Quantifying carbon dioxide exchange and understanding the response of key environmental factors in various ecosystems are critical to understanding regional carbon budgets and ecosystem behaviors. For this study, CO2 fluxes were measured in a variety of ecosystems with an eddy covariance observation matrix between June 2012 and September 2012 in the Zhangye oasis area of Northwest China. The results show distinct diurnal variations in the CO2 fluxes in vegetable field, orchard, wetland, and maize cropland. Diurnal variations of CO2 fluxes were not obvious, and their values approached zero in the sandy desert, desert steppe, and Gobi ecosystems. Additionally, daily variations in the Gross Primary Production (GPP), Ecosystem Respiration (Reco) and Net Ecosystem Exchange (NEE) were not obvious in the sandy desert, desert steppe, and Gobi ecosystems. In contrast, the distributions of the GPP, Reco, and NEE show significant daily variations, that are closely related to the development of vegetation in the maize, wetland, orchard, and vegetable field ecosystems. All of the ecosystems are characterized by their carbon absorption during the observation period. The ability to absorb CO2 differed significantly among the tested ecosystems. We also used the Michaelis-Menten equation and exponential curve fitting methods to analyze the impact of Photosynthetically Active Radiation (PAR) on the daytime CO2 flux and impact of air temperature on Reco at night. The results show that PAR is the dominant factor in controlling photosynthesis with limited solar radiation, and daytime CO2 assimilation increases rapidly with PAR. Additionally, the carbon assimilation rate was found to increase slowly with high solar radiation. The light response parameters changed with each growth stage for all of the vegetation types, and higher light response values were observed during months or stages when the plants grew quickly. Light saturation points are different for different species. Nighttime Reco increases exponentially with air temperature. High Q10 values were observed when the vegetation coverage was relatively low, and low Q10 values occurred when the vegetables grew vigorously.