The availability of spatial data sourced from either field-derived or satellite-based systems has created new opportunities to estimate and/or monitor changes in carbon sequestration rates, climate change impacts or the potential habitat alterations occurring across large landscapes. However, an effort to create models is not standardized, in part, due to different needs and data sources available for the models. For example, data may have different spatial resolutions with varying degrees of complexity in regards to inputs and statistical methods. This study determines effects of 20, 15, 10, five and one km sampling resolutions on detection of changes in net primary productivity (NPP), occupancy selection criteria for areas to be included in the sample and identification of significant variables impacting NPP in Indonesia forests. Production forest designated for selective harvest was used to define the sampling areas. Variances explained by predictive models were similar across cell sizes although relative importance of variables was different. Partial dependence plots were used to search for potential thresholds or tipping points of NPP change as affected by an independent variable such as minimum daytime temperature. Applying different cell occupancy selection rules significantly changed the overall distribution of NPP values. The magnitude of those changes within a cell size varied with changes in cell size. The mean estimated NPP for production forests across Indonesia differed significantly at every sampling resolution and occupancy selection criteria. Lows ranged from 1.107 to 1.121 kg C m−2 yr−1 for the 1-km cell size for the three occupancy selection criteria with highs ranging from 1.245 to 1.189 kg C m−2 yr−1 for the 20-km cell size. The difference in NPP values between these two cell sizes for the three occupancy selection criteria extrapolates to a range in annual biomass of 132 × 106 to 66 × 106 t for the total area of production forests in Indonesia.