Abstract
Water supply to crops is increasingly crucial in the face of climate change. This thesis aims to enhance our understanding of water dynamics in the fields of agriculture, hydrology, and ecosystem ecology. The initial studies were composed of three sub-chapters, collected under the title "Developing proper methods for ET studies" (Chapter 3), addressing important questions about the methodology used in evapotranspiration (ET) studies. Sub-chapter 3.1, "Assessing the correlation between gross primary production (GPP) and ET at ecosystem-scale using the Eddy Covariance method," explored the concept of water use efficiency at the ecosystem level to evaluate the coupling of ET and GPP, considering methodological decisions related to energy balance closure and algorithms for partitioning the net ecosystem exchange (NEE) to obtain GPP. Results showed that even though the energy imbalance is a site-specific issue, forcing the energy balance closure was useful to improve the GPP – ET coupling. Furthermore, we confirmed that both the day-time and night-time NEE partitioning methods can be used in studies related to GPP – ET coupling. Sub-chapter 3.2, "Dynamics of vine transpiration and vineyard evapotranspiration and their response to meteorological driving variables along dry days," focused on understanding the effect of meteorological variables on grapevine transpiration flux (Tv) and the diurnal dynamics of vine contribution to bulk ecosystem water fluxes (Tv/ET). By contrasting Tv estimates originated by 7 heat‐balance sap flow gauges to eddy covariance bulk evapotranspiration (ET) in a grassed vineyard, we found that Tv/ET varied throughout the day, responding rapidly to environmental stimuli. It increased with irrigation and decreased after midday, suggesting stomata closure. These results emphasized the importance of irrigation practices in viticulture, especially in the context of climate change. Sub-chapter 3.3, "Defining a protocol for measuring evapotranspiration with ground flux chambers," addressed methodological issues related to using chambers for obtaining evapotranspiration measurements. The first issue focused on the reliance of understorey evapotranspiration (ETu) on the regression approach and interval used on the gas concentration time series to obtain the flux. Transparent ground flux chambers were deployed in a vineyard to compute ETu fluxes using different models (linear and exponential) and fitting intervals (05-30 s, 15-30 s, 20-40 s and 20-80 s within the 90 s observation length). The analysis suggested that a linear regression of water vapor concentration time series using the 20-40 s interval provides more stable ETu flux patterns. The second issue is concerned about sorption/adsorption of water vapor on equipment internal surfaces, which affects the reliability of ETu values. A comparison of chamber-estimated ETu with the total exchanged amount of water obtained from a gravimetric method showed that chamber-obtained ETu were underestimated. We suggest that ETu can be corrected by an averaged scalar factor of 1.76, that accounts for the attenuation effect due to the sorption and desorption issue in the tubbing. The third topic addressed the challenges of deploying chamber systems in areas with tall herbaceous vegetation. An approach based on the green leaf area (GLA) is proposed, where the initially computed flux (ETu*) is multiplied by a collar-specific factor obtained from the quotient of original total GLA and the GLA fit within the dome after cutting the vegetation at 10 cm height (GLAtotal/GLA<10cm).
In a second part of this work, the comprehensive study “Evapotranspiration dynamics and partitioning in a grassed vineyard: ecophysiological and computational modelling” (chapter 4) took place in Caldaro (South Tyrol, Italy), from May 2021 to end August 2022. Eddy covariance, sap flow gauges, ground flux chambers and two models (the Transpiration Estimation Algorithm - TEA and a site-trained random forest model for Tv) were deployed aiming to assess i) which process, evaporation (E) or transpiration (T), had greater influence on ET dynamics; ii) which component among grapevines and understorey portion dominated the ET; and iii) how rainfall influences ET components. Two approaches were performed: a top-down approach that combined the eddy covariance method to estimate ET, and the TEA method to partition it, and a bottom-up approach integrated the understorey evapotranspiration (ETu) with modelled vines transpiration (Tv(mod)). The mean ETEC was 2.54 ± 0.89 and 3.34 1.10 mm d-1 (2021 and 2022), being TEC the higher contributor (TEC/ETEC of 0.78 ± 0.13 and 0.89 ± 0.33, same years). From the bottom-up approach, ETu during campaign days (0.62 to 1.52 mm d-1) was lower than Tv. A high agreement (R2= 0.85) was found between the eddy covariance ET hourly values and ET by summing Tv(mod) and ETu. We concluded that the T process represented major fluxes in the agroecosystem during summer, furthermore, the bottom-up approach indicated the vines as primary contributors to total T, especially after rainfall, as the system became drier and understorey transpiration contribution increased.