Soil carbon flux is a vital part of the terrestrial ecosystem carbon cycle. Carbon dioxide released from soil is 10% to 20% of the total amount of that in atmosphere, equal to carbon assimilation by photosynthesis in plant canopies. Thus accurate measurement of regional soil carbon flux is significant for research on carbon cycling in terrestrial ecosystems. Remote data combined with models to analyze regional carbon balance is important but unreliable. The data from direct measurement of soil carbon flux turns to be more accurate. During the measuring process, random sampling results in inaccurate estimates of the regional soil carbon flux because of strong spatial heterogeneity. Another method, multi-point sampling in a wide area, requires much more human labor and equipment costs. However, a wireless sensor net (WSN) is a new method that has been applied to the measurement of soil carbon flux. Its advantages are low expense, wide coverage, multi-point synchronization, continuous monitoring, and be applicability to a wide range of areas. By using a self-regulating instrument based on WSN, Lr100GE-6400, to measure regional soil carbon flux, this paper presents a new layout strategy, Multiple Directional Interpolation (MDI). The strategy is as follows: (a) Set up the original well-distributed sampling sites and measure the carbon flux of these sites. (b) Calculate the intersections of different connection lines between sampling points by the radial point interpolation method and regard the points with maximum radial differences as new sampling points. Consequently, points are gradually increased based on this method to achieve accurate estimates of regional soil carbon flux. MDI layout strategy considers the spatial heterogeneity of soil carbon flux. It determines the sampling points based on dispersion, so that MDI can provide more accurate and comprehensive spatial information. Based on simulations in 20 fields of the 50×50 mesh region, layout is set up by MDI, random layout strategy, and uniform layout strategy, respectively. Keeping the same sample points, the three strategies were compared, and results were as follows: (1) MDI layout strategy can reflect the density of sampling points in terms of the variation gradient of soil carbon flux, decreasing use of sampling points and increasing the accuracy of measurement. (2) Error analysis of the experiments verified that the MDI layout strategy is more accurate than the average and random layout strategies. If there are fewer sampling points (n=10), errors in the MDI layout strategy are less than the random and more than the average strategies; if sampling points increase, the three types of layout strategies tend to be more accurate. If there are 40 sampling points, errors in MDI (0.028) are 12.5% less than that of the average (0.32),and 30.0% less than that of the random strategy. In conclusion, MDI layout strategy generally allocates the sampling points more reasonably based on the variation gradient of soil carbon flux. More sampling points are selected in the region with a large variation gradient in soil carbon flux, which matches the variation of soil carbon flux accurately, while fewer are located in the region with a small variation gradient in soil carbon flux but these do not affect the accuracy of soil carbon flux measurement. This method maximizes the effect of every sampling point and minimizes errors.