Research, Department of Recent studies about soil erosion show that erosion rates in winter can reach or exceedSoil Science, Brueckstr. 3a, erosion induced by summer events. A factor of particular importance is the incidence of 39114 Magdeburg, Germany. frozen soil, which modi.es the surface runoff generation and also the erodibility of the soil. E-mail: firstname.lastname@example.org In this paper, investigations are conducted to characterize the snowmelt erosion events in the 1·44 km2 ‘Schäfertal’ research catchment that is locatedc. 150 km southwest of Berlin. Two runoff events of different initial conditions are compared and analysed. Although the net erosion rate in the catchment is relatively low, the described event with soil frost is considerably different from the event without soil frost. For example, the net erosion, maximum and median suspended sediment concentrations are signi.cantly higher for the event with frozen soil. The results presented suggest that the sediment source areas differ for both situations. On one hand the channel or sediment .ushing is identi.ed as the source, while on the other hand hill slope processes and intra-storm variations are recognized for the soil frost situation. A model modi.cation is presented to improve the estimation of spatial differentiation of surface runoff within the continuous hydrological model WaSim, which is linked to the erosion model AGNPS. Based on measurements of topsoil temperature at locations with diverse exposition and land use, an algorithm is developed. The average agreement of air temperature to the calculated topsoil temperature is r2 0·75. The spatial information about soil frost is utilized to modify the in.ltration characteristics of the soil. Hence, the spatial aspects of runoff generation in winter conditions and the related transport of sediment is considered in a more realistic way. The spatial results of the modelling are plausible
Snow, as well as soil frost and thawing, is of importance in mid-latitudes and mountain catchments. Sharratt et al. (1997; after Pikul and Aase, 1998) estimated that approximately 50 per cent of the Earth’s surface is affected by frozen soil at least during part of the year. Spring snowmelt or frequent melting periods during winter are responsible for a large percentage of annual runoff and also matter .ux (Rekolainen, 1989). The temporal variability of the snow cover and spatial heterogeneity of soil freezing – sometimes together with rain-on-snow – causes a complex and dynamic runoff generation (Sui and Koehler, 2001).
Results of erosion studies in northern, central or eastern Europe and North America indicate that the erosion rate during snowmelt events can reach or even exceed the rainfall erosion rate (e.g. Demidov et al., 1995; Edwards et al., 1998; Lundekvam, 2001). Baade (1996) documents the importance of sediment transport during winter for a low mountain environment in SW Germany without further explanation. This study will present another example of the dominance of snowmelt erosion processes in a low mountain region of Germany.
G. Olleschet al.
However, most erosion models are developed for rainfall erosion. Snowmelt erosion and related processes of snow accumulation, snowmelt dynamic or soil frost are not described in an appropriate way. Renard et al. (1997) exemplify the weakness of the Revised Universal Soil Loss Equation and point out that the estimation of snowmelt erosivity and soil erodibility in cycles of freezing and thawing is problematic. Several deterministic models simulate rill processes but do not take into account the modi.cations caused by soil frost (Morgan et al., 1998; La.en et al., 1991). Moreover, in these models the concepts of runoff generation have limitations to describe winter conditions. Therefore, the underlying process to estimate runoff dynamics, runoff erosivity and transport capacity is inadequately modelled.
The main objective of this paper is to characterize the spatial dynamic of soil erosion processes and sediment source areas during snowmelt. The experimental catchment ‘Schäfertal’ (NE Germany) will be used as an example. A model approach to simulate the runoff generation in the catchment during soil frost conditions, which is passively coupled to an erosion model, is presented.
Material and Methods
The experimental catchment Schäfertal is located in the Harz Mountains, NE Germany, approximately 150 km southwest of Berlin. The outlet of the 1·44 km2 catchment is at an elevation of 392 m a.s.l. and the catchment ranges within 83 m. The Luvisols and Cambisols that have developed on the loess sediments on the slopes are intensively used for agriculture (Figure 1). Major crops are winter grain and rape in a rotation of various lengths. Triticale and peas have become more important in the last three years. The Eutric Gleysols and Dystric Gleysols at the valley bottom are utilized for pasture or meadow (see Figure 1). Average annual rainfall is approximately 640 mm; the annual average temperature is 6·8 °C, ranging from −1·8 °C in January to 15·5 °C in July.
The evident seasonality in discharge is mainly caused by variations of evapotranspiration. The discharge varies from less than 10 l s−1 in summer to above 200 l s−1 during the winter period. To a certain extent, a lowered groundwater table that is caused by mining activities in the region since 1970 disturbs the base .ow generation. As a consequence, inter.ow and tile drains dominate the runoff. During .ood events an important portion of runoff is generated by fast runoff components such as Horton type, saturated area runoff or preferential .ow. On average two or three major snowmelt runoff events occur between January and April.
Since the early 1960s measurements of meteorological and hydrological parameters have been conducted by the University of Applied Science, Magdeburg (Germany) including measurements of soil temperature and snow characteristics (depth, density and water equivalent). A monitoring programme for erosion as well as sediment and nutrient loads was set up in 1998. The regular biweekly sampling scheme for major nutrients and suspended sediments in runoff at the catchment outlet is supplemented by an automatic sampler (ISCO 6700 Series), which collects 24 samples at high water levels at equal time steps. The starting water level for the sampling and the time interval of sampling is seasonally differentiated (2 hours during winter). In addition to water temperature and electric conductivity, the relevant parameters are suspended sediment concentration, total phosphorus, dissolved phosphorus, reactive phosphorus, nitrate, ammonium and dissolved organic carbon. The treatment of the samples and laboratory analysis follow standard procedures. Special attention is paid to measurements of winter runoff generation and erosion since autumn 2000. Erosion is mapped after relevant events.
For the winter period 2001/2002 a measurement scheme was set up to characterize the temporal variability as well as the spatial heterogeneity of soil surface temperature. Results of a preliminary study of temperature in different soil depths show a low thermal conductivity of a typical soil type in the catchment. In consideration of the standard surface cover for meteorological stations (grass), the surface temperature is responding directly to changes in air temperature, which was measured 2 m above ground in time steps of 5 min (Figure 2). Soil surface temperature exceeds the extreme low and high values in the period presented. In contrast, the temperature regime in a depth of 0·1 m is even-tempered and is not following any daily rhythm or other short-term variations. Consequently, .fteen automatic temperature sensors (Tinytalk, Geminidataloggers) were installed in a cross-section through the valley at a depth of 0·05 m for the period from 23 November 2001 until 17 April 2002 with two hours registration time. Different soil types, land use and expositions are represented.
Soil moisture measurements in a depth of c. 0·2 m with TDR probes were conducted at the positions of soil temperature measurement (EASY-test FP/mts). Daily readings of the period from 24 November 2001 until 30 April 2003 were used for this study. A longer duration of observation was chosen compared to the temperature values because of probe failures or tillage measures which interrupted the measurement period. The integrated temperature sensors did not indicate situations of soil frost in the relevant depth which might in.uence the measurement principles.
Figure 1. Land use and soil type (FAO classi.cation) of the Schäfertal catchment; inset map shows location of the catchment within Germany.
During the winter periods 2000/2001 and 2001/2002 .ve periods of snow accumulation and melting occurred; one summer storm event in May 2002 was monitored. Compared to one extensive summer event in June 1999 and data from literature (i.e. Steegen et al., 2000), the suspended sediment concentration (SSC) during the .ve storms is low and does not exceed 1500 mg l−1. These values are below those which were measured during winter events in a 62 ha catchment in the Kraichgau region (SW Germany) but in the range of concentrations from a 702 ha catchment (Baade, 1996). Dif.culties of comparison exist because of differences in slope and portion of arable land. The results from the Schäfertal correspond to measuring data in similar catchments with loess cover (i.e. van Dijk and Kwaad, 1996) or in those with riparian vegetation zones (i.e. McKergow et al., 2003). However, due to runoff volume, the major portion
G. Ollesch et al.
Figure 2. Characterization of air temperature, soil surface temperature and soil temperature at a depth of 0·1 m below the surface in the Schäfertal catchment for a period in February and March 2001.
Table I. General characteristics of two snowmelt–erosion events in the Schäfertal catchment