Published on January 17, 2008
Slide1: Acknowledgements Many thanks to Peter Clark, Brian Irvine, Mark Smith and Andy Turner for their help during fieldwork, data analysis and for research inputs. Runoff Generation SE Spain E. N. Dalen¹, M.J. Kirkby¹, P.J. Chapman¹ and L.J. Bracken² ¹ School of Geography, University of Leeds, UK ² Department of Geography, Durham University, UK Approach Study area Project & Objectives We are working to improve a hydrological model for prediction of runoff in medium-scale semi-arid catchments in SE Spain, by investigating the impact of different landscape elements on runoff, to estimate runoff within two 150 km2 catchments (Figure 3), Rambla Nogalte and Rambla de Torrealvilla. The study areas have been studied since early 1990’s with rainfall (at 7-9 sites) and some runoff data regularly collected over this period. The approach is to use the concept of Hydrologically Similar Surfaces (HYSS), which are defined as areas with similar1-D (vertical) partitioning of net rainfall between infiltration and overland flow. HYSS are identified from field measurements of soils, micro and macro-topography and infiltration rates; combined with analysis of existing multi-spectral airborne Remote Sensed (RS) images. HYSS are selected to minimise internal variability in the relationship between rainfall and local runoff generation. HYSS characteristics are based on measurements of plot characteristics, and have been scaled up to cover larger areas. Most storms within these catchments are typically with intense bursts of 30 - 60 minutes. The aim is to develop and improve the understanding of runoff generation in semi-arid areas and hopefully improve modelling of runoff in semi-arid areas (Figure 1 and 2). Objectives are to investigate the influence of geology, landuse and seasonality on infiltration rates and use RS and GIS to classify an area into HYSS categories. The overall sampling strategy for measurements was to undertake constant intensity rainfall simulator measurements within provisional HYSS categories, and to augment this with a large number of minidisk infiltrometer measurements. This strategy was to capture as much of the variability in the landscape as possible (Figure 4), and provide data on both local and coarser scales. The wide variability within even small areas led to the final adoption of only a few large classes that could be effectively distinguished. The Rambla Nogalte catchment is located in SE Spain (37˚ 35’ N, 01˚ 56’ W) (Figure1). The area is 171 km2 (Figure 3). The geology is dominated by red mica schist and local outcrops of blue mica schist (Figure 1). The lithology is dominated by metamorphic rocks and conglomerates. The main land covers are almond and olive cropping and scrub. Rambla de Torrealvilla is 200 km2 (Figure 3) and the geology is dominated by marls (Figure 2) and other sedimentary rocks and is characterized by steep sized channels cut into a flat piedmont surface. The main land uses are wheat, watermelons and lettuce. The rest of the catchment is scrubland. Sprinkler 4 field campaigns were used to cover all important surfaces with infiltration measurements. Surface properties were measured: porosity, bulk density, surface strength and soil moisture, in addition to soil sample analysis. Fieldwork Philip’s: Green-Ampt: B* = sorptivity GA (B) = constant A = steady long term infiltration rate Philip’s (A*) = Green-Ampt (A) The initial hypothesis developed prior to fieldwork was that soil type: blue schist, red schist, intermediate and marl, should produce significantly different responses to rainfall. The hypothesis that cultivation was important was also investigated; infiltration was expected to differ between cultivation levels. The strategy of the fieldwork in April/May 2005 and November/December 2005 was to measure a wide range of surfaces of different characteristics in the Rambla Nogalte and the Rambla de Torrealvilla with the rainfall simulator and the minidisk. The idea was to capture as much of the variability in the landscape as possible. The sampling was designed to provide both data on infiltration locally and on a larger scale and capture variability in infiltration within and between sites. The challenge was to translate the dataset into HYSS classes, which have fairly uniform response to precipitation. The intention was to differentiate infiltration levels between sites and use remote sensed data to map areas with different runoff thresholds. The rainfall simulator measured infiltration and runoff. The choice was a spray type simulator built according to design by Calvo (Cerda et al 1997)(Figure 5). The simulator is designed for rugged terrain and has been used in semi-arid areas for over a decade with good results. Minidisk The Decagon Mini Disk Infiltrometer was used; it has a radius of 1.59 cm, and a suction rate of two cm (Figure 6). The method used was that of Zhang (1997). This method of measuring hydraulic conductivity has proved successful and is an effective method to measure infiltration over large areas. The method measures cumulative infiltration (I) vs. time (t) and fitting the results with the Philips equation. School of Geography Faculty of Environment Figure 4: Rambla Nogalte almond crop The final part of the research was to link the spatial partition of the two catchments into HYSS with the detailed rainfall records for the areas, and combine these two sets of data into a grid-based model for runoff generation across the area The sprinkler was run twice; the double run, one dry and one wet. After the dry run was storage (S) estimated for the second run. f = A + B/S was used to get the infiltration rate (f). A and B could be changed to let the model fit the raw data, using cumulative infiltration = rainfall – runoff, and model of cumulative infiltration F = f*time can be seen in figure 11. Sprinkler: method and model Figure 12: Relationship sprinkler and minidisk The relationship between sprinkler and minidisk values showed high variation at each site and the correction values sometimes gave extreme outliers at both ends of the scale. Each location had a high variation of minidisk measurements (Figure 12). This local variability was seen at all sites measured, non-uniform soils was the general trend in both catchments. Figure 13 shows values from blue schist uncultivated and the variability on presumed similar surfaces. Each location has variability as seen at measurement Los Perez, which had two outliers GA B-value close to 300, however there is a central tendency represented by the trend line. The investigation of lithology and its influence on infiltration rates was crucial to determine the best possible estimates of infiltration rates in the study catchments. The blue schist in the Rambla Nogalte was used as an example, since these surfaces were more prone to crusting than other surfaces in this catchment (Figure 7). Results show that the cultivated surface with well developed crust did have significantly lower infiltration than both uncultivated surfaces and cultivated surfaces without significant crusting. The difference between cultivated surfaces crusted are very difficult to establish, since this is an issue of lithology, rainfall and agricultural practise. Infiltration data and HYSS selection There was no evidence to support the hypothesis that HYSS categories could be distinguished based on red schist, blue schist and intermediate surfaces (Figure 8). Infiltration rates for marl sites tended to be significantly lower in some parts of the data set. Results from summer 2005 were corrected for viscosity effect to 15°C. There was a central tendency in this figure, where uncultivated surfaces and former cultivated surfaces had overlapping GA B-values, the blue schist cultivated surfaces with crust was also located in this cluster of overlapping values. Figure 1: Rambla Nogalte Figure 2: Rambla de Torrealvilla Results pointed towards three HYSS categories in the Rambla Nogalte based on the difference in cultivation level (Figure 9). The data and analysis stressed no need to include lithological differences, which was one of the initially presumed important environmental factors in the catchments. Cold weather and even snow occurs on a regular basis if not annually, and low temperatures just over freezing point are common as a result of the high altitude. These low temperatures in the air, ground and precipitation are very likely to influence infiltration rates in cold periods in winter time (Figure 10). The is an important relationship between temperature and viscosity and infiltration rates. The digitizing of categories proved to be more accurate and easier to control (Figure 15). The image has five metre resolution (Figure 14) and the zoom tool and high resolution Google Earth were tools, which combined assured high accuracy. Areas of uncultivated surface, forest, river channels and surrounding alluvial material were strikingly small compared with the extensive cultivated areas in the upper part of the catchment. However, uncultivated areas tended to be close to the river channels, and the lack of cultivation implied that this land was poor and not worth cultivating, hence there is proportionally higher potential runoff from these areas. The digitized uncultivated surfaces and forest areas were scattered over the catchment, however the lower part of the catchment tended to be less cultivated than the upper part. The extensive cultivated areas in the upper part of the catchment indicated a relatively low contribution of runoff to the stream from this part of the catchment. Mapping Modelling A first attempt to model at-a-point runoff has been applied to three extremes: forest, natural surfaces and cultivated surfaces. The approach is using Green-Ampt: including Evapotranspiration E-T, deep percolation, rainfall intensity vs. infiltration capacity, and storage (S). More complex and advanced modelling will take into account non-uniform precipitation over larger areas and re-infiltration of runoff when water moves from one surface to another. Relationship: Sprinkler vs. minidisk Minidisk (Philips) corrected to sprinkler and converted to Green-Ampt Figure 3: Study areas Figure 5: Calvo sprinkler. Figure 6: Decagon minidisk Figure 7 Blue schist Figure 8:Summer values corrected to 15°C Figure 9: Main HYSS categories Figure 10: viscosity Figure 11: Sprinkler run raw data and modelled Figure 13: Blue schist uncultivated Figure 14: Rambla Nogalte Figure 15: Rambla Nogalte map.