Space-Borne Retrieval of Solar-Induced Plant Fluorescence and its Relationship to Photosynthetic Parameters

Abstract

Studies have shown that chlorophyll fluorescence is directly linked to the photosynthetic efficiency of plants. The excess absorbed energy by leaves which has not been used in photosynthesis is re-emitted to the environment, either as heat or fluorescence. Therefore, any potential stress in plants is technically visible through monitoring fluorescence and the Solar-Induced plant Fluorescence (SIF) can thus be monitored as an indicator for vegetation growth and health status. SIF is a broad band spectral feature exhibiting two maxima at about 680 and 740 nm respectively, also known as red and far-red SIF. In the recent decades, there have been several studies addressing SIF, its importance and approaches to measure its value over vegetated regions. Among several measurement approaches, satellite-based remote sensing of SIF is particularly valuable, since the covered (spatial) area can be explicitly larger than is the case with in-situ measurements. With current space-borne instruments, even a full global coverage is attainable within a few days. In the framework of this thesis, two novel methods have been developed, tested and utilized to retrieve SIF from hyper-spectral satellite measurements. In particular, the first developed method, makes use of the Fraunhofer absorption lines in the far-red spectral region (748.5 - 753 nm), to retrieve SIF via its in-filling effect on these absorption lines. However, the satellite-based remote sensing spectrometers, used in this work, typically exhibit an additive spectral feature, which is not fluorescence. This is often accompanying the actual SIF retrieval and can significantly deteriorate the results. To account for this effect, a correction method has been developed and is combined with the retrieval algorithm. The model-based sensitivity studies confirmed the feasibility of the method to disentangle SIF from this additive feature. Additionally, the potential influences of the atmospheric and measurement conditions on the retrieval results have been assessed. Finally, the method has been applied to ten years of SCIAMACHY data and the retrieved results have been mapped on seasonal base. On a global scale, the obtained values are between 0 to 4 mW ma 2 sra 1 nma 1 . In absence of large area ground based validation data, final judgment of the results obtained in the framework of this study, is not possible. Alternatively, comparison of the achieved results with those published by the US National Aeronautics and Space Administration (NASA) Goddard Space Flight Center (GSFC) has been performed. Despite some differences, the comparison also exhibited close results, both qualitatively and quantitatively. It should be noted that comparisons among the retrievals provided by other research groups (not only GSFC) over the same spatial region is also variable depending on the instruments and methods utilized (ranging on average from a few tenths to more than 4 mW ma 2 sra 1 nma 1 ). To further assess the reliability of retrieved SIF, monthly average values have been compared to ground-based flux-tower measurements of Absorbed Photosynthetically Active Radiation by plants (APAR) and Gross Primary Production (GPP), for a time span of several years. The agreement between the seasonal trends of SIF and these parameters was significant. Although the main focus of this PhD work was on retrieving SIF in the far-red wavelength region using a spectral micro-window, there are clear scientific benefits in having an estimation over the full spectral emission range of SIF. Therefore, the second retrieval method, developed in the framework of this work, was to obtain the full spectrum of the emitted SIF by retrieving the leaf and canopy parameters, utilizing a combination of two radiative transfer models. The model-based studies showed the feasibility of the method to retrieve SIF with high accuracy. Moreover, the first results of applying this approach on GOME-2 measurements demonstrated promising outcomes. Examples of the fit quality and retrieved SIF over two different vegetation coverage types have been presented in this thesis, showing clear applicability of the method to retrieve SIF over its full spectral emission range and the potential to derive other vegetation parameters (e.g. Chlorophyll content of the leaves and the so-called leaf area index).