Sediment yields in a changing environment

Human Impact on Erosion and Sedimentation (Proceedings of the Rabat Symposium, April 1997). IAHS Pub!, no. 245, 1997

169

Sediment yields in a changing environment: a historical reconstruction using reservoir bottom-sediments in three contrasting small catchments, North York Moors, UK JOAN LEES, IAN FOSTER, DAVID JONES Centre for Environmental Coventry CV1 5FB, UK

Research and Consultancy,

Coventry University, Prior)' Street,

PHIL OWENS, DES WALLING Geography Department,

Exeter University, Rennes Drive, Exeter EX4 4RJ, UK

GRAHAM LEEKS Institute of Hydrology,

Wallingford, Oxfordshire OXW 8BB, UK

Abstract The last one hundred years have witnessed a significant change in land use in northeast England resulting from an intensification of cultivation, increased areas of improved pasture and a significant expansion of managed upland forests. Three small contrasting reservoir catchments have been selected for a detailed investigation of the impact of changes in land use on erosion and sediment yields over the last ca. 100 years. Sediment yields have been reconstructed using multiple core correlation of reservoir bottomsediments and Pb-210 and Cs-137 dating of a master core from each reservoir. A number of fingerprinting techniques will be used in the future to document the changing sediment sources within the catchments but, in this preliminary paper, radionuclide fingerprinting of the catchment soils has been employed to determine the sources of the most recent sediments accumulating in the reservoirs.

INTRODUCTION Lake and reservoir bottom-sediments have been widely used by hydrologists and geomorphologists for reconstructing recent (ca. last 100 years) changes in suspended sediment yields and for identifying sediment source changes in the contributing catchments (cf. Foster et al, 1986; Laronne, 1990; McManus & Duck; 1993; Desloges & Gilbert, 1994; Foster & Walling, 1994; Foster, 1995; Foster et al., 1996). Many of the sediment yield changes identified can be directly attributed to human impact and changes in land use within catchments; especially afforestation (cf. Leeks, 1992). However, in the UK there is uncertainty first, as to the significance of headwater catchments in contributing to the sediment yields of major lowland rivers and, secondly, whether sediment yields are sensitive to what have been generally relatively small land use changes. Features such as riparian buffer strips may, for example, reduce the sensitivity of sediment yields to changes in soil erosion through their influence on sediment delivery (cf. Walling, 1983). This paper presents preliminary results of the application of palaeoenvironmental reconstruction methods to three reservoir-catchments in the North York Moors,

170

Joan Lees et al.

northeast England, which were selected to encompass a range of topographic and land use contrasts

in the area. The study forms part of a broader investigation

of

short- and longer-term changes in suspended sediment fluxes to the North Sea via the Humber and Tweed estuaries being undertaken by the authors. The specific objectives of this paper are: (a) To present a preliminary assessment of recent historical changes in suspended sediment yields. (b) To identify the most significant environmental factors which control changes in sediment yield through time. (c) To identify the most likely source of contemporary reservoir sediment deposits using the radionuclide signatures of potential source materials. STUDY AREA, RESEARCH METHODOLOGY AND CORE CHRONOLOGIES The study area The North York Moors (Fig. 1) is an upland region of northeast England ranging in altitude from below 20 m in the Derwent Valley to over 430 m above Ordnance Datum (AOD) at Danby High Moor, centred on the Cleveland Dome. Geologically, the region comprises a thick sequence of Middle Jurassic sandstones, grits and shales with thin coal seams, covering Lower Jurassic shales. The whole region is folded into a series of domes and basins and the major part of the upland consists of an eastward tilted scarp-edged plateau (Fig. 1). Extensive areas of boulder clay fringe the northeast coast and large areas of alluvium are to be found in the Derwent valley and in the southerly draining river valleys whose drainage originates from the high moors. The diverse land use embraces upland moorland and forestry plantations at high altitudes and on the heavily dissected scarp slopes; and sheep grazing and arable cultivation on the lower ground and in the Derwent valley. Three reservoirs, Elleron Lake, Boltby Reservoir and Newburgh Priory Pond, were selected for detailed reconstruction of the historical record of suspended sediment yield (Fig. 1). The reservoirs all have simple bathymetries; their drainage basins are all less than 7 km2 and the catchments range in altitude from 76 m to 366 m AOD. Elleron Lake, impounded in 1919, has a catchment dominated by pasture and parkland whilst Boltby Reservoir (built in 1882) has an almost entirely afforested catchment with upland coniferous plantations. Newburgh Priory Pond (built in 1760) drains a lowland catchment of mixed agricultural land use comprising arable (mainly cereal), improved pasture and some planted forestry. Summary information describing the reservoirs, their contributing catchments and land use histories is provided in Tables 1 and 2. Field survey All three lakes were surveyed and sampled in October and November 1994. A Mackereth type pneumatic corer was used to retrieve 1.2 m long undisturbed

Sediment yields in a changing environment

111

zs

o

m

•a a o

m

I o

o

Z

XI a ,_;

• ^

are ory

a < •a o PH «

>.OH •a m43 CO

•5 S3 -i -J o

*-< > 0.2 mm min"1) and short duration (T < 24 h) and with daily depths of precipitation P > 30 mm, the lack of good vegetation cover, the small water storage capacity of the soil and the high slope of the terrain, result in the occurrence of torrential floods. The kinetic energy of such torrential floods mobilizes large quantities of bed load and suspended material, which are transported through the river system and deposited downstream. Reduction of erosion and trapping of material on the slopes, before it reaches the river system is possible by application of anti-erosive afforestation. Newly

Ratko Ristic & Grigorije Macan

192

Table 1 General hydroloj ical characteristics of the experimental catchment areas. D C P L T (a.m.s.l.) (a.m.s.l) (km km"2) (km) (km) 0.400 1.075 6.25 M-I 0.0760 925 835 M-II 0.325 0.875 8.35 922 862 0.0635 982 780 0.625 1.300 6.76 M-III 0.0843 A = area; P = perimeter; L = length; D = drainage density; T = highest point; point; Isr = average slope of the terrain; It = slope of the bed. Catchment

Isr It (%) 27.1 22.50 14.9 18.46 33.0 32.32 C = confluence

established forest stands have a strong impact on the runoff regime. On the Goc mountain in central Serbia, an experimental hydrological station, M-III, has been established. The catchment area was bare land at the beginning of the research (1980), and in the same year it was afforested. The effects of the anti-erosive afforestation on runoff processes have been evaluated by comparison with two neighbouring catchment areas, M-II and M-III.

THE STUDY AREA On the Goc mountain, in central Serbia, three experimental catchments have been investigated from 1980 to 1995. Each of them is equipped with a limnigraph, pluviograph, thermograph and hygrograph. The flow measuring structures

T [year] Fig. 1 Annual specific discharge [1 s"1 km2] for the three study catchments during the period 1980-1995.

The impact of erosion control measures on runoff processes

193

incorporate suitable compensation pools with sediment traps (bed load measuring), and measuring weirs (combination of Cipolleti-Thomson). The main hydrological characteristics of the catchments are presented in Table 1. The experimental catchment M-III was originally bare land on serpentine rock, with 40% of its surface under Lasiagrostis calamagrostis grass. In 1980 it was afforested with Pinus nigra. The ground cover now ranges from 40.3-71.9%. The present soils range from the initial phase of soil development on serpentine to skeletal brown soil in the depressions. The tree density is 3000 per ha. The experimental catchment M-I is planted with 35 year old Pinus nigra, Pinus silvestris, Picea abies and oak. The ground cover ranges from 53-95%, and the area is underlain by serpentine and peridotite rocks. There are five types of soil, ranging from genetically weakly developed (skeletal silicate soils) to well developed brown soils (with layer of litter up to 12 cm deep). The tree density is 2500-3000 per ha. The microcatchment M-II is under natural meadow-pasture with Helleboro serbicae, and Danthonietum Calycinae. It is underlain by serpentine rock, on which brown humus-silicate soils have developed. The natural vegetation of the catchments would be native stands of Quercetum montanum serpentinicum, and degraded Juniperus oxicedrus. The climate is of a mountain type, with a mean annual precipitation of 822-949 mm, and an average air temperature of 6.01-8.32°C.

o o

1 C

1 D

0

3

1 (

X

1 i

C

Q

1 C

O

1 C

O

C

1 O

C

1 C

r

a

1 j

t

1 X

i

C

1 T

r

C

1 n

C

1 T

>

C

1 n

C

1 T

>

O

T

T [year] Fig. 2 Annual runoff duration for the three study catchments during the period 19801995.

194

Ratko Ristic & Grigorije Macan

RESULTS Specific discharge During the period of observation (1980-1995) the highest values of specific discharge [1 s"1 km"2], were found on the catchment areas under meadow-pasture vegetation (M-II) and with bare land on serpentine rock (M-III) (Fig. 1). When the stable stand of Pinus nigra (M-III) was established (1987 — 8 years after afforestation) the specific discharge decreased, to values similar to catchment area M-I (stable forest stand, 35 years old). Values of specific discharge were lowest for all catchment areas in 1990, as a result of the higher mean annual temperature (6.9-10.6°C) and the extremely low annual precipitation (542.8-579.2 mm). A high value of specific discharge was recorded for catchment M-III in 1991 as a result of the lower mean annual air temperature (6.4°C) and the higher annual precipitation (1018.9 mm).

Runoff duration Runoff duration (Fig. 2) is greater for the catchment areas under forest vegetation (M-I, M-III) than for the catchment area under meadow-pasture vegetation (M-II). Until 1984 the shortest runoff duration was associated with the afforested bare land (M-III). During the period 1985-1986 it was associated with the area of meadowpasture (M-II), whereas from 1987 to 1995 it was associated with the 35 year old stable forest stand (M-I).

CONCLUSIONS The establishment of stable forest stands (on bare land and instead of degraded forests or pastures) must be seen as a key anti-erosive measure in order to protect reservoir storage capacity from sedimentation. Generally, forest vegetation increases transpiration and interception but reduces the loss of water by evaporation. It also, influences the development of the soil, and especially its infiltration capacity. The specific discharge is lower but the runoff duration is longer. A regional research project undertaken on the Goc mountain in central Serbia, indicates that anti-erosive afforestation (with Pinus nigra) of bare land on serpentine rock, produced significant effects on the runoff regime after 7 years.

Human Impact on Erosion atul Sedimentation (Proceedings of the Rabat Symposium, April 1997). IAHS Publ. no. 245, 1997

195

Narrow stiff grass hedges for erosion control JERRY C. RITCHIE USDA-ARS Hydrology Laboratory, BARC-West, Bldg-007, Beltsville, Maryland 20705, USA

W. DORAL KEMPER USDA-ARS National Program Staff, BARC-West, Bldg-005, Beltsville, Maryland 20705, USA

JOHN M. ENGLERT USDA-NRCS National Plant Materials Center, BARC-East, Bldg-509, Beltsville, Maryland 20705, USA

Abstract Concentrated flow erosion is a major concern in agricultural areas around the world. Many methods have been used to reduce or slow soil loss from areas of concentrated flow erosion. Narrow, stiff grass hedges have been used to slow runoff and reduce soil loss caused by concentrated flow erosion in many countries. However, few quantitative data are available on the effectiveness of these hedges. This study was developed to study the effectiveness of narrow, stiff grass hedges for reducing soil loss from agricultural fields. Miscanthus (Miscanthus sinensis Andress) and eastern gamagrass {Tripsacum dactyloides L.) were used to establish stiff grass hedges on the contour across concentrated flow erosion areas in agricultural fields. Miscanthus hedges were established in 1991 and 1992 using transplants. Eastern gamagrass hedges, used to supplement the miscanthus hedges, were established in 1994 from seed. The miscanthus grew rapidly and formed dense hedges within two years that slowed runoff. Comparison of land survey measurements made in 1991 and 1995 found 8 to 15 cm of sediment deposition above miscanthus hedges. Deposition patterns were related to the original topography with low areas having the greatest deposition. Crop yields were reduced in the two rows closest to the hedge in two of three years. This study found that stiff grass hedges were an alternative conservation practice for reducing soil loss and dispersing runoff from areas of concentrated flow erosion in agricultural fields.

INTRODUCTION Soil erosion and concentrated flow erosion are major concerns in many parts of the world (Brown & Wolf, 1984). Grass filters and buffer strips, planted in 5-15 m wide strips, have been widely used and have been effective barriers for trapping sediment and some chemicals (Magette et al., 1989; Daniels & Gilliam, 1996). However, the effectiveness of these strips is reduced as flow increases and in areas of concentrated flow (Flanagan et al, 1989). Narrow, stiff grass hedges planted on the contour across areas of concentrated flow are an alternative method for using vegetated barriers to slow runoff and to reduce soil loss. Grass hedges have been used in many countries to reduce soil loss (National Research Council, 1993). In recent years a renewed interest has developed in the use of narrow, stiff grass hedges as a conservation practice for reducing sheet, rill, and ephemeral gully erosion from eroding fields (Kemper et al, 1992; NRC, 1993; McGregor & Dabney, 1993). Research has shown that narrow, grass hedges disperse water, trap sediment, reduce ephemeral gully development (Dabney et al., 1993, 1995), and

196

Jerry C. Ritchie et al.

reduce wind erosion (Aase & Reitz, 1989; Aase & Pikul, 1995; Siddoway, 1970). Grass hedges slow concentrated flow, thus dispersing runoff and promoting deposition in the ponded backwaters above the hedges (Dabney et al., 1993, 1995; Meyer et al, 1994) and enhancing terrace formation (Aase & Pikul, 1995). Grass hedges are an inexpensive biological conservation technology compatible with many tillage systems when planted along the contour (McGregor & Dabney, 1993). Grass hedges differ from other types of grass barriers (i.e. buffer strips, filter strips) in that they are narrow, planted with stiff, erect grasses, and are designed to stimulate the formation of terraces by deposited materials. A dense stand of coarse, stiff, grass stems planted in hedges across concentrated flow paths, causes ponding of runoff water above the hedge that allows time for eroded particles in the concentrated flow to be deposited. The deposited material fills low places in the field so that future runoff is even more broadly dispersed and less erosive. Narrow, stiff grass hedges are planted in lines along the dominate contours and across concentrated flow areas of the field (Kemper et al, 1992). The design, spacing, and lateral extent for these grass hedges in concentrated flow area depend on runoff rates, topography, and other factors (Dabney et al., 1993; Kemper et al., 1992). Vétiver (Vetiveria zizanioides (L.) Nash) is the most famous grass used for hedges to reduce erosion. Vétiver has been used in many tropical countries, but most of the reports about its effectiveness are based on empirical observations and anecdotal reports rather than quantitative studies (National Research Council, 1993). The World Bank promotes the use of vétiver for erosion control (World Bank, 1990). In 1991 the United States Department of Agriculture (USDA), Agriculture Research Service (ARS) in cooperation with the USDA Natural Resource Conservation Service (NRCS) and several Universities began a programme to evaluate narrow, stiff grass hedges for controlling soil loss from concentrated flow erosion areas. Vétiver was a candidate species, but it quickly became evident that this species could not withstand the low temperatures in temperate regions. Other grasses included in these studies were miscanthus {Miscanthus sinensis Andress) and indigenous grasses such as eastern gamagrass (Tripsacum dactyloides L.), switchgrass (Panicum virgatum L.), tall fescue (Festuca arundinacae Schreb.), perennial tall wheatgrass (Elytrigia elongata (Host) Nevski) and others. The purposes of this study were to establish narrow, stiff grass hedges across developing concentrated flow erosion areas in agricultural fields at Beltsville, Maryland, USA and to determine the effectiveness of these hedges to slow ephemeral gully development, capture eroded material, and reduce the loss of soil from the fields. METHODS AND STUDY SITES Two study sites were chosen. The first study site was on the South Farm of the USDA, ARS, Beltsville Agricultural Research Center (BARC), Beltsville, Maryland. This agricultural field has a history of strip cropping on the contour and row cropping with alternating years of corn (Zea mays L.) and soybean (Glycine max (L.) Merrill) on alternate strips. Slopes in the field are between 10-15% with a total slope length of 250 m. Cropping strips are approximately 50 m wide. Corn is no-till

Narrow stiff grass hedges for erosion control

197

planted into the soybean stubble while soybeans are planted after minimum tillage (surface disking) to incorporate the corn residue. Two concentrated flow erosion areas were observed in the field, starting near the crest of the slope and crossing three cropping strips before joining near the base of the slope. On 17 April 1991, miscanthus was transplanted along the contour between strips of crops and across concentrated flow erosion areas below where the two concentrated flow areas joined near the base of the slope. Miscanthus was transplanted using 2-5 cm clumps at 10 to 15 cm intervals. Transplants were made on the contour between strips of crops to reduce interference with farm operations and to reduce disturbance to the hedge during the field and harvest operations. In May 1994, the hedges were repaired by transplanting miscanthus to fill gaps in the original hedges. Also in May 1994, eastern gamagrass was planted using seeds in gaps and used to extend the length of this miscanthus hedge. Corn and soybeans yields were measured on either side of the miscanthus hedge in the South Farm field in 1993, 1994, and 1995. Sampling was done by harvesting crop rows 1, 2, 4, 8, 16, and 32 away from the miscanthus hedge. In each sample row four (4) 5 m sections of crop were harvested. Samples of the harvested crops were dried and yields were determined. In 1993 no treatment was made to the hedge. In 1994 and 1995 half the length of the hedge was kept trimmed to 75 cm or the maximum height of the soybeans during the growing season in an attempt to reduce shading of the crops by the hedge. In April 1991, shortly after the original transplanting of the miscanthus hedge, a topographic survey was made at the hedge site. In August 1995, a second topographic survey was made for comparison with the original topographic surveys. Lines were surveyed 5 cm below and 5 cm and 1 m above the hedge. The second study site is on the East Farm of Beltsville Agricultural Research Center. This agricultural field has a slope of 10 to 15% with a total slope length of about 200 m. The field had a history of being planted in either corn or soybeans. A concentrated flow erosion area was visible in the field. In April 1991, a tile drain was installed in the approximate location of the concentrated flow area. On 23 May 1991, after the installation of the tile drain, miscanthus was transplanted into a hedge at the lower edge in this field where overland flow exited the field and entered a wooded area. Miscanthus was transplanted in 2-5 cm clumps at 10 to 15 cm intervals. In 1991 and 1992, the field was surface ploughed and planted in corn. After the corn was harvested in September 1992, clover was no-till planted in the field to provide a winter cover. During the winter and spring of 1992/1993, a conservation plan was developed for this field to reduce soil loss. This plan directly affected our activities at the East Farm field by changing the farming practices from a single field to a field with five strips of crops. This plan did not affect the original miscanthus hedge. On 24 March 1993, two new miscanthus hedges were transplanted on the contour between the newly developed strips of crops. The three miscanthus hedges grew actively during 1993 and are now well established. After the 1993 growing season, row crop agriculture was stopped in the field. The field has been planted with small grain/clover since 1993 that provides continuous cover. In 1994, gaps in the hedge were filled and the hedge was extended in length by planting eastern gamagrass seeds. In 1995, a topographic survey was made along the hedge at the edge of the

198


field for comparison with surveys made in 1993. Lines were surveyed at 5 cm below and at 5 cm and 1 m above the hedge. RESULTS Miscanthus rapidly formed a dense hedge. Beginning with 2-5 cm clumps planted 10 to 15 cm apart in 1991, hedges in both fields have grown to a width of 20-30 cm and a height of 2.5 to 3.5 m by 1994. At the South Farm, miscanthus was 2 m tall at the end of the first (1991) growing season. Trimming this hedge to 75 cm along half its length during the 1994 and 1995 growing seasons did not affect its growth or expansion. Miscanthus quickly developed new growth and continued to expand after each trimming. Each spring, the hedges were trimmed to a height of approximately 30 cm. Trimmed material was left in the field where it fell. In 1994, eastern gamagrass was planted from seed to fill gaps and expand the length of the hedges at both field sites. These plantings were successful with high rates of seed germination, although eastern gamagrass is considered difficult to germinate. Dewald et al. (1996) discuss the proper techniques for establishing eastern gamagrass. Eastern gamagrass grew rapidly to form a hedge 30-60 cm tall and 10-15 cm wide by the end of the 1994 growing season. In 1995 and 1996, these eastern gamagrass hedges continued to grow and are developing into dense hedges of 0.2-0.3 m width and 3 m height. Topographic surveys (Fig. 1) of the miscanthus hedge at the East Farm in 1995 showed 10 to 15 cm of deposition above the hedge after four years. The deposition area extends at least 1 m above the hedge. More detailed surveys are needed to determine the full extent of the deposition pattern near this hedge. This hedge at the lower edge of the field is capturing eroding particles that have moved from the field. However, this field is now in continuous grass/clover cover so that the soil loss and 0.0

-0.3

o *-

iu -0.6

-0.9

ABOVE + 1 M ABOVE BELOW

GRASS HEDGES BARC-East Farm 0

10

20 30 METRES Fig. 1 Topographic survey made in 1995 of the grass hedge at the BARC East Farm at Beltsville, Maryland, USA. Survey lines were 5 cm below the hedge, 5 cm above the hedge, and 1 m above the hedge.


199

GRASS HEDGES BARC-South Farm

0.0

ABOVE HEDGE

-0.06

/""" /

-

e/J

g -0.12 ILU S

-0.18 "

\

''•• ••••"•••/ '

-0.24 -0.30

y

^ ^

BELOW HEDGE

Vv

METRES Fig. 2 Topographic survey made in 1995 of the grass hedge at the BARC South Farm at Beltsville, Maryland, USA. Survey lines were 5 cm below the hedge and 5 cm above the hedge.

concentrated flow area development have been reduced over the past two years. Observations in 1995 and 1996 noted the lack of well defined concentrated flow channels in the area near this hedge that were evident in early spring of 1991 and 1992. Only during large rainfall events does surface erosion occur in this field. The lack of development of the concentrated flow area could be due to a) reduced runoff from the field due to the continuous crop cover or b) the development of the terrace area above the hedge that further disperses the water as it approaches the hedge. Surveys above and below the miscanthus hedge on the South Farm site in 1995 also showed 8 to 15 cm of deposition above the hedge (Fig. 2). Greater deposition occurred along the hedge in the areas where concentrated flow areas (depressions in the topographic survey) cross the border between the strips of crops. A general smoothing of the topography above the hedge is occurring. This smoothing is attributed to the hedge slowing and spreading the water across a wider area as it crossed the hedge barrier. This ponding of water would allow the sediment load carried in the runoff to be deposited over a wider area. A comparison of topographic surveys made in April 1991 and August 1995 along the same survey line (5 cm upslope of the hedge) at the South Farm also showed an 8-15 cm depth of deposition (Fig. 3). Again the deposition is greatest in areas where concentrated flow had eroded the deepest channels before the establishment of the hedge. The measured deposition along this survey line is greater in the low areas. That is also evident in Fig. 2 where a comparison was made between above and below survey lines. The August 1995 survey line in Fig. 3 is the same as the "above" survey line in Fig. 2 showing that the comparison of above and below the hedge survey line probably give a conservative estimate of total deposition. At the South Farm site an extensive area of deposition of material is present approximately 50 m above and west of the centre of the hedge. Whether this deposition area is due to the hedge is not clear, since survey data are not available for comparison, but this area of deposition has developed since the hedge was

200


GRASS HEDGES

0.0

BARC-South Farm

AUGUST 1995

-0.06

5 -0.18 APRIL 1991 -0.24 -0.30 0

10

20 30 METRES Fig. 3 Comparison of topographic surveys at the BARC South Farm of the same line (5 cm above the hedge) shortly after the hedge had been transplanted in April 1991 and again in August 1995.

established. There is also evidence of the development of new concentrated flow areas below the hedge that is of concern. More extensive surveys of the area above and below the hedge need to be made to evaluate the full extent of the deposition patterns and rates. Studies of crop yield near the miscanthus hedge at the South Farm site were begun in 1993 after the hedge was well established. These studies were to determine the effect of the hedge on crop yields near the hedge and to determine the distance the hedge effects could be measured. In 1993, substantial decreases in yields of corn and soybeans occurred near the hedge. For soybeans, yields reached maximum level at approximately 2 m from the hedge. For corn, yields reached maximum level at approximately 6 m from the hedge. Rainfall was below average during the 1993 growing season. The hedge grew to heights of 2-3 m shading the adjacent rows of soybeans. In the early part of the growing season the hedge shaded adjacent rows of corn although the corn eventually grew to heights greater than the hedge. Plant populations were reduced in the first two rows near the hedge. In 1994, half the length of the hedge was kept trimmed to a maximum height of 75 cm. Only the first row of soybeans showed yields that could be related to the proximity and height of the hedge. Shading was the probable cause of decrease in yields of soybeans next to the hedge that was not kept trimmed. Few soybean plants developed and matured in the row closest to the hedge. The differences in yield patterns with distance from the hedge between 1993 and 1994 were probably due to rainfall. In 1994, rainfall was slightly above average and was adequate to meet the needs of both the hedge and the crops growing near it during the growing season. The well-established roots of the perennial miscanthus hedge could deprive the crop of moisture in dry years. Yields in 1995 were similar to 1993 yields, with yields lower in the rows near the miscanthus hedge. However, in 1995 yields were only affected in the first two


201

rows for the soybeans and corn. Keeping the hedge trimmed to 75 cm did not affect yields of the corn or soybeans. DISCUSSION Miscanthus hedges were established easily and expansion by shoot production has been rapid and vigorous. During the five-year study no evidence has been found that miscanthus produced viable seeds. While these miscanthus hedges are very robust and are capturing eroded materials, the cost of purchasing miscanthus shoots and the labour needed to transplant the shoots may reduce farmer acceptance and application of miscanthus hedges as a conservation practice. The use of an indigenous grass that could be seeded and used for other purposes should have greater farmer acceptance. Switchgrass and eastern gamagrass are good candidates for use in the eastern United States. Eastern gamagrass was planted successfully thus making the establishment of hedges easier and cheaper. Farmers can use conventional farm equipment to plant eastern gamagrass hedges using seeds (Dewald & Louthan, 1979; Dewald et al., 1996). The use of indigenous grasses also reduces the chances of the introduction of an unwanted competitor to agricultural fields. While our emphasis has been on narrow (< 1 m) grass strips, developing grass strips 1-5 m wide with indigenous grasses could provide both erosion protection and the potential for harvesting as a feed crop. Miscanthus hedges are trapping eroded materials under field conditions. Two to four centimetres of eroded material have been deposited annually immediately above the hedges. Grass hedges have been found to trap two-thirds of the sediment from small plots (McGregor & Dabney, 1993; Dabney et al., 1993). The hedge row is working as a porous filter that slows the water but lets it pass. Water is ponded above the hedge row slowing its velocity and allowing time for part of the material in suspension to be deposited. The deposition is occurring in the area above the hedges rather than in the hedge row itself. Flume studies have shown that hedges of switchgrass, vétiver, and miscanthus caused backwater depths of up to 40 cm and trapped more than 90% of sediment greater than 125 /xm. Trapping efficiency was more related to particle size than to flow rates. Sediment trapping was upslope of the hedge rather than being filtered by the hedge. Once the material reaches the hedge, it passes through (Meyer et al., 1995). Over time, the hedge can cause the development of terraces (Aase & Pikul, 1995) that flatten the slope and broadens the flow area resulting in larger ponding areas and greater storage capacities, increased settling times, and lower flow rates through the hedge (Dabney et al, 1996). Soil deposited above established hedges will flatten the slope. Concentrated flow areas above the hedge fill rapidly. However, incised areas below the hedge may be increased due to the increased erosive power of the water passing through the hedge that has increased carrying capacity due to the sediment deposited above the hedge. This erosion below the hedge should be controlled. In time, terraces may be complete so that the areas between hedges are flattened and erosion reduced. Stiff grass hedges should not be seen as a panacea to reduce erosion but as another tool in the arsenal of weapons to manage the landscape. Conservation practices should be in place on the field to prevent the movement of soil so that the need for hedges is reduced.

202


Three years of yield studies suggest that stiff grass hedges will probably affect yields of crops in the first few rows from the hedges. Lyles et al. (1984) found that vegetative barriers could affect yields at distances up to 2 times the height of the barrier. The reduction of yields probably is dependent on shading and water use although many other factors (i.e. nutrient supply, deposition) may contribute to changes in yields (Lyles et al., 1984). At the South Farm site, field observations noted fewer plants grew near the hedge and in areas where deposition was the greatest. CONCLUSIONS In a series of recent studies, quantitative data have been collected that show that narrow, stiff grass hedges do act as a filter to slow and broaden the flow area, resulting in ponding that increases settling times for suspended material. This causes the development of terraces that further reduce the steepness of slopes giving even larger areas for the water to spread. Narrow, stiff grass hedges should not be seen as a panacea but as another tool to control soil loss from agricultural fields. Continued efforts to control soil loss at the point of detachment are critical. Proper management of stiff grass hedges is required. With the development of terraces, there is an increased potential for the development of sediment patterns that may concentrate flow passing through the hedge creating conditions for the development of erosion problems immediately below hedges. While grasses such as vétiver and miscanthus are good candidates for narrow, stiff grass hedges, indigenous grasses should be used when possible to reduce the potential for the introduction of exotic material into new environments. Planting hedges of indigenous grasses in wider strips (2 to 5 m) also raises the potential for harvesting or grazing these strips. Thus soil loss could be reduced and the farmer could have a crop that could provide added income.

Acknowledgements The authors express our appreciation to personnel from the NRCS National Plant Material Center at Beltsville, Maryland who helped transplant the miscanthus. We also thank Tim Badger, Farm Manager at the Beltsville Agricultural Research Center, who helped find study sites and allowed us to plant strange grasses in the fields where he attempts to eliminate competitors for the field crops. Carole and Karen Ritchie provided field assistance during the topographic surveys.

REFERENCES Aase, J. K. & Piku], Jr, J. L. (1995) Terrace formation in cropping strips protected by tall wheat grass barriers. J. Soil Wat. Conserv. 50(1), 110-112. Aase, J. K. & Reitz, L. L. (1989) Conservation production systems with and without grass barriers in the northern Great Plains. /. Soil Wat. Conserv. 44(4), 320-323. Brown, L. R. & Wolf, E. C. (1984) Soil Erosion: the Quiet Crisis in the World Economy. WorldWatch Paper #60, Washington, DC. Dabney, S. M., McGregor, K. C , Meyer, L. D., Grissinger, E. H. & Foster, G. R. (1993) Vegetative barriers for runoff and sediment control. In: Integrated Resource Management and Landscape Modification for Environmental Protection (ed. by J. K. Mitchell), 60-70. Am. Soc. Agric. Engrs, St Joseph, Michigan, USA.


203

Dabney, S. M., Meyer, L. D. Dunn, G. H., Foster, G. R. & Alonso, C. V. (1996) Stiff-grass hedges a vegetative alternative for sediment control. In: Proc. 1996 Interagency Sedimentation Conference. Dabney, S. M., Meyer, L. D., Harmon, W. C., Alonso, C. V. & Foster, G. R. (1995) Depositional patterns of sediment trapped by grass hedge. Trans. Am. Soc. Agric. Engrs 38(6), 1719-1729. Daniels, R. B. & Gilliam, J. W. (1996) Sediment and chemical reduction by grass and riparian filters. Soil Sci. Soc. Am. J. 60(2), 246-251. Dewald, C , Henry, J., Bruckerhoff, S., Ritchie, J., Dabney, S., Shepard, D., Douglas, J. & Wolf, D. (1996) Guidelines for the establishment of warm season grass hedge for erosion control. J. Soil Wat. Conserv. 51(1), 1620. Dewald, C. L. & Lougthan, V. H. (1979) Sequential development of shoot systems components in eastern gamagrass. J. Range Management 32(1), 147-151. Flanagan, D. C Foster, G. R., Neibling, W. H. & Burt, J. P. (1989) Simplified equations for filter strip design. Trans. Am. Soc. Agric. Engrs 32(6), 2001-2007. Kemper, D., Dabney, S., Kramer, L., Dominick, D. & Keep, T. (1992) Hedging against erosion. J. Soil Wat. Conserv. 47(4), 284-288. Lyles, L., Tatarko, J. & Dickerson, J. D. (1984) Windbreak effects on soil water and wheat yield. Trans. Am. Soc. Agric. Engrs 27(1), 69-72. Magette, W. L., Brinsfield, R. B., Palmer, R. E. & Wood, J. D. (1989) Nutrient and sediment removal by vegetated filter strips. Trans. Am. Soc. Agric. Engrs 32(2), 663-667. McGregor, K. C. & Dabney, S. M. (1993) Grass hedges reduce soil loss on no-till and conventional-till cotton plots. In: Proc. Southern Conserv. Tillage Conf. For Sustainable Agriculture, 16-20. Meyer, L. D., Dabney, S. M. & Harmon, W. C. (1994) Sediment-trapping effectiveness of stiff-grass hedge. Trans. Am. Soc. Agric. Engrs 38(3), 809-815. National Research Council (NRC) (1993) Vétiver Grass: A Thin Green Line Against Erosion. Board on Science and Technology for International Development, National Academy Press, Washington, DC. Siddoway, F. H. (1970) Barriers for wind erosion control and water conservation. J. Soil Wat. Conserv. 25(2), 181-184. World Bank (1990) Vétiver Grass - The Hedge Against Erosion. Washington, DC.

Human Impact on Erosion and Sedimentation (Proceedings of Rabat Symposium S6, April 1997). 1AHS Publ. no. 245, 1997

205

Use of satellite imagery to determine the land use management factors of the USLE ZAINAL ABIDIN ROSLAN Department of Civil Engineering, MARA Institute of Technology, Shah Alain, Selangor, Malaysia

KIA HUI TEW Faculty of Civil Engineering, Universiti Teknologi Malaysia, Skudai, Johor, Malaysia

Abstract The Cameron Highlands, an idyllic highland resort in Malaysia, are now being threatened by landslips and flash floods due to numerous development projects as well as intensive agricultural activities being associated with tourism. A study of the land use management factor (Cover and Management factor, C, and Support Practice factor, P) of the Universal Soil Loss Equation (USLE) has been undertaken, as this parameter reflects the land cover in the study area and its effect on soil erosion. Using remote sensing satellite imagery, the Ringlet area in the Cameron Highlands has been identified as having the highest erosion risk/loss based on the CP factor and this is confirmed by its highest ranking for residential and construction areas compared to other locations. The results of this study highlight the important land uses associated with erosion risk, and provide guidance to ensure that development is carried out to ensure a quality environment for the future. INTRODUCTION The Cover and Management factor, C in the USLE represents the combined effect of all the interrelated cover and management variables and is defined as the ratio of soil loss from land cropped under the specified conditions to the corresponding loss from clean-tilled continuous fallow. In general, whenever sloping soil is to be cultivated and exposed to erosive rains, the protection offered by grass cover or close-growing crops needs to be supported by conservation practices that will slow the runoff and thus reduce the amount of soil that it can transport (Roslan, 1993). The most important of these supporting cropland practices are contour tillage, strip cropping on the contour, and terrace systems. The Support Practice factor, P in the USLE is defined as the ratio of soil loss with a specific support practice to the corresponding loss with up-and-down-slope cultivation (Wischmeier & Smith, 1978). Improved tillage practices, grass-based rotations, fertility treatments and the leaving of greater quantities of crop residues on the field can contribute significantly to erosion control. With recent developments in remote sensing technology, colour infrared imagery can be used to determine the combined land use management factor, C and P of the USLE as shown in Table 1. DATA The data used for this study were derived from a Landsat Thematic Mapper image

206

Zainal Abidin Roslan & Kia Hui Tew

Fig. 1 The 11 locations in the Cameron Highlands with their respective zoning within a 5 km radius.

Satellite imagery to determine the land use management factors of USLE

207

dated 23 April 1990, which was obtained from MACRES (Malaysian Center of Remote Sensing) for the area covering the Cameron Highlands catchment in the State of Pahang, Malaysia.

METHOD Eleven locations have been identified in the Cameron Highlands as indicated in Fig. 1 and these are listed in Table 2. Attention has focused on a 5 km radius zone surrounding each of the locations. The satellite image covering the Cameron Highlands is shown in Fig. 2.

RESULTS The Landsat Thematic Mapper image has been processed using the colour infrared interpretative key presented in Table 1 to obtain values of the CP factor for the 11 locations in the Cameron Highlands. Unclassified and water body areas are not included in the calculation of the CP factor since information on land use and cropping or land management was not available. As an example, the results for the Table 1 The colour infrared interpretative key for the land use management factors C and P of the USLE. Land use management Grassland/hay Residential Forest Rangeland (grass and weeds) Cropland Construction areas Impervious areas a b

Photo characteristics Pink tones, smooth texture Pink tones, houses/streets Red tones, coarse texture Variable colours Pink tones, cultivation White tones, coarse texture Bluish-white, smooth texture

CP factor value 0.003 0.003 0.003 0.007-0.450" 0.300-0.400" 1.000 0.005

For rangeland, the average CP factor is taken as 0.229 For cropland, the average CP factor is taken as 0.35

Table 2 The 11 locations in the Cameron Highlands with their respective 5 km radius zones. Zone no. 1 2 3 4 5 6 7 8 9 10 11

Location Ringlet Ldg. Teh Boh Tanali Rata Brinchang Tringkap Ldg. Sg. Palas Gunung Brinchang Kuala Terla Kg. Raja Sg.Ikan Blue Valley


208

Ringlet location (Zone 1), with its associates zoning within a 5 km radius, is shown in Fig. 3. Appendix 1 provides further details concerning the procedure for ranking the various land use management factors C and P for the 11 locations and a typical result (residential areas) is shown in Table 3. The weighted CP factors for each location are as shown in Table 4.

i

• a.

1

i

a,

•

., \A

•l

•

"•

•

S

•

.

m

J

1.

1

#

* *

r

••

.Ê

•

'

J

.

&

•

.7-•

-,

•

•'. .•

• e

r

f».*

•- •

v ;

,ig'-..

i »>.

•

"»,

< r

'• .-

: • ji

;

'f. '•••

-

I

.

' • / /

•

•'

*

..x *.,• . a

v?"'.?

e 1 1

...

••

i -/•:"*.«ii'-"

-v

.'•'..,.&

•7*«.:-:»î

•'•'

•••••

'

-

•

Fig. 2 The Landsat Thematic Mapper image of the Cameron Highlands.


209

COMMENTS AND DISCUSSIONS The observations carried out using the remote sensing image have a resolution of 30 m which means that each pixel inside the image will represent an area of 30 m x 30 m. Although this image will provide a greater accuracy compared to land use maps, roads and other buildings with a width of less than 30 m will not be able to be identified. Further work using a higher resolution image could be carried out in the future to provide improved representation of the CP factor. However, this study

mB* . ' ".- '/

I

••

a n

:^His

.,:,*i0:$M RANGELAND

IMPERVIOUS

CROPLAND

RESIDENTIAL

GRASSLAND

CONSTRUCTION

Fig. 3 Land use management within a 5 km radius of the Ringlet location (Zone 1).


210

Table 3 Ranking of the 11 locations in the Cameron Highlands in terms of the area of residential development. Rank 1 2 3 4 5 6 7 8 9 10 11

Location Ringlet Kuala Terla Ldg. Sg. Palas Tringkap Ldg. Teh Boh Brinchang Tanah Rata Gunung Brinchang Kg. Raja Blue Valley Sg. Ikan

Area (ha) 0.46 0.46 0.44 0.42 0.26 0.26 0.25 0.24 0.24 0.18 0.15

% Image 5.39 5.34 5.15 4.87 3.00 3.00 2.90 2.78 2.76 2.13 1.71

Table 4 Ranking of the weighted CP factor for the 11 locations in the Cameron Highlands. Rank 1 2 3 4 5 6 7 8 9 10 11

Location Ringlet Kuala Terla Ldg. Sg. Palas Tringkap Ldg. Teh Boh Gunung Brinchang Brinchang Tanah Rata Kg. Raja Blue Valley Sg. Ikan

CP factor 0.0875 0.0830 0.0795 0.0763 0.0762 0.0543 0.0540 0.0504 0.0501 0.0471 0.0355

highlights the dominant land use at various locations in the Cameron Highlands and provides some indication of the possibility of reducing future erosion risk/loss, based on the CP factor of the USLE. CONCLUSION From the analysis undertaken, it has been found that the Ringlet location has the highest erosion risk/loss based on the CP factor of the USLE with a value of 0.0875. This result is supported and justified by the fact that this location produced the highest ranking for residential (5.39%) and construction (2.07%) areas compared to other locations. Conversely, the location at Sg. Ikan has the lowest erosion risk/loss, as the CP factor is only 0.0355. This result is again supported and justified by the fact that this location has the largest area covered by forest (84.21%). This area has yet to be developed and therefore has a lower erosion risk/loss. In terms of other land use categories, Sg. Ikan recorded the lowest value for residential (1.71%) and


211

construction (0.73%) areas which further justifies the results obtained. Knowledge of the ranking of the CP factor at various locations in the Cameron Highlands, will ensure that the development of any new projects will be well planned and designed to ensure that the quality environment is maintained. Further work with a higher resolution image should be carried out in order to provide the best representation of the land use in the Cameron Highlands and liaison with the State Government is required to implement stricter laws in developing particular areas, in order to ensure sound and sustainable development in the future. Acknowledgement The authors wish to thank the MACRES (Malaysian Center of Remote Sensing) for providing the satellite imagery and the Universiti Teknologi Malaysia Center of Remote Sensing (PRSUTM) for processing the imagery.

REFERENCES Roslan, Z. A. (1993) Environmental assessment on soil erosion using USLE model. In: Technical Workshop on Hydrological Research in a Changing Environment in Sub-Humid and Humid Tropical Areas (Bangi, Selangor, Malaysia, 15-18 June 1993). Wischmeier, W. H. & Smith, D. D. (1978) Predicting Rainfall Erosion Losses — A Guide to Conservation Planning. USDA, Agriculture Handbook no. 537.

APPENDIX Sample calculation of the land use management factors C and P of the USLE for the Ringlet location (Zone 1) No.

Land use management

1. 2. 3. 4. 5. 6. 7. 8. 9.

Water body Forest Rangeland Cropland Grassland Impervious Residential Construction Unclassified Total

% Image (1) 0.34 62.06 9.43 12.27 5.10 2.37 5.39 2.07 0.97 100.00

CP factor from Table 1. (2)

CP factor (1)*(2)

0.003 0.229 0.350 0.003 0.005 0.003 1.000

0.0019 0.0216 0.0429 0.0002 0.0001 0.0002 0.0207 0.0875

Weighted CP factor, [(0.003) x 62.06%] + [(0.229) x 9.43%] + [(0.350) x 12.27%] + [(0.003) x 5.10%] + [(0.005) x 2.37%] + [(0.003) x 5.39%] + [(1.000) x 2.07%] = 0.0875

Human Impact on Erosion and Sedimentation (Proceedings of Rabat Symposium S6, April 1997). IAHS Pubf no. 245, 1997

91o LlD

Application of a distributed Shallow Landslide Analysis Model (dSLAM) to managed forested catchments in Oregon, USA WEIMIN WU Cacheva International Corporation, Santa Clara, California, USA

ROY C. SIDLE

LOICZ Core Project, Netherlands Institute for Sea Research, PO Box 59, 1790 AB Deb Burg-texel, The Netherlands

Abstract A distributed, physically-based slope stability model (dSLAM) for analysing rapid, shallow landslides and the spatial distribution of the safety factor (FS) was tested in two catchments of the Cedar Creek watershed in the Oregon Coast Ranges. Simulated volumes and numbers of failures for a large storm in 1975 agreed closely with field measurements. The uncertainty associated with sensitive input parameters may complicate this comparison. For example, when soil cohesion values of 2.0 kPa and 3.0 kPa were used, the failure volume changed by factors of 2.04 and 0.41, respectively, for basin A, and 2.93 and 0.29, respectively, for basin B, compared with the standard condition of 2.5 kPa used in the simulation. Changes in values for soil depth and root cohesion that reflect potential natural variabilities, produced differences of up to several fold in simulated landslide volumes. As the simulated storm sequence progressed, FS declined sharply in hillslope hollows of basin A, the sites of most landslides. A similar response was noted in basin B, although the most unstable elements in this basin were associated with steep slopes. All areas with FS < 2.0 were clearcut in 1968, 7 years prior to the storm. INTRODUCTION Most research on shallow landslides has focused on understanding temporal and onsite conditions and processes (Wu et al., 1979; Sidle, 1992). To better understand the processes controlling landslides and to design appropriate land use strategies, it is necessary to evaluate slope stability and predict the occurrence of landslides in both temporal and spatial dimensions because of the distributed properties of site variables and land uses (Ward et al, 1982; Carrara et al., 1991). When a distributed, physically-based approach is applied in basin slope stability analysis, not only are the distributed properties of site parameters of concern, but also the model output presents a spatial problem, because we need to determine the locations of slope failure. This is different from most runoff and erosion models which only predict outputs at basin outlets. Although geographic information system (GIS) technology is regarded as ideal tool for landslide analysis in terms of spatial data extraction and display (Carrara et al., 1991), no progress has been reported in integrating distributed, physically-based slope stability modelling with GIS. In this paper we discuss the application of dSLAM in two unstable forested basins in coastal Oregon subject to recent clearcutting. The effects of potential natural variabilities of input parameters on slope stability predictions are evaluated.

214

Weimin Wu & Roy C. Sidle

THEORETICAL BACKGROUND The dSLAM model is a distributed, physically-based model that combines an infinite slope model, a kinematic wave groundwater model, and a continuous change vegetation root strength model to analyse shallow, rapid landslides and the spatial distribution of the factor of safety in steep forested terrain. The infinite slope model is modified to include the effects of vegetation root strength and tree weight as follows (Sidle, 1992) _C+AC+{[(Z-/z)ym+/zYsa,-ftyw]cos2P+Wcos[3}tan4) [(Z-A)y m +AyJsinPcosP+Wsinp where C is the effective soil cohesion, j is the effective internal friction angle, AC is the cohesion attributed to root strength, W is the vegetation surcharge, ym and ysaI are the unit weights of soil at field moisture content and saturation, respectively, Z is the vertical soil thickness, h is the vertical height of the water table, P is the slope angle, and yw is the unit weight of water. In the application of dSLAM to steep forested areas, it is assumed that the infiltration capacity of the soil is always in excess of the rainfall intensity, thus only subsurface flow and non-Hortonian overland flow occur. Combined slope-parallel subsurface and saturated surface flows are routed by the kinematic wave model through topographically generated streamlines which lie orthogonal to the slope contours. For evaluating the continuous changes in vegetation root strength and surcharge following timber harvesting, Sidle's (1991, 1992) models are used. We assumed that all landslides are totally mobilized into debris flows. The extent of debris flow runout was based on channel gradient (70°) criteria (Wu & Sidle, 1995). The dSLAM model is integrated with a contour line-based topographic analysis and a geographic information system (GIS) for spatial data extraction and display. Moore et al.'s (1988) TAPES-C model that subdivides an area into irregular polygons bounded by adjacent contour lines and adjacent streamlines is adopted. The GIS used in the study is the ARC/INFO system.

SITE CONDITIONS The dSLAM model was tested in two second-order basins (A=1.18 km2 and B = 1.12 km2) within the Cedar Creek watershed in the Oregon Coast Ranges in which both the temporal and spatial patterns of timber harvesting were known and different. The study basins have relatively shallow mantles of permeable, lowcohesion soils overlying sandstone bedrock. Forest cover consists of Douglas-fir with small pockets of red alder. Other site details are given by Wu & Sidle (1995). From the early 1950s through to the late 1960s timber harvesting and development of logging roads increased in this area. Because the slopes are very steep and the soil strength is very low, periods of heavy rainfall often trigger landslides. A series of high intensity storms in the mid-1970s triggered a large number of landslides in the area, both from roads and within clearcuts.

Application of a distributed Shallow Landslide Analysis Model (dSLAM) in Oregon

215

METHODOLOGY The 7.5' USGS digital elevation model (DEM) of Goodwin Peak Quadrangle was used to derive the vector DEM input for the TAPES-C model, a topographic analysis program developed by Moore et al. (1988). Once the vector DEM was generated, it was used by TAPES-C to derive the contour line-based network and to calculate topographic attributes. The network was then used as a framework for generation of other distributed information and visualization of results. Details of the operational procedures and discussion of DEM accuracy are given by Wu & Sidle (1995). There are 13 595 elements (average element size = 82.4 m2) and 66 contours in basin B. In basin A, the average element size is 100.5 m2 with 52 contours and 11 764 elements. The spatial data for soil and vegetation were extracted from a GIS maintained by the Siuslaw National Forest. Past harvesting patterns and the spatial distribution of different soil types are documented by Wu (1993). Soil depths and saturated hydraulic conductivities (£sat) used in simulations were 1.0-1.5 m and 0.5-1.0 m h"', respectively. Soil engineering properties, root cohesion, and vegetation surcharge are documented by Wu & Sidle (1995). Heads of perennial channels were defined as having a contributing area of at least 2500 m2 based on nearby studies (Montgomery & Dietrich, 1988). No adequate method was found to assess FS for channelized elements, and they were therefore excluded from the slope stability analysis. Most channels in this area are eroded to bedrock, thus this assumption seemed reasonable. A major rainstorm in late November of 1975 that caused widespread landsliding near Cedar Creek was chosen for the simulation. Antecedent moisture conditions were relatively wet (18 mm of rain in the 24 h preceding the storm). Three landslides with a total volume of 734 m3 occurred in basin A and six landslides (total volume=749 m3) occurred in basin B according to a post-storm survey (Greswell et al, 1979).

RESULTS OF SLOPE STABILITY SIMULATIONS During the 1975 rainstorm, a total landslide volume of 733 m3 was simulated in basin A. These four landslides occurred on steep slopes (36.5 to 47.9°) that were clearcut in 1968. The three landslides in the eastern portion of the basin converged into a single debris flow. A debris flow initiating further to the west joined this debris flow at the channel junction. The combined debris flow produced by these two source areas was deposited near the outlet of basin A due to the gentle channel gradient in that reach (see Wu & Sidle, 1995). In basin B, seven discrete landslides were simulated with a total volume of 801 m3. Slope gradients of failed elements in basin B ranged from 38.4 to 45.1°. Again, all landslides occurred in areas harvested in 1968. The seven landslides occurred close together in the same tributary and the total volume of these failures moved out of the basin as a debris flow. The simulated landslide volumes and the numbers of landslides are very close to the inventory data from the 1975 landslide survey: 734 m3 and 3, in basin A, and

216


749 m3 and 6, in basin B. One more landslide was simulated in each basin than actually occurred during the 1975 storm. However, the scale of resolution for separating individual landslides both in the simulations and in the field surveys negates this small difference. The locations of the failed elements are reasonable, although it appears that the sites of landslides in basin A are more consistent with sites where groundwater flow converges; in basin B, some failure sites are more related to steep slopes. Spatial and temporal changes in FS for basin B during the November 1975 storm are shown in Fig. 1. The temporal sequence (0, 22, and 62 h from the beginning of the storm) represents pre-storm, end of storm, and post-storm FS distributions. Similar data for basin A are presented by Wu & Sidle (1995). In both basins, prestorm patterns of FS were largely controlled by slope gradient and timber harvesting patterns. The most unstable sites corresponded to areas clearcut in 1968, 7 years prior to the storm. This timing corresponds to the predicted minimum in net root cohesion following timber harvest and subsequent regrowth (Sidle, 1991). During the rainstorm, areas with low FS values (1.0-1.6) expanded dramatically and after rainfall ceased, these areas slowly decreased in size. Prior to the storm, channel heads were relatively stable in both basins, however, as rainfall progressed FS declined sharply in the vicinity of these areas. This rapid decline is more apparent in basin A than in basin B. The location of the most unstable elements in basin B appear to be more related to steep slopes (Fig. 1) rather than to the unstable head hollow sites which appeared to be the primary initiation zones in basin A.

EFFECTS OF PARAMETER UNCERTAINTY ON SLOPE STABILITY Even though dSLAM closely predicted numbers and volumes of landslides during the simulated 1975 storm, we need to test the sensitivity of the model to variations in input parameters for reasonable ranges that may occur in the field. Although some of the input data were spatially distributed (e.g. soil depth, vegetation, timber harvesting patterns), other data (e.g. soil engineering properties) were averaged over the entire basin due to lack of spatially distributed data. Furthermore, many data that were available in a spatially distributed form did not adequately represent the desired degree of resolution for this type of analysis. Changes in soil cohesion in the range reported by Schroeder & Alto (1983) for the area had the greatest effect on FS of all parameters tested (Fig. 2). When the highest value of C (6.86 kPa) was used, no landslides occurred in either basin. However, when cohesion was reduced to 0 kPa, failure volume increased 29- and 85fold in basins A and B, respectively, above the volumes obtained when the average cohesion value (2.5 kPa) was used in the simulations (Fig. 2). If soil cohesion is decreased from 2.5 kPa to 0 kPa, more land area in basin A becomes marginally stable than in basin B. If C values of 2.0 and 3.0 kPa are used, the failure volume changes by factors of 2.04 and 0.41, respectively, for basin A and 2.93 and 0.29, respectively, for basin B compared with the standard condition. These are quite large variations (up to 10-fold differences in basin B) in failure prediction for relatively small changes (±0.5 kPa) in input values of C.


217

Fig. 1 The spatial distribution of FS simulated for the 1975 winter storm in basin B(a), at 0 h (before the storm); (b) at 22 h (at the end of the storm); (c) at 62 h (40 h after the storm).

218


Factor of Safety Factor of Safety Fig. 2 The effects of changes in soil cohesion on slope stability in basins A and B.

To estimate the effect of natural variability of soil depth within soil mapping units on FS, soil depth was increased and decreased by 30% with respect to the average soil depth assumed for each unit. Failure volumes for the 30% increase in soil depth increased by factors of 2.0 and 3.9 (compared with average depth) in basins A and B, respectively; for the 30% decrease, failure volumes declined by factors of 0.27 and 0.06 in these respective basins (Fig. 3). The greater response in FS observed in basin B is likely to be due to the steeper slopes. For the internal friction angle, the relative range of data (35.3°-41.4° ) is smaller than that of soil cohesion and the effect of this variation on FS is much smaller (Fig. 4). Using the maximum value of 41.4°, failure volumes change by factors of 0.32 for basin A and 0.35 for basin B; and 2.25 and 2.64, respectively, for the minimum value of 35.3°. The relative temporal and spatial response of FS in basin A was similar to that in basin B for similar changes in internal friction angle. It appears that the natural variability in internal friction angle within these basins would not produce large deviations from FS calculated based on average values. Changes in kS3t affected FS only slightly compared to other variables, possibly due to the relatively high ksa values of these forest soils. The timing of failure, however, is slightly affected: landslides occur earlier during the storm for larger values of £,.,.

Factor of Safety

Factor of Safety

Fig. 3 The effects of a 30% increase and decrease in soil depth on slope stability in basins A and B.


219

too 90

40-

iff iff nf iff

3 20-

o

100- • • • •

y ^—

i

n>

ra ? < ?

60-

basin A A standard . » max i.f.a (41.4°) -B- min i.f.a. (35.3°)

50-

20-

o

10-

i

—

i

2

—

i

—

i

—

-

i

// 4/

ft

B

30-

umu

6050-

nulative Area (%)

70-

90 80-

(%)

80-

/

J

T

basin B A standard -i» max i.f.a (41.4°) •B- min i.f.a. (35.3°)

!*/ / ?/y

ïï

f

0-

—

s?

3

Factor of Safety

Factor of Safety

Fig. 4 The effects of changes in internal friction angle on slope stability in basins A andB. 10090-

60bl> 40 30 20

8070-

basin A A standard • AC = 15kPa -B- AC = 10kPa

$

ut iff

•1

/// tT

4U30-

iff W

20101

2

3

4

1

5

7

1

/

^

E

8

basin B A standard » AC = 15kPa -E3- AC = 10kPa

Iff

f

0-

1

6

if/ iff W /i

•a. 50-

10 0-

S rea

70

Cumuls

80-

e Area (%)

90-

0

1

2

3

4

5

6

7

Factor of Safety Factor of Safety Fig. 5 The effects of ±2.5 kPa changes in maximum root cohesion on slope stability in basins A and B.

When the porosity changes from 0.37 (standard condition) to 0.30, failure volumes increase by 3.13- and 4.05-fold in basins A and B, respectively. For a porosity of 0.40, failure volumes in basins A and B decrease by factors of 0.66 and 0.61, respectively. However, the changes in the cumulative area-FS distributions are small. If the maximum root cohesion changes from 12.5 kPa to 10 kPa, failure volumes increase by 1.73 and 2.65-fold for basins A and B, respectively. For the case of AC=15 kPa, failure volumes decrease by factors of 0.55 and 0.29, respectively. Figure 5 shows the changes of FS distribution due to a 20% change in maximum root strength from the standard condition. The widest spread in FS values between the two AC extremes (10 and 15 kPa) occurs in the FS range from 2.0 to 3.5. Even though FS decreases in this range did not contribute to landsliding for this storm simulation, they could affect landsliding in larger storms. Figure 6 shows the effect of: (a) changing the root strength regrowth inflection point (r,) for Douglas-fir (18 to 25 years) and alder (15 to 20 years) and (b) changing the root decay curve constant (n) from 0.73 to 0.70. In the first case, failure volumes increased 1.56-fold (basin A) and 1.82-fold (basin B) compared with standard conditions. Thus, slower regrowth of regenerating trees can significantly increase landslide potential in steep forested terrain. In the second case (decreasing ri), failure

220


m m 807060-

basin A A standard • » root n = 0.7 -B- tj ( 25 & 20 yrs )

5040-

îulative Are a (%)

90-

30-

t:

20-

o

basin B A standard - » root n = 0.7 -B- t i ( 2 5 & 2 0 y r s )

f

102

3

4

5

6

7

00

1

2

3

4

5

6

7

Factor of Safety Factor of Safety Fig. 6 The effects of increases in the root strength regrowth inflection point (from 18 to 25 years for Douglas-fir and from 15 to 20 years Tor alder) and decreases in root decay curve constant: (n from 0.73 to 0.70) on slope stability in basins A and B (see Sidle, 1991 for description of constants).

volume was unchanged in basin A and decreased by a factor of 0.82 in basin B compared with standard conditions. When the threshold contributing area for propagation of a stream channel is reduced to 1000 m2, no failures occur in basin A and failure volume is drastically reduced (by a factor of 0.05) in basin B. By increasing the threshold contributing area to 5000 m2, failure volumes increase 3.12- and 1.38-fold in basins A and B, respectively. To determine the effect of different distributions of rainfall hyetographs, the sequence of intensity of the 1975 winter storm was rearranged with the duration and rainfall volume unchanged. If the hyetograph is reversed, failure volume increases by a factor of 1.13 in basin A and is unchanged in basin B. There is no change in failure volume if a normal distribution is used. For the case of 5% initial saturation (by depth) of the soil profile (0% for standard condition), failure volumes increase by 1.62- and 1.92-fold, respectively, for basins A and B. However, changes in FS distributions throughout other portions of the basins were small. Even though the stability simulations indicate that FS is largely controlled by topography, slope gradient, timber harvesting, and groundwater flow patterns, the influence of natural variabilities in soil depth, soil cohesion, and other soil properties are apparent.

REFERENCES Carrara, A., Cardinal, M., Detti, R., Guzzetti, F., Pasqui, V. & Reichenbach, P. (1991) GIS techniques and statistical models in evaluating landslide hazard. Earth Surf. Processes andLandforms 16, 427-445. Greswell, S., Heller, D. & Swanston, D. N. (1979) Mass movement response to forest management in the central Oregon Coast Range. USDA Forest Service Res. Bull. PNW-84. Pacific N.W. Res. Stn., Juneau, Alaska. Montgomery, D. R. & Dietrich, W. E. (1988) Where do channels begin? Nature 336(17), 232-234. Moore, I. D., O'Loughlin, E. M. & Burch, G. J. (1988) A contour based topographic model and its hydrologie and ecological applications. Earth Surf. Processes and Landforms 13, 305-320. Scheroeder, W. L. & Alto, J.V. (1983) Soil properties for slope stability analysis; Oregon and Washington coastal mountains. Forest Sci. 29, 823-833. Sidle, R. C. (1991) A conceptual model of changes in root cohesion in response to vegetation management. J. Environ. Qual. 20, 43-52.


221

Sidle, R. C. (1992) A theoretical model of the effects of timber harvesting on slope stability. Wat. Resour. Res. 28, 1897-1910. Ward, T. J., Li, R. & Simons, D. B. (1982) Mapping landslide hazards in forest watersheds. J. Geotech. Engng Div. ASCE 108(GT2), 319-324. Wu, W. (1993) Distributed slope stability analysis in steep forested basins. PhD Dissertation, Utah State University, Logan. Wu, T. H., McKinnel, W. P. & Swanston, D. N. (1979) Strength of tree roots and landslides on Prince of Wales Island, Alaska. Can. Geotech. J. 16, 19-33. Wu, W. & Sidle, R. C. (1995) A distributed slope stability model for steep forested basins. Wat. Resour. Res. 31, 20972110.

Human Impact on Erosion and Sedimentation (Proceedings of the Rabat Symposium, April 1997). IAHS Publ. no. 245, 1997

223

Application of a pedogeomorphic approach to sediment management strategies in the High Atlas mountains, southern Morocco J. P. NEWELL PRICE, H. R. FOX, H. M. MOORE Centre for Land Evaluation and Management, Division of Geography, School of Environmental and Applied Sciences, University of Derby, Kedleston Road, Derby DE22 1GB, UK

M. C. HARROUNI & A. EL ALAMI Institut Agronomique et Vétérinaire Hassan II, Complexe Horticole d'Agadir, BP 121, Ait Melloul, Morocco

Abstract In a study of the 97 km2 Askaouen sub-basin in the High Atlas Mountains, it is shown that a pedogeomorphic approach based on characteristic soil toposequences can provide a framework for understanding the spatial and temporal character of erosion processes. It provides information on the pedogeomorphic units which are dominant in determining sediment yields at each point in the basin and at a scale which is more suited to the formulation and implementation of soil conservation strategies. INTRODUCTION Since the development of the Politique des barrages programme in 1967, the sedimentation of reservoirs has been a priority environmental problem for the Moroccan government. While sediment yield figures for the High Atlas region are lower than those for northern Morocco (Heusch & Milliès-Lacroix, 1971; Probst & Amiotte Suchet, 1992) they are still a considerable cause for concern both in terms of the loss of reservoir capacity and reductions in land productivity. This paper focuses on the Aoulouz basin project in which a 103 million m3 reservoir was created at the mouth of a 4446 km2 drainage basin in the head waters of the Oued Souss in 1991 (see Fig. 1). Sedimentation rates in the Aoulouz reservoir have been estimated at 2 154 441 t year"1 by the Direction Régionale de l'hydraulique d'Agadir. However, less is known about the actual erosion rates occurring throughout the drainage basin itself. This paper concentrates on two sites within the 97 km2 Askaouen sub-basin (shown in Fig. 2) where sediment yields were monitored and discusses the value of a pedogeomorphic approach in the development of strategies for sediment management.

PEDOGEOMORPHOLOGY Pedogeomorphology is essentially the study of the interaction between pedological and geomorphic properties and processes. It is concerned with contemporary processes of soil and water movement relevant to soil erosion and the reflection of

224

/. P. Newell Price et al.

Fig. 1 Location of Aoulouz and Askaouen in southern Morocco.

these processes in features within the soil and at the surface (Conacher & Dalrymple, 1977). Through the establishment of characteristic pedogeomorphic unit sequences, at any particular point within a drainage basin, it is possible to assess the propensity of a land surface to both generate runoff and provide sediment to the water course at its base (Newell Price, 1996). It is also possible to establish relationships between certain lithologies and the typical catenary sequences that are produced from them in different stream order positions and under certain bioclimatic conditions (Watson, 1964; Mapa & Pathmarajah, 1995). Indeed, the true characteristics of the drainage basin can be defined by the sequences of land units that occur at various points within the catchment. However, in terms of runoff generation and erosion mechanisms it is not simply the systematic variation of soil properties or the sequence of land surface units that is important, but the relative dominance and function of certain units within the drainage basin. This paper illustrates how, within the Askaouen sub-basin, the effective control of such units is crucial to the overall success of soil conservation strategies. THE STUDY AREA Situated in a mountainous area, between 1600 and 2800 m, the Askaouen sub-basin has a sub-humid to semiarid climate with hot dry summers and cool humid winters. Mean annual precipitation between 1972 and 1994 was 281 mm (Berkaoui, personal communication). The eastern part of the sub-basin is dominated by the micaceous ash and trachytic tuffs of the Sirwa volcanic shield. To the west, a further complex of volcanic and volcano-sedimentary rocks gives way to large areas underlain by Askaouen granite.

Application of a pedogeomorphic approach to sediment management strategies

225

Fig. 2 Land surface types and monitoring sites within the Askaouen sub-basin.

Steeper slopes in the eastern half of the sub-basin, along with shallow and skeletal soils restrict land use to degraded rangeland. Soils throughout the sub-basin are characterised by a high stone content, thick surface crusts, low organic matter contents and large angular blocky structure. As a result, agriculture is concentrated in irrigated, terraced fields adjacent to the principal streams. Wheat and barley are intercropped with turnips, potatoes, broad beans, carrots, or alfalfa. Apple, walnut and almond trees are grown in plantations and on agroforestry plots with intercropped barley and vegetables. Barley is also sown on rain-fed land, most of which is found on the light textured soils and gentler slopes of the Askaouen granite.

METHODOLOGY This paper concentrates on two rangeland sites, Abane and Izguern, where contrasts in pedogeomorphic unit sequences have given rise to contrasting erosional response and, as a result, contrasting requirements in terms of soil conservation management. Abane is the site of a gully microcatchment on the weathered granite at 1950 m, while Izguern at 2200 m is a non-gullied hillslope underlain by andesitic tuff and andésite. The physical characteristics of the two sites are summarized in Table 1, while Figs 3 and 4 illustrate the three-dimensional arrangement of the various soil horizons at each site.

226

J. P. Newell Price et al.

Table 1 Land surface characteristics for the land surface units at Abane and Izguern. Land surface type Abane

Parent materials

Weathered granite Abane Weathered granite Izguern Andesitic tuff Izguern Andesitic tuff

Land surface Soil types unit 11.5° rilled upper slope 10.5° rilled lower slope 7.5° upper slope 9.5° lower slope

eutric eutric eutric eutric eutric eutric eutric eutric

anthrosols cambisols cambisols leptosols leptosols cambisols leptosols cambisols

Mean rock fragment cover (%) 37.2

Mean vegetation cover (%) 6.7

26.9

4.0

Mean rill density (rills lOOrn') 22 (1993-1994) 8(1994-1995) 8

77.0

6.0

16

75.5

5.4

18

Sediment yields were measured using Gerlach troughs. Troughs were arranged in pairs at the base of each principal land surface unit so as to be representative of sites of flow convergence. In addition, at Abane, at the mouth of the gully microcatchment, a granite and concrete dam was constructed on 4 December 1993, in order to trap sediment derived from the gully network and from the 5° to 19° slopes within the 4.43 ha micro-catchment. Measurement of the amount of sediment trapped by the dam permitted comparison of hillslope sediment yield with gully sediment yield.

eutric cambisols & oniric anthrosols

Ab2 trough site

©

A0 10YR4/3 brown BwCw

Êà

: XI .CO

3

'

J

/

CO

•a J oi

E

238

H. R. Fox et al.

Aoulouz basin was approximately 485 t km"2 year"1 and that over 46% of the sediment was produced by less than 25% of the catchment. Sub-basin 7, the Askaouen sub-basin (Fig. 2), was calculated to have the highest contribution to sedimentation at 7.15%, which corresponded to a specific yield of 1592 t km"2 year"1. Sub-basin 29 on the other hand contributed only 1.32% to reservoir sedimentation with a specific sediment yield of 232 t km"2 year"1. Despite the seemingly accurate estimates of sub-basin sediment yield, they are nevertheless predictions of potential sediment supply based upon simple descriptors of the main factors influencing the processes of erosion and sediment delivery. In an effort to evaluate predicted erosion values and to describe the pattern and process of erosion, field measurements of soil erosion were made in the Askaouen sub-basin over the period 1993 to 1995. The 97 km2 sub-basin was subdivided into a number of land surface types and pedogeomorphic toposequences using air photograph cover and field survey (Fig. 3). Gerlach troughs were used to monitor soil loss caused by surface wash and rill erosion in the land surface types and gully erosion was measured with a check dam. The results of these measurements confirm the importance of gully erosion in the sediment production system. For the season 1994-1995, for which the most complete records are available, mean annual sediment yields for rill erosion and slopewash ranged from 9.0 t km"2 year"1 to 16.3 km"2 year"' for the lower slope units of five different land surface types (Table 2). This compares with a sediment yield in excess of 1030 t km"2 year"1 produced by a single rainfall event of 36 mm for a monitored gully system. Analysis of the frequency of such storm events for the sub-basin suggests that the annual sediment yield from gully affected land may be over 2050 t km"2 year"1. This is a figure comfortably in excess of the predicted potential yields of 1592 t km"2 for the sub-basin. The two orders of magnitude difference between rill erosion and gully sediment yield corroborates the findings of Heusch (1970) in the Rif Mountains of Morocco and the study by Roose (1991) in northern Algeria. Gully texture in the five monitored land surface types ranged from 2 to 5 channels per km. Flow in the main badland area at Abane (Fig. 3) was activated by precipitation events in excess of 16 mm and occurred in seven out of 16 storm events during the monitoring period 1993-1995. Strong links between lower slope units of adjacent catenary sequences mean that sediment delivery to gully channels is very high and hence overall sediment yield from this gullied area is also high. In terms of reservoir sedimentation in the Aoulouz reservoir and the contribution of the Askaouen sub-basin, the Abane badland area probably accounts for a significant proportion of the total sediment yield. This is not only because it is an area of high sediment production, but also because it is close to the sub-basin outlet in the Oued Tizguiy (Fig. 3). Other land surface types in other parts of the sub-basin have lower gully textures, weaker slope channel connections and are at greater distances from the basin outlet giving greater potential for deposition and storage within the sub-basin. Consequently any efforts to reduce sedimentation rates through control of sediment supply from the Askaouen sub-basin should be focused upon the Abane gully system. Many workers have drawn attention to the importance of lithology in determining erosion rates. Probst & Amiotte Suchet (1992) and Lahlou (1993) produce multiple regression relationships describing the significance of varying

Soil erosion and reservoir sedimentation in the High Atlas Mountains

239

Table 2 Sediment yields from different land surface types (lower slope units) Askaouen sub-basin. Land surface type

Lithology

Abane Aoulouz

Granite ([weathered) Granite

Sediment yield 1994-1995 (t km2) 16.30 12.25

Imi-nTagdalt

Granite

Mean rill density (rills 100 iff2) 8 10

Mean rill cross section area (cm"2) 625

Slope (°) 10.5

140

10.3

470

4.5

13.47

9

Tastouite

Volcano-sedimentary siltstone and schist

15.23

44

70

5.0

Izguern

Andesitic tuff

13.97

18

135

9.5

lithology to rates of sediment yield. However such relationships are not simple and as noted by Gerrard (1989) and Fox & Moore (1993) similar lithologies may have quite different susceptibilities to erosion due to their physical and chemical make-up and past denudational history. Hence whilst igneous rocks are generally regarded as being amongst the least erodible lithologies, in the case of the Askaouen sub-basin erosion rates are high. The sub-basin is underlain principally by granites and volcanic lithologies, but the erodibility of these materials is determined by the degree of weathering to which they have been subjected. Where weathering depths reach several metres, as at Abane, gully systems have developed and as noted rates of sediment yield are at minimum in excess of 1000 t km"2 year"1. Where relatively unweathered rock outcrops at the surface, gully development is restricted and consequently sediment yields are lower. Rill erosion rates were also higher on the weathered granite areas of Abane in comparison to the lower pedogeomorphic units of the relatively unweathered granite catenary sequences (Table 2). Rill densities were in fact lowest on the Abane weathered granite, but the rills developed were deeply incised into the colluvial footslope. Much smaller rills occurred on the other lithologies. Erosion on the less weathered granite areas was similar to that on the tuffaceous volcanic rocks and siltstone schist, volcanosedimentary formation of the other land surface types.

Acknowledgements The authors thank the officers of the Département des Eaux et Forêt et de la Conservation des Sols at the Ministry of Agriculture in Rabat and Agadir and the officials and people of Askaouen for their help in this study.

REFERENCES Arnoldus, H. M. J. (1977) Methodology used to determine the maximum potential average annual soil loss due to sheet and rill erosion in Morocco. FAO Soils Bull. 34, 39-48. El Hebil (1970) Etude hydrogèologique du haut bassin versant du Souss. Thèse Univ. Montpellier -Rapp. Inéd. MTPC/DH/DRE. El Kasri, M. (1995) Project d'Aménagements du Bassin Versant du Barrage d'Aoulouz. Ministère de l'Agriculture et de la Mise en Valeur Agricole. Mai 1995 (unpublished). Fox, H. R. & Moore, H. M. (1993) Discussion: fluvial suspended sediment transport and mechanical erosion in the Maghreb (North Africa). Hydrol. Sci. J. 38(6), 1-4. Gerrard, A. J. (1989) The nature of slope materials on the Dartmoor granite. Z. Geomorphol. N.F. 33, 179-188.

240

H. R. Fox et al.

Heusch, B. (1970) L'érosion du Pré-Rit'-Une étude quantitative de l'érosion hydraulique dans les collines marneuses du Pré-Rif occidental. Ann. Recherches Forestières de Maroc 12, 9-176. Heusch, B. & Millies-Lacroix, A. (1971) Une méthode pour estimer l'écoulement et l'érosion dans un bassin. Mines & Géologie 33, 21-39. Lahlou, A. (1993) Envasement des Barrages au Maroc. Wallada, Casablanca. Probst, J. L. & Amiotte Suchet, P. (1992) Fluvial suspended sediment transport and mechanical erosion in the Maghreb (North Africa) Hydrol. Sci. J. 37(6), 621-637. Rippey, B. (1982) Sedimentary record of rainfall variations in a sub-humid lake. Nature, Land. 296, 434-6. Roose, E. J. (1991) Conservation des sols en zones méditerranéenes. Synthèse et proposition d'une nouvelle stratégie de lutte antiérosive: la GCES. Call. ORSTOM sér. Pédol. 26, 45-181. Snoussi, M. (1988) Nature, estimation et comparison des flux de matières issus des bassins versants de l'Adour (France), du Sebou, de L'oum-er-Rbia et du Souss (Maroc). Mémoire de l'Institut de Géologie du Bassin Aquitaine, no. 22, Université de Bordeaux, France.

Human Impact on Erosion and Sedimentation (Proceedings of Rabat Symposium S6, April 1997). 1AHS Publ. no. 245, 1997

241

Sediment deposition in the Barasona reservoir (central Pyrenees, Spain): temporal and spatial variability of sediment yield and land use impacts BLAS VALERO-GARCÉS, ANA NAVAS & JAVIER MACHIN Estaciïm Experimental de Aula Dei, CSIC, Apdo 202, 50080 Zaragoza, Spain

Abstract A sedimentological study of the Barasona reservoir (central Pyrenees, Spain) illustrates the variability of depositional dynamics in the Esera-Isâbena drainage basin through time and space. A preliminary chronology based on flood events recorded in the reservoir indicates changes in sediment yield during the last four decades. Coupled reservoir-drainage basin sediment studies have enabled sediment sources and sediment production risk areas to be identified and, the assessment of the relative role of natural (topography, drainage network, lithology, and hydrology) and human (land use) factors influencing sediment yield, delivery and deposition.

INTRODUCTION Soil erosion processes are an important influence on sediment delivery to reservoirs and consequent loss of water storage capacity. Lake and reservoir sediments have been frequently used to reconstruct records of catchment processes (Dearing & Foster, 1993) and the quantification of sediment accumulation in reservoirs has proved to be a reliable means of estimating average erosion rates in their drainage basins. In this paper we present the preliminary results of a sedimentological study of a mountain drainage basin and reservoir in the Spanish Pyrenees. Using these methods we have improved the resolution of the reconstructed depositional history of the reservoir, and contributed to a better understanding of temporal and spatial variability in catchment processes, and the relative importance of anthropogenic and natural factors.

THE ESERA-ISABENA RIVER BASIN AND THE BARASONA RESERVOIR In the Spanish Pyrenees, many rivers that run through rugged topography and erodible soils and rock formations have been dammed in the foothills. The Barasona reservoir in the Esera-Isâbena drainage basin, central Pyrenees (Fig. 1A) provides a case study for an integrated investigation of reservoir drainage basin dynamics. The basin is characterized by heterogeneous topography and lithology and contains several WNW-ESE trending geologic units (Fig. IB, C). The climate is of a mountain type, and is cold and wet with both Atlantic and Mediterranean influences, and strong north-south gradients. Annual precipitation and average temperature range from more than 2000 mm year"1 and 4°C respectively in the headwaters, to less than 500 mm year and 12°C at the reservoir. Both the Esera and Isâbena rivers have

Bias Valero-Garcés et al.

242

E E

E 300

o

O ra

CM

A

A

A

A

A O

A

o O O

i\ m

>CD UJ

P

ogene ogene one (Creta aceous one (Juras

nean Rang

to tu

m m ÛL n

S E ? to

O

>. D) O O

s:

E

tO

-i>

•

o

"O

E S?

o

m ') °l — -S mF "o "O m

t

CO

S?

E

o

c

to m w

o

•s

o

CD

co O w O O

âbena

_i

1S

c -Sî*

-r m

ate

ë — J2 == J3 Io=(0

E o

S

Eser

co m

z/

n Q

V Mino

X.

\ Jucar( o

t>

c o •*-» ce u o

Ç0

Q)(\

00 ] V 00 *)

rth

n m

( /-

Uc

qu iv

c _to U)

\

o m \t-

• -

to -i

n

(

1

CO T3

S.

3

S 0

|3

Uo 1 \

O

O to O r\l M t_>

O

o O

F o o

A Ci) A O O CO

A O

o o

E 009

F ni F

1600

o o o CD o CM

-:

V

Sediment deposition in the Barasona reservoir (central Pyrenees, Spain)

243

transitional hydrologie regimes characterized by two periods of high flows, during late spring/early summer (snow melt) and late fall (Mediterranean rains). According to available historical and hydrological evidence there have been eight major floods in the Esera-Isâbena basin since 1892. The floods are caused by three different mechanisms i.e. late spring/early summer snow melt and heavy rains (1925, 1960, 1971, 1977), summer thunderstorms (1963), and late autumn heavy rains (1960, 1963, 1965, 1977, 1982, 1984, 1996). The reservoir, one of the oldest in Spain (1932), has lost one third of its initial water capacity (71 x 106 m3). The specific sediment yield of the 1224 km2 basin has been estimated as 350 t km2 year1 (Sanz Montera et al., 1996). Variations in drainage network pattern (Fig. 1C) and lithology (Fig. IB) have been proposed as key factors in explaining intrabasin contrasts in sediment yield (Fargas et al., 1996). In this paper we test this hypothesis and also the relationship between land use and sediment delivery.

METHODS AND DATA Twenty two cores and sediment sections, up to 4 m long, were described and correlated using lithology, mineralogy and sedimentary structures (Fig. 2A and C). Suspended sediment and river bed sediments were sampled after the winter 1996 floods (28 February 1996) in the Esera and Isâbena rivers. Samples of 1 litre were collected at approximately 10 cm from the stream bed, and filtered in the laboratory to calculate suspended sediment concentration (SSC). Mineralogy was determined by a Siemens D-500 diffractometer. Composition percentages were calculated using relative reflectance factors and are therefore semiquantitative. Flow values for the Isâbena and Esera rivers were provided by the Confederation Hidrogrâfica del Ebro, and land use maps were based on data derived from Manrique et al. (1987).

RESULTS AND DISCUSSION Sedimentation in the Barasona reservoir The two main depositional environments in the Barasona reservoir are firstly, deltas developed at the mouth of the Esera River and close to the dam wall, and secondly, pelagic plains (Valero-Garcés et al., 1996). Sediment thickness varies from 2 m in the littoral areas, to more than 4 m in the Esera delta, 6 m on the pelagic plain, and a few tens of metres in the dam wall delta. Sediments are admixtures of calcite, quartz, illite, chlorite and minor amounts of kaolinite, feldspar, dolomite, and pyrophyllite. The organic matter content ranges between 2-5%. The silt fraction is dominant, but sand layers also occur, specially close to the mouth of the Esera River. Sediments are primarily of allochthonous origin, and deposition is dominated by physical processes. Changes in the river flows exert a major control on the depositional dynamics. Upward fining sequences, up to 30 cm thick and composed of sand, silt, and clay define major flood episodes in the reservoir. At the Esera River mouth (profile B-21, Fig. 2C) more than 50% of the sediments are sandy silts and

244


250 m SSCsj Gravel, sands and silts (Fluvial) Fig. 2 Sediment core correlation and a tentative chronology for deposition in the Barasona reservoir. A. Location of the sediment cores. B. Maximum instantaneous flow values in the Esera and Isâbena rivers during each year at the gauging station closest to the reservoir. C. Selected sediment cores and a tentative correlation between sand layers and flood events.

sands accumulated during floods. The occurrence of sand layers in cores from the central areas of the reservoir (Fig. 2C) indicates that tractive processes still function in that area. In contrast to other reservoirs, where authigenic processes are very significant (Cobo Rayân et ai, 1996), our data support a depositional model with major flood events as the main agents of reservoir sedimentation. A relative depositional chronology can be constructed by comparing the known floods in the basin (Fig. 2B) with the sandy layers described in core B-21 (Fig. 2C).


245

246


We have ascribed the sand layer at the bottom of the profile, characterized by the coarser grain size, higher quartz content, and presence of the largest tree branches, to the largest flood (1960). According to this relative chronology, three periods can be distinguished in the Esera delta of the Barasona reservoir. Sedimentation rate was high during the early 1960s (more than 1.5 m of sediment was deposited in less than 10 years), decreased during the 1970s (up to 1 m in about 10 years) and increased subsequently during the 1980s and 1990s (about 1.5 m in 15 years), but without reaching the values of the 1960s. Taking into account that most of the sediment is deposited during floods and that these events were more frequent during the 1960s, we speculate that these changes in sediment yield have been caused by changes in flood frequency. A definitive chronology will be established using radioisotope analyses, which are currently in progress. The winter 1996 floods The floods of 1995-1996 occurred during a period of increased precipitation extending from late fall to winter (December 1995-February 1996). A preliminary sampling programme was undertaken at the end of that flood period. Our results, although not based on a comprehensive survey, clearly indicate that the Esera River delivers most of the suspended sediment to the Barasona reservoir (Fig. 3A). Suspended sediment concentrations in the inflow to the reservoir were 50 times higher in the Esera River (224 mg l"1) than in the Isâbena River (3.9 mg l"1). The mineralogy of the river sediments reflects not only their grain size distribution but also the heterogeneous lithology of the source areas (Fig. IB and Fig. 3) Clay minerals are more abundant in the Isâbena River suspended sediment due to the lower flow of this river. Quartz is generally dominant over calcite on the Esera River, whereas the opposite is true for the Isâbena River. Feldspars are more abundant in sediment from the Esera River, especially in the upper valley where granitic rocks outcrop. Sediment yields as inferred from SSC values and river sediment mineralogy (Fig. 3A and B) show a distinctive geographic distribution. We define four areas in the Esera basin (the headwaters, the Castejôn de Sos Depression, the Campo-Seira intramountain Depressions, and the Intermediate Depression), and three in the Isâbena basin: northern, intermediate, and southern. The Esera River basin Suspended sediment concentrations in the headwaters of the Esera River were below detection limits but increased to 6 mg l"1 downstream. A heavy snowpack covered the landscape and effectively reduced soil erosion. Pyrophyllite appears in all samples downstream of where tributaries draining sedimentary Palaeozoic rocks join the Esera River. The Lower Devonian low grade metamorphic shales are the only pyrophyllite-rich rocks outcropping in the basin (up to 25% of total clay minerals; Valero-Garcés et al., 1996), and consequently, we identified them as the only source area for this mineral. One of the samples is unusual in terms of both sediment concentration (2468 mg l"1) and mineral composition (higher amounts of carbonate and quartz), and is more similar to the composition of silt from the Esera River bed (Fig. 3B). Because the sample was


247

taken in an area where the river is very shallow, we explain this anomaly as the result of a higher proportion of coarser particles transported by saltation and as bed load. We interpret the low SSC in the northern Esera basin as being the result of lower erosion rates. The snow covered landscape would be a significant factor in limiting erosion during the winter months. Land use would also favour low sediment yields (Fig. 4). Most of the land is grassland (>50%), which helps to prevent soil erosion throughout the year (Ruiz Flano, 1993), and the percentage of cultivated land is small (< 10 %) (Manrique et al., 1987). The mineral composition of the suspended sediment changes markedly downstream, after the river enters the Castejôn de Sos sub-basin. The suspended sediment concentration is very high (1276 mg l"1), and its quartz and calcite content increases up to 20% for each. A higher sediment yield would be favoured by the increase in cultivated land (up to 50%), and the decrease in grassland (