Ecosystem productivity indicates the net carbon captured (or released) by primary producers and consumers in an ecosystem. It is an important biophysical parameter not just because it defines the habitability of the earth, but the role of forest ecosystem as a primary sink (or source) for the atmospheric carbon dioxide. With the uncertainty of global climate change and complex ecosystem functions, it is necessary to delineate its spatial and temporal behavior in terms of multi-scaled perspectives. In this study, Moderate Resolution Imaging Spectroradiometer (MODIS) primary productivity product (MOD17), which includes gross primary productivity (GPP) and net photosynthesis (PSNnet), have been analyzed and compared with the net ecosystem exchange (NEE) derived from eddy-covariance flux measurements. With the assistance from Taiwan Typhoon and Flood Research Institute (TTFRI), CO2 flux from four experiment campaigns have been used in this study, which covers cropland (SE1), urban and build-up (SE2), mixed forest (SE3), and evergreen broadleaf forest (SE4), with data periods range from 6 months to 1 year (2009/5-2010/5). Based on the synergic analysis of data from complementary platforms, both the micro- and macroscopic behavior, and the discrimination of primary productivity from its respiration cost could be described. With time-series analysis of MODIS data (2001-2010), the study indicated that GPP (overall forest ecosystems in Taiwan) present the highest value in summer (1.62-1.76 tC/ha/m), despite of a 3-month lag following the highest PSNnet in the early growing season of April (1.18-1.25 tC/ha/m). Such temporal difference could be explained by the seasonal dependence of available incident photosynthetic active radiation (IPAR) and autotrophic respiration (Ra )(or in terms of leaf area index). Among the forest ecosystems, evergreen needleleaf forest presents a comparably more constant PSNnet/GPP ratio than others. In all 4 tower sites, IPAR seems to play a major role for the determination of diurnal (onset and offset) and daily variation of NEE, despite of the hourly lag depending on the surrounding surface pattern. Contrary to Ra , heterotrophic respiration (Rh ) shows no clear seasonal dependence, but it tends to decrease during the rainy days, and be followed by a sharp increase once the rainy days come to an end. For the spatial context, the variation of GPP is closely related to available IPAR and ecosystem composition, which describes the latitudinal and altitudinal dependence of primary productivity. In the 4 tower sites, SE2 indicate net carbon sources, while SE1, SE3, and SE4 net carbon sinks. To validate MOD17 product, GPP were further compared with flux-derived GPP based on identical spatial and temporal scale. The comparison indicated that MODIS GPP was closely related in the temporal trend to the tower-based GPP, with correlation coefficient of 0.79 for SE4.