model or the other should be based on the availability and suitability of the required data, on preferred output variables and ease of use.

Biomass data for young, planted Norway spruce (Picea abies (L.) Karst.) trees in Eastern Carpathians of Romania.

Tree biomass data are essential for developing the biomass allometric models that are necessary for estimating carbon stock and for monitoring changes in forest biomass. In this 'data article' biomass records are presented for 240 Norway spruce trees (Picea abies (L.) Karst.). Trees were between 4 and 15 years of age and were sampled from 24 pure plantations located in Eastern Carpathians of Romania. Ten trees were sampled from each plantation using a cluster sampling method. For each tree, diameter at root collar height (D) and tree height (H) are provided as potential predictors for biomass. Oven-dried biomass is also recorded for the following partitions: stem (ST); branches (BR); needles (ND); roots (RT); as well as their combinations representing total aboveground biomass (AGB) and overall tree biomass (TB). Sampled trees were between 0.6 and 10.0 cm in diameter and between 53.0 and 552.0 cm in height. Total tree biomass ranged between 0.019 and 15.53 kg/tree. This dataset is related to the research article entitled "Site-effects on biomass allometric models for early growth plantations of Norway spruce (Picea abies (L.) Karst.)" (Dutca et al., 2018) [1].

Using clustered data to develop biomass allometric models: The consequences of ignoring the clustered data structure.

This paper investigates the consequences of ignoring the clustered data structure on allometric models. Clustered data, in the form of multiple trees sampled from multiple forest stands is commonly used to develop biomass allometric models. Of 102 reviewed papers published between 2012 and 2016 that reported biomass allometric models, 84 (82%) have used a clustered sampling design. However, in as many as 80% of these, the clustered data structure was ignored, potentially violating the independence assumption in ordinary least squares methods. The consequences of ignoring clustered data structure were empirically validated using two clustered biomass datasets (of 110 and 220 trees, with the cluster size of 5 and 10 trees respectively). We showed that when Intraclass Correlation Coefficient (ICC) was higher than zero, ignoring the clustered data structure returned underestimated standard errors, affecting further the confidence interval and t-test results. The underestimation level depended on ICC (which shows the variance proportion that was caused by the forest stand) and on cluster size (the number of trees sampled from one forest stand). We also showed that using first-order autocorrelation tests, such as the traditional Durbin-Watson statistic, to detect the autocorrelation due to clustered structure could be misleading as the test may show lack of autocorrelation even though ICC is different from zero. In conclusion, when ICC is higher than zero, ignoring the clustered data structure yields over-confident biomass predictions (due to underestimated confidence interval) and/or incorrect research conclusions (due to overestimated evidence against null hypothesis in t-test). Therefore, using a modelling approach that accounts for the hierarchical structure of the data is highly recommended when any form of clustering can be identified, even if the autocorrelation is not significant.