Extreme events controlling erosion and sediment transport in a semi-arid sub-andean valley

The importance of extreme events in controlling erosion and sediment transport in semi-arid areas has long been appreciated but in practice being by definition rare and episodic they are difficult to study. When they are observed this is frequently in catchments for which little data are available. Another difficulty is that even when catchments are being monitored the instruments that record discharge, sediment load and hillslope sediment fluxes perform inaccurately or unpredictably during extreme weather conditions. This paper describes slope and channel processes that were actually observed by the authors during a (at least) I : 30 year 30 minute event with a rainfall intensity of 240 mm h(-1) in a second-order tributary of the Rio Camacho near Tarija in southern Bolivia. During the event, it could be observed how different tributary streams and slope sections contributed sediments and flow to the main channel. Evidence for these contributions did not survive the event, which has implications for both modelling and monitoring. Before the onset of the event open erosion plots were functioning on the slopes where rainfall and runoff were being measured. Rainfall experiments were used to obtain infiltration rates. The storm began with a moderate intensity of about 5 cm per hour but increased after 5 minutes to 30 cm per hour and continued for 30 minutes. At this time, the rainfall intensity greatly exceeded the infiltration capacity and water started draining the steep slopes. The ephemeral channel rapidly filled up with runoff. Erosion by hailstones was considerable. Provisionally, the discharge during peak runoff was estimated at 43.7 m 3 s(-1) (Manning equation). On the basis of sediment loads carried by previous storms, (average concentration of 21 g l(-1)) the total suspended load discharge during the storm would have been 15 ton ha(-1). Within the ephemeral channel, 10 to 50 cm thick layers of coarse sediments were deposited. The collectors of the open erosion plots could not handle the large amounts of runoff and sediment and were completely filled to overflowing. Comparing these data with soil losses during less intense storms it can be concluded that extreme events largely contribute to erosion and sediment transport and that the majority of the rainstorms play only a minor role. The results also show how limited the values of rainfall experiments are in understanding geomorphic events. This makes modelling of erosion and soil losses a difficult and hazardous task. Copyright (C) 2002 John Wiley Sons, Ltd.