How terrestrial ecosystems respond to environmental changes affects the well-being of human society. Thus, extreme climate events, increasing the atmospheric concentration of CO2, and drastic changes in temperature are sources of major concern. However, our current capacity to understand and predict these responses is still limited because a myriad of physical, chemical, and biological processes are involved. While many mechanistic-based land surface models have been developed, their performances remain relatively poor and require continuous improvement. While ground-based and space-based observational datasets of the surface of the Earth have been available for a long time, their linkages to the functional aspects of the processes in terrestrial ecosystems often are weak. In this study, I used the approach of integrating in-situ measurements, land surface models, and remote sensing by satellites. I hypothesized that, through such integration, the impacts of environmental changes on terrestrial processes at multiple scales could be better understood and even predicted, especially the impacts related to the functions of important ecosystems. I tested this hypothesis at the local, regional, and global scales. At the local scale, i.e., at a Midwest forest site known as the isoprene volcano of the world, I examined the effects of droughts on the emissions of isoprene, which is the most abundant, non-methane, biogenic volatile organic compound. I compared flux tower observations with simulations performed by a state-of-the-art land model (CLM) coupled with the model of emissions of gases and aerosols from Nature version 2.1 (MEGAN2.1), and I used these observations to develop an understanding of how the amount of moisture in the soil and the ambient temperature affect the prediction of isoprene emissions during droughts. I found that temperature had a delaying effect on isoprene emissions, which are sensitive to variations in the moisture content of the soil. Thus, during drought conditions, both the delaying effect and the sensitivity to moisture are overlooked by the model. A better model that does not have these two shortcomings is required for realistic predictions of isoprene emissions. At the regional scale, I investigated the potential of using sun-induced chlorophyll fluorescence (SIF) retrieved from a satellite to monitor vegetation activities in an arid region and a semi-arid region in Australia. I chose these two types of regions for this investigation because the ecosystems in such regions have important effects on the global carbon cycle, while their contributions are poorly constrained in global carbon budgets. I found that SIF was synchronized better with the activity of vegetation than other indices that are commonly used for this purpose. I quantified the relationships between the various activities of plants and the amount and frequency of precipitation, and I was able to demonstrate that, over the region being studied, SIF represented the activity of vegetation in response to the availability of water better than other, remotely-sensed variables. At the global scale, I used multiple model ensembles to determine the climatic and anthropogenic controls on the terrestrial evapotranspiration trends from 1982 to 2010. After climatic influences, increases in CO2 were found to be the second-most dominant factor that affected the trend of ET. CO2 causes a decreasing trend in the canopy’s transpiration and ET, and this is especially of concern for tropical forests and high-latitude shrub lands. The increased deposition of nitrogen amplifies the global ET slightly due to enhanced growth of plants. On a global scale, land-use-induced ET responses are minor, but they can be significant locally, particularly over regions with intensive changes in the land-cover. The results of my studies demonstrated that integrating in-situ measurements, models of the surface on the land, and remote sensing using satellites can provide insights regarding the impacts of environmental changes on terrestrial processes at multiple scales. This approach is particularly important when models are imperfect and observations are lacking. My findings indicated ways that future models can be improved and identified key observational needs for the functions of terrestrial ecosystems.