Abstract:A hydrological experiment was conducted in a small headwater catchment in southeast Dartmoor, UK, to monitor the temporal and spatial variations in soil moisture content at the hillslope scale in order to determine how the spatial organization affected runoff generation. Two distinctly different types of rainfall response were found. During the dry state the soil moisture pattern was very patchy and the increase in stream discharge was relatively small for most rainstorms. The catchment response was limited to about 10% of the area, a figure that is similar in extent to the saturated area identified in the valley bottom. During the wet state, however, modest to large storms resulted in significantly higher discharge rates. The area generating the runoff increased up to 65% of the area. The division between the two 'preferred' states occurred at a catchment wetness of about 0Ð60 cm 3 cm 3 . This figure was based firstly on the exceptional increase in range, as determined by geostatistical analyses, for the soil moisture content measured associated with very high stream discharges. Secondly, it was consistent with a steep rise in gradient noted for the soil moisture characteristic curves at about 0Ð60 cm 3 cm 3 . The greater catchment responses were therefore dependent on the pore size distribution plus other soil characteristics and the connectivity between the wet areas.

The majority of geomorphological papers about Dartmoor have been essentially speculative, particularly when discussing weathering processes and the evolution of the Dartmoor landscape. In contrast, this article presents a synthesis of several experimental investigations aimed at studying the chemical weathering of Dartmoor granite through the systematic analysis of soil and water samples. This involved the computation of a geochemical budget to determine the amount of erosion in the catchment, as well as more detailed mineralogical investigations within a soil profile.The annual output of solutes due to weathering was 116 kg ha-' a-' of which the majority was silica (93 kg ha-' a-I). From an examination of the soil mineralogy, it was calculated that these solutes were derived from the dissolution of approximately 200 k g h a -l a -' plagioclase, 90 k g h a -l a -' biotite, and 40 k g h a -l a -' orthoclase. As well as the weathering of granite, there was also the production of kaolinite (150 k g h a -l a -' ) and gibbsite (002 kgha-la-'). Analysis of the soil water chemistry confirmed that kaolinite was the stable mineral phase in the regolith, although in areas where interflow was the dominant mode of water movement, the solute composition was in equilibrium with both kaolinite and gibbsite. Examination of the clay mineralogy confirmed these results.The microtexture of quartz grains was examined by the scanning electron microscope as another means of investigating the hydrochemical environment in the soil. Silica was found precipitated on all the grains examined but the maximum amount occurred in the Bs horizon. This evidence showed that, firstly, the dissolution of aluminosilicate minerals is greater than that calculated by thechemical budget and, secondly, that models of granite weathering must take localized weathering in the soil profile into account.The final part of the paper highlights the limitations of calculating denudation rates for an entire catchment and stresses the need to consider weathering as a highly localized phenomenon, particularly where there are high volumes of interflow at hill crest sites. Observations on granite decomposition in the future should be quantitative in approach and be related to the local site conditions.

No abstract