An accurate estimate of total forest carbon (C) stock and C uptake is crucial for predicting global warming scenarios and planning CO₂emission reductions. Forest inventory, based on field measurements of individual tree sizes, is considered the most accurate estimation method for forest C stock. Japan’s national forest inventory (NFI) provides stand-scale stem volume for the entire forested area based on (1) direct field measurements (m-NFI) and (2) prediction using yield tables (p-NFI). Here, we show that Japanese national and local forestry agencies and some research studies have used p-NFI and greatly underestimated the Japanese forest C stock (58–64%) and net annual C uptake (41–48%). This was because approximately 10% of the forest area was not counted in p-NFI and because the yield tables in p-NFI, which were constructed around 1970, were outdated. For accurate estimation of the forest C stock, yield tables used in p-NFI should be reconstructed or ideally field measurement campaigns for m-NFI should be continued. In the future, appropriate forest management plans are necessary to effectively use the high CO₂absorption capacity of Japanese forests and these should be compared with other industries’ CO₂reduction plans from a cost-benefit perspective.

Understanding how spatial variability in stream discharge and water chemistry decrease with increasing catchment area is required to improve our ability to predict hydrological and biogeochemical processes in ungauged basins. We investigated differences in this decrease of variability with increasing catchment area among catchments and among specific discharge (Q_s) and water chemistry parameters. We defined the slope of the decrease in the variability with increasing catchment area as the rate of decrease in the standard deviation and coefficient of variation (δ_SD and δ_CV, respectively), both of which are −0.5 for the simple mixing of random variables (random mixing). All δ_SD and δ_CV values of Q_s were less than −0.5, while those of most water chemistry values were greater than −0.5, indicating that with increased catchment area the spatial variability of Q_s decreased more steeply than for random mixing, while for water chemistry they decreased less steeply. δ_SD and δ_CV had linear relationships with both the spatial dissimilarity index and relative changes in parameters' mean values with increasing catchment area. It suggested that differences in δ_SD or δ_CV for Q_s and water chemistry can be explained by the different spatial structures, where dissimilar values of Q_s and similar values of water chemistry, respectively, are located close together in space. Differences in δ_SD and δ_CV according to Q_s and water chemistry should significantly affect the determination of representative elementary area and therefore need to be considered when predicting representative elementary area from spatial variability of low‐order streams.