Global sensitivity analysis of the SCOPE model: What drives simulated canopy-leaving sun-induced fluorescence?

This study provides insight into the key variables that drive sun-induced chlorophyll fluorescence (SIF) emanating from vegetation canopies, based on a global sensitivity analysis (GSA) of the Soil-Canopy Observation of Photosynthesis and Energy (SCOPE) balance model. An updated version of the SCOPE model was used here (v1.53) which contains novel leaf physiological modules for determination of the steady state fluorescence yield: a photosynthesis model coupled with (a) submodels having empirically derived relationships, identified as TB12 for unstressed and TB12-D for drought conditions and (b) a mechanistic (MD12) submodel based on theoretical relationships. By inspecting Sobol's total order (main effect and all the interactions) sensitivity index (STi) rankings, the influential and non-influential variables were determined. Two experiments were conducted for the different leaf physiology modules in SCOPE considering (1) only vegetation variables, and (2) all SCOPE variables, i.e., including micrometeorological, aerodynamic and geometry variables. Considering TB12-D STi results using only vegetation input variables, the canopy-leaving broadband (641–800 nm) SIF variability was determined mainly by leaf optical properties and canopy structural variables. The most important variables were (with decreasing importance) leaf chlorophyll content (Cab), leaf inclination (LIDFa) and leaf area index (LAI). These three variables alone determined 77.9% of the SIF variability. Vcmo, the variable related to photosynthetic capacity, determined 11.4% of overall SIF variability, and its importance declined considerably when moving from the first emission peak (SIFred; with maximal relevance of 17.9% at 676nm) to the second emission peak (SIFNIR; e.g., 9.6% at 740 nm). Stronger relationshipswith Vcmo were obtained when retrieving the full broadband SIF flux and calculating total fluorescence yield (Fyield, determined as the integral of the hemispherical broadband SIF flux divided by the total absorbed PAR), of which 35% of the variability was influenced by Vcmo. Using the TB12 submodel, the major drivers of SIF flux were similar to TB12-D except that Vcmo accounted for very little (b2%) variability. The MD12 submodel identified the components of long-term PSII photoprotection and photodamage as the dominant factors for SIF variability: these two variables alone accounted for 51.4% of the variability of SIF flux and 61% of Fyield, whereas Vcmo explained only 9.7% and 10.9% of variability in SIF flux and Fyield, respectively. Analysis of the relative importance of all SCOPE variables revealed that in addition to the key vegetation variables, micrometeorological variables were important in driving SIF variability, especially incoming shortwave radiation (Rin) and to a lesser extent air temperature (Ta), atmospheric vapor pressure (ea) and atmospheric CO2 concentration (Ca). Their impact further reduced the relative importance of Vcmo. The GSA experiments led to the following conclusions: (1) explicit knowledge of key variables driving the SIF flux is essential in order to achieve unbiased SIF interpretation related to photosynthetic activity at local and global scales; (2) information related to photosynthetic activity is found more in the first emission peak (SIFred) than in the second peak (SIFNIR), and more in the full broadband SIF emission, which allows calculation of Fyield, than in individual wavebands.