Date of Award




Document Type


Degree Name

Doctor of Philosophy (PhD)


Department of Atmospheric and Environmental Sciences

Content Description

1 online resource (iii, xix, 260 pages) : illustrations (some color)

Dissertation/Thesis Chair

David R Fitzjarrald

Committee Members

Vincent P Idone, Robert G Keesee, Christopher J Walcek, Geoffrey G Parker, Scott D Miller


dynamic light conditions, eddy-covariance flux, ensemble dynamic flux, NEE enhancement, whole-canopy dynamic modelling, whole-canopy NEE, Forest canopies, Photobiology, Plants, Carbon cycle (Biogeochemistry), Biogeochemical cycles, Carbon dioxide, Greenhouse effect, Atmospheric, Ecosystem health, Gas exchange in plants

Subject Categories

Atmospheric Sciences | Ecology and Evolutionary Biology | Environmental Sciences


An observed 20-30% increase in forest net ecosystem exchange (NEE) on partly cloudy days is often attributed to there being more uniform canopy illumination by diffuse radiation when clouds are present. However, the sky on such days is typically populated by fair-weather cumulus clouds, bringing dynamically changing shadow-to-light conditions on the order of minutes to the forest, with radiation alternating from 1000 W m^-2 in the clear sky to less than 400 W m^-2 in under-cloud shadows. These dynamically changing conditions cannot be investigated by the conventional time-averaged eddy-covariance flux method, which requires nearly steady-state turbulent conditions over much longer 20-30-min periods in order for the fluxes to converge to stationary values. We examine the “true” dynamics of the whole-canopy response to the light change by using a practical ensemble-flux method applied to eddy-covariance flux measurements from two distinct forest ecosystems: Harvard Forest (HF, 42.53°N, 72.17°W), temperate mid-latitude forest near Petersham, Massachusetts, USA, and Tapajós National Forest (LBA, 2.86°S, 54.96°W), an Amazonian evergreen tropical forest near Santarem, Pará, Brazil. Using the rapid change in radiative flux that occurs during the transition from cloud-induced shadow to light as a reference starting point, we combine sets of conditional illumination-change shadow-to-light and light-to-shadow transition events characteristic of cumulus-cloud conditions and parametrize distributions of light and shadow durations and rates of light change of the radiative-flux time series for different cloud conditions reported by standard weather stations. We investigate the sensitivity of the dynamics of forest response to the illumination transitions initiated by these conditional events, and identify an unexpected transient NEE maxima when NEE increases above the clear-sky steady-state equilibrium values (NEEeq) within the first 10 min of the light period after the shadow-to-light transition, that we hypothesize to be a physiological forest response to the abrupt light change due to presence of the intercellular CO2 pool in the leaf tissues. Overall NEE builds up during the sunlit periods, but in shadow heat and water stresses are reduced, thus increasing the water use efficiency (WUE). To conduct this analysis, we obtain similarity criteria for realizations defined by conditional events to combine them into the ensembles. With 300 similar realizations grab-sampled at 1-s intervals, we can reliably estimate (≤ 5% standard error) dynamic ensemble fluxes resolved on a 5-s time scale. By the successful application of the first-order system of the delay differential equations with the exponential approach-to-equilibrium solutions, we are able to justify the utility of the “Big-Leaf”-model approach to describe whole-canopy fluxes and provide the dynamic parametrizations of the “Big-Leaf” Active Thermal Layer as well as of the Transient Internal Layer above both forests when the light switches on after the cloud pass. By combining results of sensitivity analysis with modelled solutions applied to the real day-long fluctuating-light time series, we show that the variable light during fair-weather clouds (Shadow period duration < 100 s, Light period duration ≥ 300 s) is responsible for an increase in NEE above the NEEeq of 15-25% for HF and 10-15% for LBA. This indicates that there is a fluctuating-light NEE-enhancement mechanism that can be considered to be a viable alternative to the existing hypothesis of a diffuse-radiation NEE-enhancement mechanism on partly-cloudy days. We show that on such days Diffuse Fraction can be linearly-related to Cloudiness estimated using the shadow-to-light change in a conditional-event radiative-flux ratio, connecting these two NEE-enhancement mechanisms. Combination of increased NEE with the relatively high plateau in WUE allows forests to operate efficiently in partly-cloudy conditions with maxima located in the Cloudiness range [0.1-0.3] and Diffuse-Fraction range [0.35-0.6], suggesting forest adaptations to the preferred lighting conditions and fair-weather cloudiness.