Forest Hydrology and Watershed Management - Hydrologie Forestiere et Amenagement des Bassins Hydrologiques (Proceedings of the Vancouver Symposium, August 1987; Actes du Co11oque de Vancouver, Aout 1987):IAHS-AISH Publ.no.167,1987. Deforestation and erosion in the Nepalese Himalaya - is the link myth or reality? W. J. H. RAMSAY Land Capability Consultants Ltd., Times House, Willingham, Cambridge CB4 5LH, UK ABSTRACT Despite numerous popular accounts, data on erosion processes in the Himalaya are scarce and unre- liable. Quantitative studies of mass wasting and surface erosion in Nepal are reviewed. The literature reveals a consensus that in Nepal erosion rates are naturally very high, mass wasting is the dominant hillslope process, and geological factors are the most important determinants of slope stability. Deforestation is linked to surface erosion, gullying and shallow (~3 m deep) landslides. Locally this soil loss is important agriculturally. Currently there is no evidence to link deforestation to the large, complex slope failures which mobilize most sediment. Deboisement et erosion dans les Himalayas du Nepal: lien fictif au reel? RESUME En depit de nombreux recits sur Ie sujet, les donnees sur les processus erosifs dans les Himalayas sont rares et peu fiables. Les etudes quantitatives sur les mouvements de masse et l'erosion de surface au Nepal sont passees en revue. Ces etudes s'accordent sur Ie fait que Ie processus des mouvements de masse domine la forte erosion naturelle du Nepal, et que la stabilite des pentes est determinee par des facteurs geologiques. Le deboisement est relie a l'erosion de surface, au ravinage et a des glissements de terrain peu profonds (~3 m). Ces pertes de sol sont importantes localement du point de vue agricole. II n'y a actuellement aucune evidence reliant Ie deboisement aux destabilisations complexes de pentes qui fournissent la masse des sediments. INTRODUCTION Deforestation and erosion are widespread in the Himalaya. This environmental deterioration is widely seen as the principal cause of severe flooding and sedimentation downstream in the Gangetic Plain. The perceived cause and effect relationship between upstream ac- tivities and downstream damages has had major impacts on both the decision-making process and on development-project design (lves & Messerli, 1984). However, this relationship is an assumption based on extremely poor data, and it may well reflect political and in- 239 240 W.J.H.Ramsay stitutional influences rather than prevailing bio-physical processes (Thompson & Warburton, 1985). The Himalaya-Ganges-Brahmaputra system is one of the world's largest highland-lowland interactive systems (Ives & Messerli, 1984), and any analysis of environmental problems in the region must recognize both the dynamic nature of the geomorphological environ- ment and the extreme heterogeneity of the area. Periods of record of geophysical events in the Himalaya are short. Statistics derived from such records may yield biased results due to intermittency, i.e. the variable geomorphological response to forcing events (Church, 1980). To avoid the hazard of generalizing locally-derived data over wider physical and social environments, what data there are should be treated with caution, and original sources referred to in order to clarify the context and methodology of the original study. This paper reviews existing studies on current slope processes in Nepal, and discusses these in relation to the effects of deforesta- tion. EROSION IN NEPAL Mass wasting Despite a wealth of anecdotal information, quantitative studies on current rates of mass wasting in Nepal are scarce. Bansode & Pradhan (1975) carried out a reconnaissance survey of landslides along part of the channels of the Sun Kosi and Tamur rivers above Tribeni in 1963. Their impression was of high levels of mass move- ment activity contributing to high sediment loads in the rivers, "mostly due to heavy precipitation, deep weathering, steep dip- slopes of the valley walls, under-cutting of the banks due to high velocity of these rivers, unstable nature of the rocks due to their structural disposition, high seismicity of the area, unplanned deforestation, etc." (Bansode & Pradhan, 1975; p.253). Failure surfaces were 30-70° (n = 19). Prasad (1975) reported on 10 years' observations of seismicity, rainfall and landslide occurrence in the Durbasha watershed near Chatra in eastern Nepal. Overall, landslide incidence corresponded with high levels of both precipitation and seismic activity in July and August.l However, slides also occurred in years of low earth- quake activity, and so Prasad concluded that "seismic shocks by themselves are not the main cause of occurrence of landslides in regions away from the epicentre" (Prasad, 1975; p.79). Hydrological conditions, i.e. high groundwater levels and intense precipitation, were considered to be more important. Williams (1977), investigating the east to west shift of the Kosi river, used satellite imagery to identify all slides larger than approximately 20 ha in the catchment of the Sapta Kosi in eastern lThe author does not comment on the apparent seasonality of earthquakes. Deforestation and erosion in the Himalaya 241 Nepal, correlated these with data from 1:63,360 topographic maps available for a small part of the basin, assumed a failure scar recovery period of 50 years, and used Simonett's (1967) empirical volume/area relationship derived for slides in New Guinea to give a "total slide volume" for the Sapta Kosi basin over 50 years of 0.91 x 109 m3. He estimated that these "large landslides" contributed 31% of the sediment load of the Sapta Kosi, with a further 64% coming from small slides, surface erosion and gullying. The assump- tions concerning basin homogeneity and failure age, area, and volume required for this procedure leave the accuracy of Williams' figures open to question. Laban (1979) carried out a reconnaissance slide intensity survey of the whole of Nepal, expressing his data in terms of number of failures per linear km seen from one side of a light aircraft. He attributed 5% of all slides observed to road and trail construction. Wagner (1981, 1983) used a statistical analysis of geological and other characteristics of 100 landslides, mainly along roads in the Middle Mountains, to develop a site-specific landslide hazard as- sessment and mapping methodology. He concluded that geological factors were of overriding importance in determining debris and rock-slide hazard. Brunsden et al. (1981) made the observation that in the Low Himalaya of eastern Nepal mass movement phenomena were concentrated in two locations: low level undercut situations such as ravines and the outside of meander bends, and areas of structural discontinuity, suggesting an important role for "intensely shattered rock and preferred water movements". Modal angles for debris slides were 35- 43°, and they tentatively identified a slope angle of 30° as a lower limit for first-time shallow debris slides. "Mudslides" (elongate or lobate masses of weathered debris which move in well defined tracks bounded by steeply inclined lateral and basal shear surfaces (Brunsden et al., 1981)) occurred on slopes of 25-29°. They also described "mass movement catchments", steep rapidly eroding channels with active, expanding heads supplying material to the channels by debris slides, debris flows, rock-debris chutes, and gullies. Severe gullying in areas underlain by gneiss, which weathered to depths of 20 m, was attributed to man-accelerated soil erosion as a result of cultivation following forest clearance. Caine & Mool (1982) investigated mass movements in the Kathmandu- Kakani area as part of the United Nations University Mountain Hazard Mapping Project (see Ives & Messerli, 1981). Their analysis of slide morphometry and slope material properties led them to em- phasize the importance of material controls on the landslides in their study area, particularly the brittle behaviour of the weather- ed, untransported bedrock. The high incidence of catastrophic landsliding was further explained by relief, seasonally high water tables, and recent deforestation. Rainfall was thought to be of comparatively minor importance. They gave an estimated rate of surface lowering by landsliding of 12 mm year-l. Matsuura (1985) studied landslides in a small catchment northwest of Pokhara and found a close relationship between mineralogy, weath- ering, and the development of slip surfaces. Ramsay (1985, 1986, 1987) carried out a reconnaissance survey of sediment production and transfer mechanisms in the 122 km2 Phewa 242 W.J.H.Ramsay Valley in the Middle Mountains of Nepal and identified a variety of mass movement processes. The commonest events were shallow transla- tional failures on slopes of, typically, 36-45°, with volumes <1 x 103 m3, and with slope revegetation taking less than ten years. Larger slides occurred on slopes oversteepened by fluvial action. Flows developed in areas of weak rock and unfavorable structure, and were associated with groundwater discharge. Flow velocities ac- celerated during the monsoon. The highly fractured and deeply weathered zones around faults were the sites of mass movement catch- ments, and these complex failures were responsible for approximately 90% of all sediment production by mass wasting in the basin. A first estimate of surface lowering by mass movement processes in the Phewa Valley was 2-3 mm year-i. Locally, surface erosion on severe- ly overgrazed pasture may be 5-6 mm year-i. No data were available