This study was undertaken to test the utility of a geographical information systems (GIS) approach to problems of watershed mass balance. This approach proved most useful in exploring the eects that watershed scale, lithology and land use have on chemical weathering rates, and in assessing whether mass balance calculations could be applied to large multilithological watersheds. Water quality data from 52 stations were retrieved from STORET and a complete GIS database consisting of the watershed divide, lithology and land use was compiled for each station. Water quality data were also obtained from 7 experimental watersheds to develop a methodology to estimate annual ¯uxes from incomplete data sets. The methodology consists of preparing a composite of daily ¯ux data, calculating a best ®t sinusoid and integrating the equation to obtain an annual ¯ux. Comparison with annual ¯uxes calculated from high resolution data sets suggests that this method predicts ¯uxes within about 10% of the true annual ¯ux.
Annual magnesium ¯uxes moles km À 2 yr À1 were calculated for all stations and adjusted for ¯uxes from atmospheric deposition. Magnesium ¯ux was found to be a strong function of the amount of carbonate in the watershed, and silica ¯uxes were found to increase with the fraction of sandstone present in the watershed. All ¯uxes were strongly in¯uenced by mining practices, with magnesium ¯uxes from aected watersheds being 6±10 times higher than ¯uxes from comparable pristine watersheds. Mining practices enhance chemical weathering by increasing the surface area of unweathered rock to which water has access and by increasing acidity and rate of mineral weathering. Fluxes were also found to increase with watershed size. This scale dependence is most likely caused by the sensitivity of weathering ¯uxes to even minor quantities of carbonates, which are likely to be found in all lithologies at larger scales.
Mass balances were carried out in watersheds where gauged sub-watersheds made up more than 95% of the area. The calculations show large magnesium ¯ux and water balance discrepancies. These errors may be a result of signi®cant groundwater inputs to streams between gauges. The results suggest that improvements in how we measure discharge and estimate ¯uxes may be required before we can apply mass balance techniques to larger scales.