Assessing chlorophyll fluorescence and light use efficiency in boreal forest canopies using imaging spectroscopy data

The photosynthetic light use efficiency (LUE) is used to estimate the rate of photosynthesis in green vegetation. LUE is generally defined as the carbon uptake divided by absorbed photosynthetically active radiation (PAR). Several proxies for LUE estimation from remotely sensed data have been introduced, while the most promising results have been achieved by the Photochemical Reflectance Index (PRI) and chlorophyll fluorescence (ChlF). Both PRI and ChlF can serve as a proxy for photosynthetic downregulation and photoprotection, and can theoretically be measured and calculated from hyperspectral remote sensing data. Fluorescence and PRI were measured at leaf-level for sample trees of Pinus sylvestris and Betula pendula during two field measurement campaigns at Hyytiälä Forestry Field Station in 2014 and 2015. The spectrometer measurements were performed at crown height, separately for sunlit and shaded foliage. The measurements were taken at different hours of day to observe the diurnal variation of fluorescence and PRI. Along with the spectral measurements, CO2 exchange was constantly measured in shoot chambers attached to the sample trees. The diurnal trend of CO2 assimilation and LUE was then projected against the spectral indicators. An airborne campaign was conducted concurrently with the field measurement campaign in 2015. Hyperspectral data (< 5 nm spectral, 60 cm spatial resolution) was collected over the study area using AISA Eagle II line scanner. ChlF was estimated from the data using Fraunhofer line discrimination (FLD) approach. A change in ChlF between morning and noon acquisitions was estimated using different flight lines covering the same area. The reliability of the estimation was tested in several ways including noise reduction and a change analysis of non-fluorescent targets. Moreover, the non-physiological impact of shadow fraction on apparent fluorescence was estimated. The leaf-level results show a distinctive diurnal trend in fluorescence values measured with the spectrometer. The diurnal variation of PRI appears more unpredictable within the sample trees. Statistically, fluorescence values and PRI were only able to predict CO2 assimilation and LUE during the midday measurements with higher values of PAR, fluorescence performing slightly better. The results obtained at canopy level show an increase in radiance emitted as fluorescence and a decrease in PRI between morning and midday acquisitions. The change was observed in all the canopy types, including low vegetation, broad-leaved forests and coniferous forests. The average change of FLD-derived fluorescence (F763) between different canopy types varied between +24.0 units (14.9 %) and +6.3 units (4.9 %). Furthermore, it was observed that illumination conditions and shadow fraction had a major impact on F763 values, which must be considered when estimating ChlF using the FLD approach.