GISdevelopment.net ---> Application ---> Natural Resource Management


The notion that intensification of land use (e.g. deforestation, urbanization) in mountain environments is rampant and leads to regional-scale degradation of the land and increased flooding of lowland areas has been the focus of many studies (e.g. Eckholm, 1975; Ives, 1987; Hofer, 1993; Ives and Messerli 1993; Lauterburg, 1993). Population growth due to high birthrates, in-migration, and/or improved access to healthcare facilities often necessitate the development of infrastructure such as roads, trails, and buildings and an intensification of agricultural/horticultural, timber extraction (for fuel and building material), and grazing activity. The resulting impacts are purported to have increased the risks associated with landslides, floods, and other erosion processes. Mass wasting, flooding, and increased sedimentation are often seen as direct consequences of the over-population and poor land management practices of the hill-people (Gupta, 1990).

A number of authors (e.g. Hofer, 1993; Lauterburg, 1993) have identified the need for more reliable empirical data and research in order to properly evaluate the role of land use change and increased population on the activity of erosion processes in the Himalaya. Ives and Messerli (1993) warn against extrapolation of local results to the entire Himalayan region and Lauterburg (1993) concludes that the impact of human activity on the landscape in terms of erosion processes is heavily scale-dependent. The lack of reliable data is demonstrated when comparing the results of research carried out by Kuster (1993), who concludes that there has been an increase in the forests of Himachal Pradesh, India and Gupta (1990) who warns of the impending total deforestation of the very same area. There is little doubt that the consequences of increased land pressures are having a negative impact on parts of the Himalaya but care must be taken not to generalize this to the entire region. Micro- and meso-scale differences in climate, geology, vegetation, and human land use all play a role in determining the extent to which human activity affects natural erosion processes.

The Kullu District in the state of Himachal Pradesh, where this research project is focused, has undergone significant land use/cover change and intensification. The Kullu District, like all mountainous areas, is subject to natural processes of erosion such as mass wasting, flooding, and avalanche activity. The extent of economic and human losses due to flooding and mass wasting activity, especially during the previous two decades, is alarming and provides the stimulus for this research. A single flood event in the Kullu Valley in 1993 caused an estimated US$18 million in damages and claimed the lives of 6 people (Sah and Mazari, 1998). A flood of similar magnitude in 1905 caused similar geomorphic damages, but had virtually no direct impact on people or infrastructure. The unchecked expansion of built-up areas into geomorphologically active regions such as the river flood plain, unstable slope deposits, and debris flow fans has been accelerating and is a major cause for concern. The consequences associated with rapid land use change due to uncontrolled development in mountain areas has been studied by Gardner et al. (1997), Singh and Pandey (1996), Gupta (1997), Rao (1997), and Yudhbir (1997). Kuster (1993) and Gardner et al. (1997) assessed the effects of deforestation and urbanization on the state of natural hazards in this area and concluded that:

There is relatively little evidence (to suggest) that human activities have materially altered the frequency, magnitude, and location of hazardous processes so as to increase risk except in a few, very localized situations. The primary causative factors in the increased risk are growth in tourist demand and intensification and diversification of commercial agriculture. Gardner et al. (1997, p.251)

Thus the research question central to this project is to determine the validity of these claims and identify the main driving forces behind the dramatic increase in natural hazard activity. In addition the paper aims to explore the interactions between and consequences of land use intensification, environmental degradation, and activity of erosion processes. The methods employed and developed in this study area are intended to be globally applicable.

Study Area
The Kullu District located in state of Himachal Pradesh, India is in the transition zone between the lesser Himalaya to the south and the greater Himalaya to the north. Elevation of the valley floor rises northward from ~1,100m near Bhuntar to 4,000m at Rohtang Pass with an average relief of ~2,000m (Fig. 1). The temperate climate is characterized by dry winters and hot, rainy summers. Average annual precipitation amounts of more than 4,000mm in the valley are due to the monsoon, which brings persistent rains beginning in June and lasting until late September. Temperatures range from a mean daily minimum of between -15°C and 0°C in January to maximums of 20°C to 30°C in June. Topography plays a key role in modifying the local climate in terms of lower temperatures and higher precipitation amounts on north/northeast-facing slopes.

Geologically the area is composed of predominantly Precambrian phyllites, schists and gniesses, and granite of the Jutogh and Chail Formations (Mehta, 1976). Southern areas of the valley are composed of heavily folded and faulted sedimentary rocks including shales, quartzites, dolomites, and sandstones. Earthquakes continue to play a significant role in reshaping the valley, which falls into the highest category of earthquake risk (zone V). Intense physical weathering is the primary agent of erosion with chemical weathering playing a minor role in the lower elevations. Surficial materials are composed mainly of colluvium, alluvium, and morainal deposits. Soils are of the sub-montane podzolic type with average depths of 50 to 100cm (Singh, 1992). It can be assumed that the geology of the Kullu Valley has not changed appreciable over the relatively short time frame of this research.

Present surface cover consists mainly of forests which cover ~38% of the total land area of the state, agricultural fields account for ~11%, orchards and grazing land each occupy ~10%, 5% of the land is abandoned, and the remaining 25% is regarded as unfit for vegetation (i.e. snow, ice, and rock). The most productive agricultural fields are located on the relatively flat valley bottom and on fans where nutrient-rich alluvial soils are most abundant. Up slope from these are terraces where maize, wheat, barley, and rice are grown. Mixed-in with these fields are orchard terraces, which are typically situated on previously abandoned or less productive soils on the lower to mid-slopes. Grazing and forested areas are typically found above the agricultural fields and orchards, and in some cases extend all the way to ridgeline. 

The Kullu District has been settled for several millennia with relatively long periods of social and political stability (Gardner et al., 1997). However parts of the area have undergone noticeable land use change over the past century, particularly since about 1970. Land use change is usually brought about by a change in the socioeconomic situation of an area. Agriculture has been the dominant economic and land use activity in the region for centuries, at times employing ~90% of the population (Singh, 1992). The construction of National Highway 21 in the 1950's linked this remote region with centers to the south. This increase in transportation capacity led to a shift from subsistence crop agriculture to commercial horticultural orchards. Cultivated and abandoned fields were converted to stands of various types of fruit trees, and orchardry became an important economic as well as land use activity. Conversion of land to orchards continued until a second major land use change occurred during the 1980's when the district experienced a large influx of tourists avoiding the unstable political situation in the Kashmir region. This led to the construction of hotels, restaurants, roads, and development of other tourist-related facilities. By the early 1990's, agricultural land was being developed or abandoned in favor of tourism-related opportunities. The region has recently (past decade) become popular with foreign tourists as well, further promoting development of urban infrastructure.

The increasing population, rapidly expanding infrastructure, and continued intensification of land use are superimposed onto a physically unstable landscape characterized by frequent mass wasting and erosion, activity and floods. Risk to life and property is a consequence of natural hazards, which result from this juxtaposition of the human and physical environments. Disasters are the realization of risk and often occur at specific sites (i.e. hazard sites). Methods used to locate and identify hazard sites are discussed below.


Hazard site identification and mapping
Current and potential hazard sites were located and identified through fieldwork because in most cases maps, satellite imagery, and air photos are not accurate enough to yield sufficiently detailed information on the activity of various processes, their location, and proximity to settlements or other infrastructure. Hazard maps do not exist or are unreliable for many mountainous areas. Although maps, air photos, and historical information should always be evaluated and used when available the main source of information should come from work carried out on the ground.

A hazard site is defined as the area of release/failure that can potentially comes into contact with settlements and/or infrastructure. In many cases the initial point of contact will be a road, trail, railroad, or other linear feature which cuts across the path of an erosion process. Examples of geomorphic processes examined in the field include landslides, composite and progressive failures, and rockfall. All accessible roads and settlements in the study area were surveyed in order to collect information on all current and potential hazard sites. Thus the dataset represents the entire population of hazard sites in the Valley. The severity of potential damages in terms of human and economic losses at each hazard site was not addressed due to a lack of reliable socioeconomic data and only the possibility of occurrence was noted.

Geomorphic evidence was used to identify the type of active process operating at each site. Geomorphic characteristics examined in the field included, slope, aspect, type of surficial materials, slope surface morphology, and microrelief (Table 1). Aspect and slope were measured using a compass and inclinometer respectively. Unconsolidated, poorly sorted materials such as colluvium usually indicate unstable slopes especially in sparsely vegetated areas. Highly jointed, steep, and exposed rock faces indicate areas of rockfall activity, particularly in locations where fresh deposits are frequently found near a cliff base. In many instances fresh deposits at the base of a slope or cliff following a heavy rainfall or other destabilizing event (e.g. earthquake) provided clear evidence of the type and magnitude of process operating at that location. Other areas with similar morphological and vegetative characteristics were investigated in order to assess their potential for failure. Mitigative structures such as retaining walls, gabion baskets, deflection bars, and channel baffles are another source of evidence that a certain type of erosion process has recurred there with some frequency in the past. 

The geographical location of each known and potential hazard site was recorded using a GPS unit and plotted onto a field map. The GPS locations were subsequently plotted onto a base map showing roads, hydrological features, and settlements to produce a map of hazard sites in the valley. Field observations and measurements recorded at each site such as those described above (e.g. slope, aspect, vegetation type, etc.) were then entered into a database (Table 2) and linked with the GIS hazards layer.


Table 1. Identification criteria for each process.

	Landslide	Prog./Comp. Failure	Rock fall
Deposits	x	x	x
Landform	x	x	x
Mitigative structure	x	x
Historical information	x	x

Table 2. Specific variables/attributes collected for each type of erosion processes.  

Layer Hazard		Polygons
Hazard Type	Landslide	Rock fall	Prog./Comp. Failure
Site ID	x	x	x
Northing	x	x	x
Easting	x	x	x
Elevation	x	x	x
Activity	x	x	x
Angle	x	x	x
Aspect	x	x	x
Pic.#	x	x	x
Soils	x		x
Height	x	x	x
Width	x	x	x
Structure		x
Boulder		x
Angularity		x
Vegetation	x		x
Land Use	x	x	x
Water	x		x

As discussed above, each hazard site represents the area of potential release. However, the mobilization of an erosion process will affect a much larger area down slope of the release area known as the deposition zone. To account for this, a procedure capable of approximating the deposition zone for each release area was developed. The procedure is essentially an anisotropic disperse operation that uses slope, aspect, elevation, and distance information from field measurements, a DEM, and historical records as input parameters and generates a map showing the potential run-out/deposition area. First, the 52 sites were grouped according to average aspect and plotted on separate layers. Next, two friction surface were generated and combined. One is based on a range of run-out distances derived from field measurements and literature and the other a gravity potential image. To produce the latter friction surface a slope image was recoded using a regression formula which relates slope steepness to gravitational potential. The disperse operation calculates a cost surface from each site based on the friction surface and an image defining the direction in which the dispersing force is acting. The output essentially represents the potential run-out or deposition zone for each site (Figure 2).



The magnitude, indeed the activity of all erosion processes, is controlled to varying degrees by geologic, climatic, and human factors. Geologic factors are assumed to have stayed constant over the past 27 years. Climate and certainly human population factors may have changed over time period and thus need to be examined in terms of the role they play in changing the magnitude and activity of erosion processes. 

Climate over the past Century
The main physical triggering mechanism for most of the erosion processes mentioned above is moisture. Although earthquakes can also play a large role in mobilizing various mass wasting processes, they are not nearly as frequent as precipitation and runoff events that area capable of destabilizing a slope. Intense, prolonged monsoon rains and abundant winter snow fall combined with a steep, geologically weak and weathered environment such as the Kullu Valley leads to the activation of a variety of denudation processes. An increase in the magnitude and/or frequency of precipitation or runoff events could potentially lead to higher moisture conditions and likely an increase in the magnitude and/or frequency of erosion processes. However, previous literature and the statistical analysis of a 100-year precipitation record for three urban centers in the valley (i.e. Manali, Nagar, and Kullu town) indicate that there has been no net change in the precipitation amount over the past century (Fig. 3). Although the intensity and frequency of isolated storm events may have changed, the available climatic data precludes any detailed analysis of individual precipitation events.


Figure 3. Average monthly precipitation at Manali, Nagar, and Kullu over the past century. Grey line indicates the linear trend over the 99 year time period (Source data: Singh, 1995).

Changes in temperature may also affect moisture availability. Gupta et al. (1995) indicate that temperatures in the valley have increased, possibly due to environmental climate change. A wetter moisture regime in the valley would result from the accelerated melting of alpine glaciers and more precipitation falling as rain (higher up slope). Potentially, higher runoff fluxes could contribute to an increase in the magnitude and frequency of floods, debris flows, and associated slope failure processes. Nevertheless, according to Sah and Mazari (1998), geomorphic damages to the landscape caused by floods in 1902, 1945, 1988, 1993, 1995, and 1996 are almost identical, only damages to roads, buildings, and other structures have increased. The apparent lack of an obvious hydraulic adjustment by the Beas River and it's tributaries means that either a large portion of the moisture is being evapotranspirated back into the atmosphere before reaching the surface or that the channels themselves are absorbing the impacts of the excess runoff. In either case, the climatic and geomorphological evidence available suggests that the physical environment is not contributing significantly to the reported loss of lives and property as a result of natural hazards. The following section addresses the role played by land use/cover change on the activity and distribution of natural hazard sites over a period of 27 years in the Kullu Valley.

Mapping land use/cover
As discussed previously, the Kullu Valley has experienced several periods of rapid socioeconomic change, which are reflected in the land use/cover of the area. Land use/cover change analysis in mountain environments focuses mainly on detecting changes between non-built-up and built-up and between forested and non-forested areas. The built-up land use/cover includes not only urban infrastructure within towns and cities, but also individual dwellings, roads linking settlements, and other human-built structures. The stabilizing effect of a forest on mountain slopes has been well documented (Brookes et al., 1997; Ives and Messerli, 1993). The removal of a forest cover from a steep slope often leads to accelerated surface erosion and dramatically increases the chances for landslides as well as runoff. The consequences of deforestation include raised riverbeds due to increased channel siltation, which ultimately leads to more flooding in low-lying areas. Destruction of aquatic habitat and a reduction in the quality of the water, which is an important resource for the local population, are other negative impacts of deforestation.

Historical land use/cover maps are often not available for a given region or the required period of time. Historical records, pictures, and oral history provide a somewhat limited and sporadic insight into the land use patterns of the past, if available at all. A common approach to this problem is to use a time series of satellite images to derive historical land use/cover information for the required time period(s). For this research three satellite images from 1972, 1980, and 1999 and a 1999 land use/cover map were used to reconstruct the land use/cover for the past three decades.

All three satellite scenes were acquired within the first week of November of each respective year using sensors aboard the Landsat 1 (1972), Landsat 3 (1980), and Landsat 7 (1999) satellites. The autumn acquisition date was selected so as to avoid heavy cloud cover and to maximize the spectral differences between coniferous and deciduous vegetation, the latter of which losses its foliage by mid-October. The 1972 and 1980 scenes were acquired using the 4-band multi-spectral scanner (MSS) sensor, while the 1999 scene was recorded using the 8-band enhanced thematic mapper (E-TM) sensor which has improved spectral and spatial capabilities. All bands were spectrally standardized, geo-registered to UTM zone 43N, and cropped in order minimize differences in reflectance and image geometry. Because the 1999 scene was affected by clouds and cloud shadows over approximately 10% of the area, a cloud/shadow mask was developed and applied to the other two satellite scenes in order to maintain an equal study area size Schanzer (1992). The basis for a supervised classification of the satellite scenes was a land use/cover map and information about locations where land use has not changed appreciably over the past 27 years.

A land use/cover map produced by staff at the Science and Technology Office based on a manual interpretation of a series of 1998 7.5m resolution IRS satellite images and detailed field investigation was used as a training layer for the 1999 satellite image. Information about areas where land use/cover did not changed appreciably over the past three decades was used in order to accurately classify the 1972 and 1980 satellite imagery. During an informal interview, local long-term residents were asked to identify areas on photographs and maps which exhibited the same land use/cover characteristics as they did 27 years ago. This information was compiled on a map and used as training data for the 1972 and 1980 satellite scenes.



Classification of satellite imagery in steep, mountainous terrain requires that a topographic component be included in the procedure in order to account for differences in spectral reflectance as a result of topographic shadows due to aspect and slope (Wilson and Franklin, 1992; Fleming, 1988; Frank, 1988; Franklin and Wilson, 1992). Topographic shadows generally reduce or scatter the reflectance from a given land use/cover leading to errors of misclassification. An aspect layer derived from a digital elevation model produced from a series of 1:250,000 and 1:25,000 scale topographic maps with contour intervals of 150 and 40m respectively was included as a "band" in a maximum likelihood classification procedure. As input into the maximum likelihood classifier, bands 4, 5, 6, and 7 for the earlier satellite scenes and bands 1, 2, 3, 4, 5, and 7 for the 1999 scene were used in conjunction with the aspect image. The resulting land use/cover maps contain 5 land use/cover categories and were smoothed using a 3 x 3 median filter.

Land use/cover change detection and hazards analysis
The overlay and change detection analysis was carried out using the ERRMAT (error matrix) module in Idrisi GIS software and the three land use/cover maps. Results of the pair-wise comparison of land use area are given in table 3. 

Table 3. Percent cover of each land use/cover type for the three time periods and the net change between 1972 and 1999.

	1972	1980	1999	D 1972 - 1999
Forest	47.2%	44.0%	40.1%	-15.0%
Clearing	5.2%	4.2%	3.0%	-42.3%
Settlement	2.3%	3.7%	4.7%	104.3%
Agriculture	41.2%	44.1%	48.2%	17.0%
River	4.0%	4.0%	4.0%	0.0%

The results indicate that there has been an increase in settlement and agricultural areas from 1972 to 1999 at the expense of a decrease in forest and clearing (i.e. wasteland/abandoned) areas. The expansion of infrastructure onto geomorphologically unstable and hazard-prone areas was also detected using overlay analysis between the release area, deposition zone, and land use maps. Table 4 summarizes the results of this overlay operation which entailed overlaying the hazard site (release area) and deposition zone maps onto each of the three land use maps.

Table 4. Both tables show the percent land use/cover for each time period and the net change between 1972 and 1999. Table A. shows results of the release area map overlay and table B shows the deposition zone overlay results.
 

A. Release Areas					B. Deposition Zones
	1972	1980	1999	D 1972-1999		1972	1980	1999	D 1972-1999
Forest	20.3%	14.6%	9.2%	-54.7%	Forest	18.9%	17.7%	15.4%	-18.5%
Clearing	3.7%	5.5%	7.4%	100.0%	Clearing	5.1%	7.9%	8.5%	66.7%
Settlement	3.7%	9.1%	6.1%	64.9%	Settlement	2.4%	3.1%	7.8%	225.0%
Agriculture	71.1%	69.5%	76.0%	6.9%	Agriculture	44.9%	41.5%	45.1%	0.4%
River	1.2%	1.3%	1.3%	8.3%	River	27.7%	32.8%	23.2%	-16.2%

Results from the time-series analysis support the theory that the apparent increase in natural hazard activity and the consequent destruction of property and lives is due to a major increase (104% between 1972 and 1999) in the total area under the settlement land use category. This is further corroborated by results from the overlay operation using both the release area and deposition zone maps which show a 64.9% and a 225% increase, respectively, in the total settlement area affected in the same time period. Perhaps more telling is the rather sharp difference between the amount of forest and clearing area loss detected by the time series analysis (15% and 42.3% respectively) and the loss of forest and increase in the clearing category detected in the release areas (54.7% and 100% respectively). A reduction in the forest cover coupled with a large increase in clearing areas in the already unstable release areas would certainly promote further instability and increase the risk of failure. Agricultural areas do not have as clear or direct an impact on slope stability in this area as they do in other Himalayan valleys (i.e. parts of Nepal) due to the gentler slopes and careful terrace construction and maintenance practices. 

Although errors due to classification, registration, conversion, and data quality/accuracy are obviously present, the main question of what is the dominant cause of the apparent increase in the activity of natural hazard processes can be determined with confidence. 

Conclusions
Has the intensification of land use due to population growth led to an increase in the frequency and impact of mass wasting processes on people and infrastructure in the Kullu Valley? This central question can be answered in the affirmative. An examination of the physical (geologic), environmental (climate and precipitation), and cultural (land use/cover) variables has demonstrated that only the latter variable (land use) could exert enough force to alter the relationship between people and natural hazards in the Kullu Valley over the 27-year time period considered. Geology was treated as a static variable given the relatively short study period. Precipitation, the main triggering mechanism for mass wasting events, showed no sign of change based on a trend analysis of mean monthly precipitation for a 100-year period. 

A time series of land use maps (1972, 1980, and 1999) derived from a combination of field work, satellite imagery, and historical documents and records was compared in a pair-wise manner. The time-series comparison of the 1972 and 1999 maps revealed that the most significant changes took place in the clearing (42.3% decrease) and settlement (104.3% increase) categories. More importantly, most of the expansion of the settlement category occurred at or near the valley bottom (within 200m of the riverbed) where the deposition and impact areas are concentrated.

The locations of 52 active and potentially active release sites were identified in the field based on geomorphological interpretation of landforms and formed the basis of a GIS hazards map. A database of quantitative and qualitative site variables and attributes was simultaneously developed and linked to the map to create a flexible repository of natural hazards information for the area. The hazards map then served as a starting point for a disperse operation which modeled the potential run-out or deposition areas for all sites based on an anisotropic friction surface. Forest cover has declined and clearing and settlement areas have increased within the release areas over the 27 year time frame. Similar results are reported for the deposition areas with the notable exception of a dramatic rise in settlement areas (up 225% compared with 64.3% for release areas). Settlements have expanded more than three-fold into active deposition zones near the valley bottom since 1972. This increase combined with a reduction in forest cover in release areas and the associated soil anchoring that trees provide, could certainly lead to an increase in the activity of natural erosion processes and result in more damage and loss of property and lives.

This research has made a significant contribution to the knowledge of natural hazards and their relationship with land use/cover change, not only in the Kullu Valley, but other mountain areas as well. Hazards maps which may be available for some mountain areas typically only look at the historical or current risk situation. In many cases, the release and deposition areas are represented as one entity, a practice which obscures the important differences in land use change between the two unique components of a natural hazard site. This research is currently being expanded to include the development of land use change models which rely on change direction and magnitude vectors to forecast land use change into the future. These models will be made available to land managers and used to augment the land management decision process.

References

• Eckholm, E., 1975. The deterioration of mountain environments. Science, 189: 764-770.
  
• Fleming, M.D., 1988. An Integrated Approach for Automated Cover-Type Mapping of Large Inaccessible Areas in Alaska. Photogrammetric Engineering & Remote Sensing, 54(3): 357-362.
  
• Frank, T.D., 1988. Mapping Dominant Vegetation Communities in the Colorado Rocky Mountain Front Range with Landsat Thematic Mapper and Digital Terrain Data. Photogrammetric Engineering & Remote Sensing, 54(12): 1727-1734.
  
• Franklin, S.E. and Wilson, B.A., 1992. A Three-Stage Classifier for Remote Sensing of Mountain Environments. Photogrammetric Engineering & Remote Sensing, 58(4): 449-454.
  
• Gardner, J.S., Pandey, B.W. and Singh, R.B., 1997. Risk From Natural Hazards in the Manali Area. In: F. Berkes and J.S. Gardner (Editors), Sustainability of Mountain Environments in India and Canada. University of Manitoba Campus Copy Centre, Winnipeg, pp. 404.
  
• Gray, D. H. 1970. Effects of forest clear-cutting on the stability of natural slopes. Bulletin of the Association of Engineering Geologists, 7 : 45-66.
  
• Gupta, A.K., 1997. Effect of Natural Disasters on Roads in India. In: T. Iwasaki et al. (Editors), PIARC G2 Group Seminar: Natural Disaster Reduction for Roads. Ravinder Kumar, Documentation & Library Services, New Delhi, New Delhi, India, pp. 342.
  
• Gupta, K.M., 1990. Himalaya, Man and Nature. Lancers Books, New Delhi.
  
• Hofer, T., 1993. Deforestation-Changing Discharge and Increasing Floods: Myth or Reality? In: B. Messerli et al. (Editors), Himalayan Environment Pressure-Problems-Processes 12 Years of Research. Geographica Bernensia, Berne, pp. 206 and maps.
  
• Ives, J.D. and Messerli, B., 1993. Mountain Hazards Mapping in Nepal, Introduction to an Applied Mountain Research Project. In: B. Messerli et al. (Editors), Himalayan Environment Pressure-Problems-Processes 12 Years of Research. Geographica Bernensia, Berne, pp. 206 and maps.
  
• Kuster, H., 1993. Dynamics of Forest Cover in Indian Himalaya: An Investigation in the Upper Beas Catchment (Kulu-Valley, Himachal Pradesh). In: B. Messerli et al. (Editors), Himalayan Environment Pressure-Problems-Processes 12 Years of Research. Geographica Bernensia, Berne, pp. 206 and maps.
  
• Lauterburg, A., 1993. The Himalayan Highland-Lowland Interactive System: Do Land Use Changes in the Mountains Affect the Plains? In: B. Messerli et al. (Editors), Himalayan Environment Pressure-Problems-Processes 12 Years of Research. Geographica Bernensia, Berne, pp. 206 and maps.
  
• Lillesand, T.M. and Kiefer, R.W., 1994. Remote Sensing and Image Interpretation. John Wiley & Sons, Inc., New York, 750 pp.
  
• Lull, H. W. 1971. Runoff from forest lands. In W. H. Mathews, F. L. Smith and E. D. Goldberg (eds.), Man's impact on terrestrial and oceanic ecosystems, MIT Press, Cambridge.
  
• Mehta, P.K., 1976. Structural and Metamorphic History of the Crystalline Rocks of Kulu Valley, Himachal Pradesh, in Relation to Tectonics of the Himalaya. , 21: 215-244.
  
• Rao, P.J., 1997. Landslide Hazard in Himalaya. In: T. Iwasaki et al. (Editors), PIARC G2 Group Seminar: Natural Disaster Reduction for Roads. Ravinder Kumar, Documentation & Library Services, New Delhi, New Delhi, India, pp. 342.
  
• Schanzer, D.L., 1992. An Automatic Classification Procedure for Coping With Clouds in Landsat TM Data. Canadian Journal of Remote Sensing, 18(1): 30-43.
  
• Selby, M.J., 1991. Earth's Changing Surface: An Introduction to Geomorphology. Clarendon Press, Oxford, England, 607 pp.
  
• Singh, M.G., 1992. Himachal Pradesh: History, Culture, and Economy. Minerva Book House, Shimla, India.
  
• Singh, R.B. and Pandey, B.W., 1995. Common Resources and Sustainable Livelihood of Mountain Environments: A Micro-level Experience of Upper Kullu Valley, Sustainable Reconstruction of Highland and Headwater Regions, New Delhi.
  
• Wilson, B.A. and Franklin, S.E., 1992. Characterization of Alpine Vegetation Cover Using Satellite Remote Sensing in the Front Ranges, St. Elias Mountains, Yukon Territory. Global Ecology and Biogeography Letters, 1992(2): 90-95.
  
• Yudhbir, 1997. Mitigation of Landslide Hazards. In: T. IIwasaki et al. (Editors), PIARC G2 Group Seminar: Natural Disaster Reduction for Roads. Ravinder Kumar, Documentation & Library Services, New Delhi, New Delhi, India, pp. 342.