Date of Award


Degree Type


Degree Name

Doctor of Philosophy (PhD)


Geography and Earth Sciences


Altaf Arain


Jing Chen



Committee Member

Mike Waddington


Landscape-level understanding of forest carbon (C) dynamics is required to quantify the net contribution of forest biomes to the global C cycle and to help forest managers to understand the impacts of forest management activities to the C sequestration in forests. Landscape-level estimation of C exchanges in various ecosystems is also crucial for the validation of the Moderate Resolution Imaging Spectroradiometer (MODIS) derived Gross Primary Productivity (GPP) which may help in improving MODIS GPP algorithm and the estimation of national-scale C budget to meet Canada’s international greenhouse gas inventory reporting obligations.

In this study, Carbon version of Canadian Land Surface Scheme (C-CLASS) was used to simulate historic C dynamics of a 2500 ha temperate Douglas fir forest landscape in Oyster River area of Vancouver Island in British Columbia from 1920 to 2005 and a 6275 ha boreal black spruce forest landscape at Chibougamau, Quebec from 1928 to 2005. The impacts of disturbance history and the climate variability on the landscape-level C stocks and fluxes were also investigated. The disturbance matrix of the Carbon Budget Model of the Canadian Forest Sector v3 (CBM-CFS3) was incorporated into C-CLASS to account for the removal of the C stocks by disturbance events. Study results indicate that GPP and autotrophic respiration (Ra) in the temperate Douglas fir forest landscape are sensitive to the air temperature variability. Stand replacing disturbance events can remove large amounts of C in the disturbed year, however, it takes a long period of time for the recovery of landscape-level total ecosystem carbon (TEC) to the initial state, which depends on forest age and the effects of historic climate variability. Our analysis further showed that in undisturbed forest landscape, simulated annual net ecosystem productivity (NEP) deviations were positively related to daily minimum and maximum temperatures in spring, while they were not sensitive to summer temperatures. Study results also showed that simulated landscape-level NEP is less sensitive to the changes in air temperature compared to other simulated C fluxes such as GPP, Ra and and heterotrophic respiration (Rh). Simulated landscape-level C stocks (aboveground biomass, belowground biomass, dead organic matter and soil organic matter) are sensitive to the changes in air temperature. This work suggests that the C-CLASS model can be used to investigate the impacts of climate variability and disturbance events on the historic C dynamics of forest landscapes. This study has also made it possible to analyze the importance of climate drivers and the development of methods for including climate sensitivity into inventory-based models.

In addition, C-CLASS simulated GPP overCanada’s landmass (at 1-km resolution) in 2003 and its comparison with the MODIS GPP product (MOD17) indicated overestimation of MODIS GPP compared to the C-CLASS upscaled GPP overCanada’s landmass. This overestimation was attributed to the limitations in the components of MODIS GPP algorithm. It further suggests that the parameterization of light use efficiency in MODIS GPP algorithm is amenable to improvement based on observations of light use efficiency at eddy covariance flux tower sites or the photochemical reflectance index derived from satellite remote sensing data.

This study would be helpful in calculating Canada’s national terrestrial ecosystem C budget which is important for making environmental policies and ecosystem management for enhancing the terrestrial C sink.

McMaster University Library

Files over 3MB may be slow to open. For best results, right-click and select "save as..."