Photosynthesis is the key process sustaining life on Earth. The efficiency of photosynthesis under varying environmental conditions is maintained by fast-responding regulatory mechanisms, namely chlorophyll a fluorescence emission (F) and non-photochemical quenching (NPQ). F is an electromagnetic radiation emitted at longer wavelengths (640-850 nm) than for excitation. NPQ is a complex process which includes the de-epoxidation of xanthophyll cycle pigments resulting in dissipation of excess energy as heat. This translates into changes in leaf absorbance at 531 nm, which can be detected with the photochemical reflectance index (PRI). Proximal sensing is a powerful tool for exploitation of subtle signals related to the downregulation of photosynthesis. However, F and PRI are also influenced by canopy structural and biochemical properties, illumination conditions and solar-view geometry, and its explicit interpretation is still challenging. The aim of my Ph.D. project was to exploit the methods of proximal sensing of vegetation to elucidate a link between continuous hyperspectral measurements of optical indicators related to plant physiology (F and PRI) and vegetation functioning. Multi-angular measurements were used to characterize the anisotropic response of far-red (F760), red (F687) fluorescence and PRI. I present an extensive experimental dataset of multi-angular measurements of F, PRI and R collected during a day over four different vegetation targets. Directional responses of F760, F687 and PRI in horizontally homogeneous canopies are characterized by increased values in the backward scattering direction with a maximum in the hotspot and decreased values in the forward scatter direction. The shape of F and PRI angular response is mostly controlled by leaf inclination distribution function (LIDF), while a magnitude and shape of the hotspot peak is sensitive to a combination of factors - leaf area index (LAI) and the ratio of leaf width to canopy height. Quantitative evaluation of the impact of anisotropy on fluorescence apparent yields (Fy*760 and Fy*687) showed, that, on average, off-nadir observations were overestimated by 20-67% in the backscatter direction and underestimated in other directions by 10-45%. The archived results reinforce the importance of maintaining nadir observation geometry for continuous measurements of F and PRI. Additional complexity for the interpretation of time series of optical signal lies in time-scale dependant vegetation dynamics. Here I test the applicability of highly-adaptive time series decomposition technique Singular Spectrum Analysis (SSA) for disentangling slow and fast varying components of vegetation dynamics in time series of F760, Fy*760 and PRI. First, the proof of concept was developed based on spectral and flux half-hourly time series realistically simulated with the SCOPE model. The simulations included two outputs - featuring and excluding the effect of the xanthophyll cycle de-epoxidation and fluorescence efficiency amplification factor on PRI and F760, respectively. The decomposed slow varying SSA-components of total PRI, F and Fy*760 showed a good correlation with reference constitutive variability simulated excluding the effect of physiological modulation (PRI0 , F0,760, Fy*0,760). The accuracy of disentangled fast varying SSA-components was validated with the reference physiologically induced variables (∆PRI, ∆F), and with non-photochemical quenching (NPQ) and light-use efficiency (LUE). Then, the application of this methodology on a field dataset of spectral and flux time series collected in winter wheat field allowed to significantly improve the correlation of fast components of F and PRI with the fast component of LUE in comparison with original time series. Therefore, SSA-based approach is a promising tool for decoupling physiological information from continuous measurements of optical signal which can foster the use of automated proximal sensing systems.
La fotosintesi è il processo chiave che sostiene la vita sulla Terra. L'efficienza della fotosintesi in condizioni ambientali variabili è mantenuta da meccanismi di regolazione che rispondono velocemente, ovvero l'emissione della fluorescenza della clorofilla a (F) e il quenching non fotochimico (NPQ). F è una radiazione elettromagnetica emessa a lunghezze d'onda più lunghe (640-850 nm) rispetto a quelle di eccitazione. NPQ è un processo complesso che comprende la de-epossidazione dei pigmenti del ciclo della xantofilla con conseguente dissipazione dell'energia in eccesso sotto forma di calore. L'obiettivo del mio progetto di dottorato era quello di sfruttare i metodi di telerilevamento di prossimità della vegetazione per chiarire il collegamento tra misurazioni iperspettrali in continuo di indici ottici relazionati alla fisiologia delle piante (F e PRI) e la funzionalità della vegetazione. Sono state utilizzate misure multiangolari per caratterizzare la risposta anisotropa della fluorescenza nel rosso lontano (F760), nel rosso (F687) e del PRI. Presento un ampio dataset sperimentale di misurazioni multiangolari di F, PRI e riflettanza (R) raccolte durante cicli giornalieri su quattro diversi target di vegetazione. Le risposte direzionali di F760, F687 e PRI in canopy orizzontalmente omogenee sono caratterizzate da valori maggiori nella direzione di retrodiffusione con un massimo nell'hotspot e da valori minori nella direzione di diffusione in avanti. La forma della risposta angolare di F e PRI è per lo più controllata dalla funzione di distribuzione dell'inclinazione delle foglie (LIDF), mentre la grandezza e la forma del picco dell'hotspot sono sensibili ad una combinazione di fattori - l'indice di area fogliare (LAI) e il rapporto tra la larghezza delle foglie e l'altezza della canopy. La valutazione quantitativa dell'impatto dell'anisotropia sui rendimenti apparenti della fluorescenza (Fy*760 e Fy*687) ha mostrato che, in media, le osservazioni off-nadir sono sovrastimate del 20-67% in direzione della retrodiffusione e sottovalutate in altre direzioni del 10-45%. I risultati ottenuti corroborano l'importanza di mantenere una geometria di osservazione nadirale per le misure in continuo di F e PRI. Un ulteriore elemento di complessità per l'interpretazione delle serie temporali di segnali ottici risiede nella dipendenza temporale delle dinamiche della vegetazione. In tale contesto, questo lavoro verifica l'applicabilità di una tecnica di decomposizione delle serie temporali altamente flessibile chiamata Singular Spectrum Analysis (SSA) per disaccoppiare le componenti lente e veloci delle dinamiche della vegetazione nelle serie temporali di F760, Fy*760 e PRI. In primo luogo, è stata sviluppato un proof of concept sulla base di serie temporali spettrali e di flussi a intervalli di mezz’ora simulate realisticamente con il modello SCOPE. Le simulazioni comprendevano due output - considerando ed escludendo l'effetto del ciclo di de-epossidazione delle xantofille e il fattore di amplificazione dell’efficienza di fluorescenza rispettivamente su PRI e F760. Le componenti SSA lente decomposte di PRI, F760 e Fy*760 hanno mostrato una buona correlazione con la variabilità costitutiva di riferimento simulata escludendo l'effetto della modulazione fisiologica. L'accuratezza delle componenti SSA veloci è stata validata con variabili fisiologicamente indotte di riferimento (∆PRI, ∆F), e con il NPQ e l'efficienza di utilizzo della luce (LUE). Quindi, questa metodologia è stata applicata su un set di serie temporali spettrali e di flussi acquisite in campo su grano invernale, permettendo di migliorare significativamente la correlazione delle componenti veloci di F e PRI con LUE rispetto alle serie temporali originali. l'approccio basato su SSA è uno strumento promettente per disaccoppiare le informazioni fisiologiche a partire da misurazioni in continuo del segnale ottico.
Characterization of temporal variability of Plant Traits and Ecosystem Functional Properties using optical signals
BIRIUKOVA, KHELVI
2020
Abstract
Photosynthesis is the key process sustaining life on Earth. The efficiency of photosynthesis under varying environmental conditions is maintained by fast-responding regulatory mechanisms, namely chlorophyll a fluorescence emission (F) and non-photochemical quenching (NPQ). F is an electromagnetic radiation emitted at longer wavelengths (640-850 nm) than for excitation. NPQ is a complex process which includes the de-epoxidation of xanthophyll cycle pigments resulting in dissipation of excess energy as heat. This translates into changes in leaf absorbance at 531 nm, which can be detected with the photochemical reflectance index (PRI). Proximal sensing is a powerful tool for exploitation of subtle signals related to the downregulation of photosynthesis. However, F and PRI are also influenced by canopy structural and biochemical properties, illumination conditions and solar-view geometry, and its explicit interpretation is still challenging. The aim of my Ph.D. project was to exploit the methods of proximal sensing of vegetation to elucidate a link between continuous hyperspectral measurements of optical indicators related to plant physiology (F and PRI) and vegetation functioning. Multi-angular measurements were used to characterize the anisotropic response of far-red (F760), red (F687) fluorescence and PRI. I present an extensive experimental dataset of multi-angular measurements of F, PRI and R collected during a day over four different vegetation targets. Directional responses of F760, F687 and PRI in horizontally homogeneous canopies are characterized by increased values in the backward scattering direction with a maximum in the hotspot and decreased values in the forward scatter direction. The shape of F and PRI angular response is mostly controlled by leaf inclination distribution function (LIDF), while a magnitude and shape of the hotspot peak is sensitive to a combination of factors - leaf area index (LAI) and the ratio of leaf width to canopy height. Quantitative evaluation of the impact of anisotropy on fluorescence apparent yields (Fy*760 and Fy*687) showed, that, on average, off-nadir observations were overestimated by 20-67% in the backscatter direction and underestimated in other directions by 10-45%. The archived results reinforce the importance of maintaining nadir observation geometry for continuous measurements of F and PRI. Additional complexity for the interpretation of time series of optical signal lies in time-scale dependant vegetation dynamics. Here I test the applicability of highly-adaptive time series decomposition technique Singular Spectrum Analysis (SSA) for disentangling slow and fast varying components of vegetation dynamics in time series of F760, Fy*760 and PRI. First, the proof of concept was developed based on spectral and flux half-hourly time series realistically simulated with the SCOPE model. The simulations included two outputs - featuring and excluding the effect of the xanthophyll cycle de-epoxidation and fluorescence efficiency amplification factor on PRI and F760, respectively. The decomposed slow varying SSA-components of total PRI, F and Fy*760 showed a good correlation with reference constitutive variability simulated excluding the effect of physiological modulation (PRI0 , F0,760, Fy*0,760). The accuracy of disentangled fast varying SSA-components was validated with the reference physiologically induced variables (∆PRI, ∆F), and with non-photochemical quenching (NPQ) and light-use efficiency (LUE). Then, the application of this methodology on a field dataset of spectral and flux time series collected in winter wheat field allowed to significantly improve the correlation of fast components of F and PRI with the fast component of LUE in comparison with original time series. Therefore, SSA-based approach is a promising tool for decoupling physiological information from continuous measurements of optical signal which can foster the use of automated proximal sensing systems.File | Dimensione | Formato | |
---|---|---|---|
phd_unimib_824409.pdf
accesso aperto
Dimensione
7.56 MB
Formato
Adobe PDF
|
7.56 MB | Adobe PDF | Visualizza/Apri |
I documenti in UNITESI sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.
https://hdl.handle.net/20.500.14242/73765
URN:NBN:IT:UNIMIB-73765