Partitioning Net Ecosystem Exchange (NEE) of CO₂ Using Solar‐Induced Chlorophyll Fluorescence (SIF)

Abstract

Accurate partitioning of net ecosystem exchange (NEE) of CO2 to gross primary production (GPP) and ecosystem respiration (Reco) is crucial for understanding carbon cycle dynamics under changing climate. However, it remains as a long‐standing problem in global ecology due to lack of independent constraining information for the two offsetting component fluxes. solar‐induced chlorophyll fluorescence (SIF), a mechanistic proxy for photosynthesis, holds great promise to improve NEE partitioning by constraining GPP. We developed a parsimonious SIF‐based approach for NEE partitioning and examined its performance using synthetic simulations and field measurements. This approach outperforms conventional approaches in reproducing simulated GPP and Reco, especially under high vapor pressure deficit. For field measurements, it results in lower daytime GPP and Reco than conventional approaches. This study made the first attempt to demonstrate SIF's potential for improving NEE partitioning accuracy and sets the stage for future efforts to examine its robustness and scalability under real‐world environmental conditions.

A Data-Driven Approach to Partitioning Net Ecosystem Exchange Using a Deep State Space Model

“…The study performed by Lee et al [9] uses continuous stable isotope measurements in a Pacific Northwest Douglas-fir forest ecosystem for NEE partitioning and also reports estimated daytime R eco lower than the conventional approaches. Furthermore, similar results were obtained in the most recent study on the topic by Kira et al [10], in which the authors used a parsimonious Solar-Induced Chlorophyll Fluorescence (SIF)-based approach for NEE partitioning and examined its performance using synthetic simulations and field measurements. Time series forecasting using deep learning.…”

“…Now we can interpret our daytime R eco forecasts having in mind that there is no ground truth with which to compare these forecasts directly. We compare our findings to those of the recent empirical NEE partitioning methods [6], [9], [10] which suggest that R eco is lower during the day than during the night, and can be attributed to the Kok effect [8]. In Fig.…”

The physiological basis for estimating photosynthesis from Chla fluorescence

“…Note that FvCB has larger PEU than MLR‐SIF, not because the former has a structural weakness but because the latter takes advantage of an observable photosynthetic functional ‘shortcut’ (SIF) (Gu et al ., 2019), that is, at least in theory, SIF makes estimating photosynthesis simpler, although there are still difficult aspects that need to be resolved through continuing research. MLR‐SIF effectively utilizes the mechanistic information in observed SIF to estimate photosynthesis in a forward way, independent of other existing photosynthesis estimation approaches that are known to have different levels of uncertainty (Wehr et al ., 2016; Keenan et al ., 2019; Kira et al ., 2021). Independent estimates are essential for confidence building as photosynthesis cannot be measured directly beyond a single leaf.…”

From remotely‐sensed solar‐induced chlorophyll fluorescence to ecosystem structure, function, and service: Part II—Harnessing data

“…For example, the SIF‐GPP scaling factor is derived from regressing SIF against GPP either from global gridded products or inferred from in‐situ NEE of CO 2 measured with EC techniques. Global gridded GPP products are highly uncertain, whereas the latter, commonly treated as the ground “truth”, is actually imprecisely partitioned with well documented biases (Keenan et al, 2019; Kira et al, 2021; Wehr et al, 2016). SIF‐GPP scaling derived from these GPP datasets was then used to back‐calculate GPP, which is essentially circular, and inherits uncertainties in the original GPP.…”