Are carbon flux products reliable?


Authors: Xianglin Tian, Francesco Minunno, Laura Sirro, Annikki Mäkelä, Tuomas Häme

1. What is the tool used for carbon flux computation?

In Forest Flux project the carbon flux computation is mainly based on PREBAS (Figure 1), which is a forest growth model developed at the University of Helsinki. In PREBAS two different models PRELES (PREdict Light-use efficiency, Evapotranspiration and Soil water) model and CROBAS (Tree growth and CROwn BASe from carbon balance) are coupled together to predict the growth and the carbon and water balance of the forest. To complete the carbon balance at ecosystem level PREBAS has been combined with a soil model (YASSO: Yet Another Simulator of Soil Organic matter). CROBAS provides estimates of fAPAR (the fraction of absorbed photosynthetically active radiation) that is used in PRELES to compute gross primary production (GPP). GPP is then used by CROBAS to estimate forest growth. The stand structural variables and the biomass components of the forest are updated. CROBAS also estimates the litter production that is the input of YASSO.

2. Why choose PREBAS?

The advantage of PREBAS is that it concerns physiological mechanisms and can theoretically be extrapolated to novel sites or to future climates, whereas traditional statistical models are often limited to the data conditions used to develop the model. However, unlike the parameters obtained using regression, the parameters of a process-based model are usually determined by knowledge or references about plant traits. It can be hard to ascertain the inherent variability of those parameters. Thus, we apply an inverse modelling approach, Bayesian calibration, to adjust model parameters and processes according to their ability to reproduce stand-level field observations (Figure 2). Benefited from Bayesian approaches, PREBAS has been calibrated and evaluated to varying tree species and environmental conditions for simulating ecosystem fluxes and forest growth.

4. How uncertain are the predictions?

The total uncertainty of the carbon storage and flux services is a compilation of uncertainties in the input datasets and model uncertainties. The uncertainties from PREBAS are quantified during the model calibration process by means of Bayesian statistics. For instance, the prediction uncertainty of GPP covered the variation in daily measurements (Figure 3). Measurement uncertainty, i.e., random error, comprised more than 90% of the predictive uncertainty of daily GPP.
Soil carbon estimates have a considerably higher degree of uncertainty than other variables. The initial soil carbon storage as input is difficult to obtain and has a high variation on spatial and temporal scales. When soil carbon measurements are not available, we apply the YASSO model to estimate the equilibrium of soil carbon storage with the average carbon input that is derived from PREBAS (Figure 4). Even though the variation of management schemes and site conditions propagates large uncertainty into the equilibrium calculation, soil carbon estimates from YASSO are within a reasonable range.

Relevant literature:

Liski J, Palosuo T, Peltoniemi M, Sievänen R. 2005. Carbon and decomposition model Yasso for forest soils. Ecological modelling, 189(1-2): 168-182.

Mäkelä A. 1997. A carbon balance model of growth and self-pruning in trees based on structural relationships. Forest Science, 43(1): 7-24.

Minunno F, Peltoniemi M, Härkönen S, Kalliokoski T, Makinen H, Mäkelä A. 2019. Bayesian calibration of a carbon balance model PREBAS using data from permanent growth experiments and national forest inventory. Forest Ecology and Management, 440: 208-257.

Peltoniemi M, Pulkkinen M, Aurela M, Pumpanen J, Kolari P, Mäkelä A. 2015. A semi-empirical model of boreal-forest gross primary production, evapotranspiration, and soil water-calibration and sensitivity analysis. Boreal Environment Research, 20 (2): 151–71.

Tian X, Minunno F, Cao T, Kalliokoski T, Mäkelä A. 2020. Extending the range of applicability of the semi‐empirical ecosystem flux model PRELES for varying forest types and climate. Global Change Biology 26(5): 2923–2943.