Intercomparison of methods to estimate gross primary production based on CO2 and COS flux measurements

Abstract

Separating the components of ecosystem-scale carbon exchange is crucial in order to develop better models and future predictions of the terrestrial carbon cycle. However, there are several uncertainties and unknowns related to current photosynthesis estimates. In this study, we evaluate four different methods for estimating photosynthesis at a boreal forest at the ecosystem scale, of which two are based on carbon dioxide (CO₂) flux measurements and two on carbonyl sulfide (COS) flux measurements. The CO₂-based methods use traditional flux partitioning and artificial neural networks to separate the net CO₂ flux into respiration and photosynthesis. The COS-based methods make use of a unique 5-year COS flux data set and involve two different approaches to determine the leaf-scale relative uptake ratio of COS and CO₂ (LRU), of which one (LRU_CAP) was developed in this study. LRU_CAP was based on a previously tested stomatal optimization theory (CAP), while LRU_PAR was based on an empirical relation to measured radiation. For the measurement period 2013-2017, the artificial neural network method gave a GPP estimate very close to that of traditional flux partitioning at all timescales. On average, the COS-based methods gave higher GPP estimates than the CO₂-based estimates on daily (23% and 7% higher, using LRU_PAR and LRU_CAP, respectively) and monthly scales (20% and 3% higher), as well as a higher cumulative sum over 3 months in all years (on average 25% and 3% higher). LRU_CAP was higher than LRU estimated from chamber measurements at high radiation, leading to underestimation of midday GPP relative to other GPP methods. In general, however, use of LRU_CAP gave closer agreement with CO₂-based estimates of GPP than use of LRUPAR. When extended to other sites, LRU_CAP may be more robust than LRU_PAR because it is based on a physiological model whose parameters can be estimated from simple measurements or obtained from the literature. In contrast, the empirical radiation relation in LRU_PAR may be more site-specific. However, this requires further testing at other measurement sites.

Figure 1. Median diurnal variation of T_a, T_soil, PAR, SWC, and VPD in different months during the measurement period 2013–2017.

Figure 2

Median diurnal variation of GPP partitioned using a combined nighttime–daytime method (GPP_NLR, purple line), GPP from artificial neural networks (GPP_ANN, pink line), GPP from COS flux measurements with LRU determined according to Kooijmans et al. (2019) (GPP_COS,PAR, dark blue line), and GPP from COS flux measurements using a new approach for LRU (Sect. 2.3.4, GPP_COS,CAP,light blue line) in different months during the measurement period 2013–2017. Averaging was done to the same data points, and only months with more than 55% of data coverage were included.

Figure 3

Diurnal variation of the difference of GPP_ANN (pink), GPP_COS,PAR (dark blue) and GPP_COS,CAP (light blue) to the reference GPP_NLR in different months during the measurement period 2013–2017. Averaging was done to the same data points and only months with more than 55% of data coverage were included.

Figure 4

Scatter plots of GPP_ANN, GPP_COS,PAR, and GPP_COS,CAP against GPP_NLR in 30 min, daily, and monthly timescales. The colour of data points in 30 min and daily scatter plots indicates the data density, lighter colours indicating higher point density than dark ones.

Figure 5

Relative (a) and absolute (b) difference of daily GPP_ANN (pink), GPP_COS,PAR (dark blue), and GPP_COS,CAP (light blue) to GPP_NLR in different months, averaged over the whole measurement period 2013–2017. Bars represent the median difference, and whiskers show the 25th and 75th percentiles. Numbers on top of the bars indicate how many daily flux data points have been used for calculating the medians. Median differences have been calculated using the same number of data points for each method in each month.

Figure 6

Distribution (bars) and probability density functions (lines) of daily average (a) GPP_NLR, (b) GPP_ANN, (c) GPP_COS,PAR, and (d) GPP_COS,CAP. All probability density functions are combined in (e) for better comparison.

Figure 7

Responses of the different GPP estimates (GPP_NLR (purple), GPP_ANN (pink), GPP_COS,PAR (dark blue), and GPP_COS,CAP, light blue) to environmental parameters – photosynthetically active radiation (a, d), air temperature (b, e), and vapour pressure deficit (c, f) – in spring (a–c) and summer (d–f). Data are binned to 12 equally sized bins (same number of data points in each bin), and all GPPs have the same data coverage. Only measured (non-gap-filled) 30 min flux data were used, and GPP was filtered to include only PAR > 700 μmol m² s⁻¹ in responses to T_a and VPD to avoid simultaneous correlation with PAR.