『Abstract
　Prediction of CO2-consumption by chemical weathering is important to understand the global carbon cycle, and island arcs are assumed to contribute significantly to the transfer of atmospheric/soil CO2 to the oceans. Previous work established empirical functions for bicarbonate-flux and CO2-consumption in dependence of runoff for five to six lithological classes. These functions were applied in global studies. However, it has remained uncertain, if improvements can be achieved by considering further factors or by an enhanced lithological classification scheme. This study applies a new lithological map of Japan with an enhanced lithological classification, a hydrochemical data set representing 382 catchments (covering 〜44.4％ of the Japanese Archipelago) and a multi-lithological model approach to predict bicarbonate-fluxes and CO2-consumption by chemical weathering. Because of significant carbonate contents in sediments (different from carbonate sedimentary rocks), acid plutonics and metamorphics, firstly a bicarbonate-flux model approach is established and, secondly, silicate and carbonate CO2-consumption is estimated, based on the geochemical composition of rocks.
　In accordance with previous studies on the catchment scale, the most important factors controlling bicarbonate-fluxes are lithology and runoff. The anion ratio HCO3^-/(SO4^2- + Cl^-) is identified as the third most important predictor, because anion sources and anion composition of river water have a great effect on bicarbonate concentrations. And last but not least, potential predictors such as gradient of slope, temperature and physical erosion explain some part of the observed bicarbonate-flux variation. These potential predictors have been discussed in literature, but quantification of the effects of these factors remains difficult due to their correlation with runoff and lithology. However, all tested models reproduce a total of observed bicarbonate-fluxes within 10％ on the regional scale including the simplest model which recognizes only lithology and runoff as predictors.
　Model results suggest that bicarbonate-flux from the Japanese Archipelago is about 6.61 t C km^-2 a^-1, and CO2-consumption by chemical weathering is about 6.05 t C km^-2 a^-1 (91.6％ of bicarbonate-flux). This CO2-consumption rate is 3.2 times above the global average rate. The silicate to carbonate CO2-consumption ratio is comparatively high. It amounts to 9.9 which is above the global value. The latter one being 1 to 1.5. Carbonate sedimentary rocks contribute only 1.3％ to bicarbonate-fluxes due to their low areal proportion (0.2％). Acid volcanic rocks (VA) show the lowest bicarbonate-fluxes, on average, while basic and intermediate volcanics (VB) as well as pyroclastic (PY) are in the range of the Japanese average value. The carbonate weathering contribution to bicarbonate-fluxes from mixed sediments, siliciclastic sediments, metamorphics and acid plutonics is estimated to be 69.2％, 23.3％, 63.4％ and 24.8％, respectively. Only an average carbonate CO2-consumption per lithological class is calculated because no function based on typical predictors could be established to calculate carbonate weathering contribution to silicate-dominated lithological classes. This can be attributed to large spatial differences in carbonate content in rocks and to corresponding contributions to bicarbonate-fluxes. In the following, results are compared to bicarbonate-flux models of previous studies and identified differences are discussed. In conclusion, recognition of carbonate abundance in silicate-dominated lithological classes (including trace calcite) is important to calculate silicate/carbonate CO2-consumption ratios. Presented results are relevant for studies modelling CO2-consumption by chemical weathering in context of the global C-cycle.

Keywords: Chemical weathering; CO2; Bicarbonate; Lithology; Silicate; Carbonate』

1. Introduction
　1.1. Weathering and the global carbon cycle
　1.2. Controls on chemical weathering
　1.3. Prediction of CO2-consumption by chemical weathering
　1.4. Objectives
2. Data and methodology
　2.1. Overview
　2.2. Hydrochemical data
　2.3. Handling of geodata
　2.4. Mass balance
　2.5. Preanalysis
　　2.5.1. Selection of parameters for CO2-consumption modelling
　 2.5.2. Selection of predictors for bicarbonate-flux models
　2.6. Modelling technique
　2.7. Calculation of the contribution of carbonates to bicarbonate-fluxes
　　2.7.1. Ca-proportion counterbalancing bicarbonate
　　2.7.2. The hypothetical maximal carbonate contribution to bicarbonate-fluxes
　　2.7.3. Calculation of silicate and carbonate CO2-consumption
　2.8. Application of the bicarbonate runoff model on the Japanese Archipelago
3. Results
　3.1. Overview
　3.2. Observed fluxes
　3.3. Relationships between proposed predictors and bicarbonate-fluxes/concentrations
　3.4. Predicted bicarbonate-fluxes
　　3.4.1. Monitored catchments
　　　3.4.1.1. Average bicarbonate-flux predictions per lithological class
　　　3.4.1.2. Interpretation of parameter estimates
　　3.4.2. Application to the Japanese Archipelago
　3.5. The influence of silicate and carbonate proportions on CO2-consumption
　3.6. Evaluation: analysis of model residuals
　3.7. Evaluation: comparison with previous bicarbonate-flux models
4. Discussion
　4.1. General discussion
　4.2. Observed bicarbonate-fluxes: local findings
　4.3. Discussion of model results
　　4.3.1. Discussion of predictors and potential errors
　　　4.3.1.1. Runoff
　　　4.3.1.2. Lithology
　　　4.3.1.3. Anion ratio AER
　　　4.3.1.4. Temperature
　　　4.3.1.5. Slope
　　　4.3.1.6. Physical erosion
　　4.3.2. CO2-consumption and contribution of carbonates
　4.4. Comparison with previous models
　　4.4.1. Bicarbonate-flux models
　　4.4.2. Global CO2-consumption models
　4.5. Error discussion considering factors not included in the model discussion
　　4.5.1. Metamorphic CO2
　　4.5.2. Soils
　　4.5.3. Ecosystems and land cover
　　4.5.4. Mass balance
5. Conclusion
Acknowledgments
Appendix A. Distribution of precipitation and atmospheric hydrochemical deposition patterns were calculated from the following references. Atmospheric deposition was extrapolated using the product of concentration in atmospheric deposition and annually averaged precipitation
Appendix B. Calculated Ca-excess for lithological classes SM, SS, MT and PA. Ca-excess is assumed to be equivalent to bicarbonate-fluxes from carbonates
Appendix C. Application of the Japanese CO2-consumption model on the global scale
Appendix D. Supplementary data
References