Plant productivity and evaporation from remote sensing
Due to the COVID-19 crisis the PhD defence of Egor Prikaziuk will take place (partly) online.
The PhD defence can be followed by a live stream.
Egor Prikaziuk is a PhD student in the department of Water Resources (WRS). His supervisor is dr.ir. C. van der Tol from the Faculty of Geo-Information Science and Earth Observation (ITC).
The quantification of carbon and water fluxes between the land surface and the atmosphere has practical applications in agriculture and it is fundamental for global monitoring, especially in view of climate change. The main process of carbon uptake is photosynthesis, and the main water flux from the land to the atmosphere is evaporation, where we can differentiate transpiration as the flux through pores of living organisms, mostly stomata in plants. Remote sensing products can be used for both photosynthesis and evapotranspiration fluxes estimates, but modelling is an inevitable step. This thesis assesses the Soil-Canopy-Observation of Photosynthesis and Energy fluxes (SCOPE) model for flux simulation in theory, in agricultural field experiments, and globally for different ecosystems.
Global sensitivity analysis
In the first step a global sensitivity analysis (GSA) of SCOPE was carried out to evaluate the theoretical capability of the model to be driven by satellite data, and the effects of the atmosphere on this capability. The innovative element of the study was the sensitivity analysis at satellite level, including the atmosphere. Previous GSAs of SCOPE were limited to top of canopy (or bottom of atmosphere) level. The need for atmospheric analysis was also dictated by the choice of the Sentinel-3 satellite, the data of which were only available at the top of atmosphere level then. Another innovation - and also the motivation for the use of Sentinel-3 - was the GSA in thermal infrared domain (TIR). Previous studies demonstrated and our study confirmed that the link between thermal radiance emission and the key determinant of photosynthesis rate (plant biochemical property of the CO2 assimilating enzyme Rubisco; maximum Rubisco carboxylation capacity, Vcmax) is apparent in the output of the SCOPE model. We did not succeed in retrieving Vcmax from TIR radiance with numerical optimization, but it definitely requires further investigation.
Large scale study
The following step was the global evaluation of SCOPE-simulated gross primary productivity (GPP) and evapotranspiration (ET) across herbaceous and woody plant functional types in Europe. Various seasonally static and dynamic values of Vcmax were evaluated. Surprisingly, for most of the studied ecosystems the default parametrization performed the best for the GPP flux. The largest errors were found in dry regions, which we attribute to the lack of a soil water balance computation and soil moisture-transpiration feedback in SCOPE model. Another source of uncertainty was the varying ground pixel position of Sentinel-3, inherent from the varying observation orbit. A separate chapter in the thesis discusses the constraints on the homogeneity of the area of interest in the application of time series of Sentinel-3.
Local scale study
The final step was the in-field application of SCOPE for retrieval from hyperspectral data and potato plant productivity simulation from the CO2 assimilation flux. The performance of the retrieval algorithm was not as successful as on the synthetic dataset, in particular only leaf chlorophyll content (Cab) and leaf area index (LAI) were retrieved with acceptable accuracy. At the same time, exactly those two parameters are the main drivers of photosynthesis, therefore the retrieval of LAI and Cab enabled us to model potato yield as the sum of mean daily net primary productivity (NPP) fluxes corrected with the harvest index.
This thesis confirms that SCOPE can be applied not only as a radiative transfer model, as it mostly has been applied up to now, but also as photosynthesis and energy balance model. SCOPE provides physically consistent approach where the satellite reflectance input and traits and fluxes output occur within a single model. The release of the faster version 2.0 facilitates operational applications further. The future improvement of the SCOPE model might include (1) a soil water balance representation relevant for drought-subjected ecosystems, where ET is determined by water availability, rather than energy availability, and (2) leaf clumping relevant for forests and row crops, whose leaves form patches in contrast to grasses, where leaves are evenly distributed. Finally, the link between thermal emitted radiance and Vcmax should be investigated in the coming research.