The spatial heterogeneity of the land surface affects water, energy, and greenhouse gas exchanges with the atmosphere. Designing observation networks that capture land surface spatial heterogeneity is a critical scientific challenge. Here, we present a geospatial approach to capture the existing spatial heterogeneity of soil organic carbon (SOC) stocks across Alaska, USA. We used the standard deviation of 556 georeferenced SOC profiles previously compiled in Mishra and Riley (2015, Biogeosciences, 12:3993-4004) to calculate the number of observations that would be needed to reliably estimate Alaskan SOC stocks. This analysis indicated that 906 randomly distributed observation sites would be needed to quantify the mean value of SOC stocks across Alaska at a confidence interval of ± 5 kg m^-2. We then used soil-forming factors (climate, topography, land cover types, surficial geology) to identify the locations of appropriately distributed observation sites by using the conditioned Latin hypercube sampling approach. Spatial correlation and variogram analyses demonstrated that the spatial structures of soil-forming factors were adequately represented by these 906 sites. Using the spatial correlation length of existing SOC observations, we identified 484 new observation sites would be needed to provide the best estimate of the present status of SOC stocks in Alaska. We then used average decadal projections (2020–2099) of precipitation, temperature, and length of growing season for three representative concentration pathway (RCP 4.5, 6.0, and 8.5) scenarios of the Intergovernmental Panel on Climate Change to investigate whether the location of identified observation sites will shift/change under future climate. Our results showed 12-41 additional observation sites (depending on emission scenarios) will be required to capture the impact of projected climatic conditions by 2100 on the spatial heterogeneity of Alaskan SOC stocks. Our results represent an ideal distribution of observation sites across Alaska that captures the land surface spatial heterogeneity and can be used in efforts to quantify SOC stocks, monitor greenhouse gas emissions, and benchmark Earth System Model results.