About GAIA GAIA Case Studies Global GIS Agenda 21 Country Data Model Database

Case Study





Macro-scale, Multi-temporal Land Cover Assessment and Monitoring of Nepal



1. INTRODUCTION

1.1 General

This project is directed in providing information about land cover at a regional scale for selected countries in the Asian region (e.g., Cambodia, Vietnam, Lao P.D.R., Myanmar, Bangladesh and Nepal). The investigation of the NOAA AVHRR satellite data is directed towards the detection and monitoring of ecologically important vegetation types at the national level. This similar activity also highlights the areas where there is a major change of land cover (i.e. "hot spots"), both in temporal and spatial aspects.

The use of remotely sensed data particularly the NOAA AVHRR data was considered a major tool for such purpose. On a much wider scale, this kind of information could also be useful for global research and modelling, macro-economic studies, and assessment of the earth's state of environment. The harmonization of land cover categories was adopted in the interpretation of satellite data for these countries towards a comprehensive regional resource assessment and information aggregation, an important decision input for the regional and national context.

Under this volume, the land cover of Nepal and its related information have been presented while the overall rational behind this activity and its associated methodology appeared in Volume 1-A. This Nepal's Case Study made possible through the ICIMOD/MENRIS-Kathmandu contribution. It is envisioned that such results will serve as a valuable information for a more direct and appropriate formulation of policies and sensible resource management strategies.


1.2 Nepal: Study Area

1.2.1 Location and Physical Characteristics

Nepal is divided into five major physiographic regions which run in more or less parallel bands from northwest to southeast. Each of these regions has a distinctive agricultural and forestry land utilization pattern. These regions are known as Terai, Siwaliks, Middle Mountains, High Mountains and High Himal from south to north direction. Nepal was once extensively covered by forests. Demand for fodder, overgrazing and uncontrolled cutting of timber and fuel wood, have significantly reduced the original forest cover. The composition of vegetation is closely related to the climate, which in turn is related to the physiographic region. There is not only a difference in vegetation from north to south, but also from east to west. The latter is caused by the decrease of monsoon rains in the western part and to some extent by the latitudinal differences between the eastern and western regions of Nepal.

a. The Terai - it forms a long strip of alluvial deposits along Nepal's southern border with India. The strip is an extension of the broad Gangetic Plain, and includes the Bhabar regions, which consist of alluvial fans of the Siwaliks. Originally, the Terai was covered with dense jungle, mainly composed of sal (Shorea robusta) and mixed hardwoods. Following a steady migration of population, most of the forest have been cleared for cultivation, predominantly of rice. Large tracts of forest in the Terai can be found in the Bhabar areas, which are well drained and have a low water table. The soils of the Bhabar areas are generally coarse in texture, subject to erosion if mismanaged, and are not therefore as suitable for agricultural production as the other areas of the Terai. The Terai is not only important for its agricultural production, but also plays an important role in timber production and fuelwood supply. It also provides a rapid and economic east-west transportation corridor. Although the Terai represents only 14% of the total area of Nepal, it contains about 42% of the total cultivated land of the country. The forests consist mainly of high value sal and a mix of tropical and subtropical species.

b. The Siwaliks - its hillslopes have little potential for agricultural production. Soils are shallow and erodable.The majority of agricultural land is restricted to river terraces, alluvial valleys and, despite their limited areas to the Dun valleys. However, occasional hillslope cultivation may be found. The forests consist mainly of chir pine (Pinus roxburghii) and tropical mixed hardwoods of which sal is often a major component. The Siwaliks has very little agricultural land in the valley. The main crops are maize, millet, wheat and mustard. Rice is found where irrigation water is available.

c. Middle Mountains - this region has the highest population density in relation to cultivated land. The height of the Middle Mountains ranges from a few hundred meters to approximately 3,000 meters. The region is generally heavily dissected and has a great variety of soil types, geology and microclimates. Because of these, the area has many different agricultural landuse patterns. The farming systems incorporate a number of such land patterns as well as extensively used public forest and grazing areas. The land in the Middle Mountains is intensively cultivated. Existing pasture land is heavily overgrazed and forests are stripped for fodder and fuelwood. Most slopes are terraced and support maize, millet, rice, wheat or potatoes. It was found that 85% of the cultivated land of the Middle Mountains consists of some form of hillslope cultivation. The badly degraded forests of the Middle Mountains consist mostly of hardwoods with some conifers, mainly pine.

d. High Mountains - unlike the boundary between the Terai and Siwaliks, there is no clearly defined boundary between the High Mountains and the Middle Mountains. The steep slopes of the High Mountains often show intensive agricultural terracing. River terraces are less extensive than they are in the Middle Mountains since rivers tend to develop deep incisions which leave little room for valley cultivation. Upper limits of agricultural land are reached in this physiographic unit. On large, gently sloping fields, one crop of potato, buckwheat or barley may be grown once every year or every two years. Many of the snow fields serve as buffer reservoirs for irrigation water used in the lower regions. Extensive areas of grazing land are found in the form of alpine pastures. Migrating livestock from lower areas utilize the pastures of this area during the monsoon. The forests of this region include both coniferous and hardwoods. Some of the least disturbed forest, especially the conifer forests including fir (Abies spectabilis) and hemlock (Tsuga dumosa) are found in this region. In the High Mountains, upper limits of agriculture are found at about 4,200 meters. In the high regions, fields can support only one crop of buckwheat, barley or potato once a year or once every two years. The High Mountain forests contain a higher proportion of different conifers and in general, are in a better condition than forests elsewhere in the country.

e. High Himal - its climate is predominantly arctic, with permafrost, permanent snow fields and many glaciers. The little agricultural land available is found in the valleys and in some cases in sheltered pockets of the hillslopes. Pasture lands are used by migratory livestock in the High Mountain regions, and by yaks along the trade routes connecting Tibet.

The climate of Nepal varies from subtropical to arctic, all within a distance of approximately 180 kilometers. In addition to the broad differentiations in climate, there is a great variety of micro climatic conditions, resulting in a diversity of landuse and land practices within the country. In general, the climate of the Terai, Dun valleys, and part of the Siwaliks (up to 1000m) is subtropical. The climate of the Middle Mountains (1000-3000m ) ranges from warm temperate to cool temperate, and the high Mountains (2600-4000m) from cool temperate to sub-alpine.

The monsoon, which lasts from June to September, is the most outstanding feature of the climate in Nepal. The monsoon peaks in July and is accompanied by a northwesterly airflow from the Bay of Bengal. Hence, its onset is first experienced in the east of the country while the period from October to March is mainly dry. Occasional precipitation occurs in the form of winter rains, caused by an eastward airflow from the Mediterranean, with local surface heating and/or orographic effects. The influence of these winter rains tends to be stronger in the west than in the east. The rains are of great importance for winter crops such as wheat and barley. In the east of the country the pre-monsoon rains generally last longer, facilitating two monsoon crops in many areas.

The snowline lies around 2500 meters during the winter. Snow rarely falls below the 1500 meter level. On shaded north slopes, snow lingers on considerably longer than on south facing slopes. Farmers make use of this feature since irrigation water is released at a slow and steady pace. Many of the higher snow fields supply irrigation water to the lower agricultural land during most of the year.

1.2.2 Present State of Land cover

The total forest area is estimated to have declined from 6.4 million ha in 1964 to 5.5 million ha by 1985. About 0.2 million ha of the Terai and Siwaliks forests were cleared under planned settlement and because of illegal felling between the 1950s and 1985. (National Report on Nepal, UNCED-92). Further degradation of the forest lands has been observed in the years after 1985, and the reasons can easily be recognized by observing the population movements and the general physiography of the region. As reported by FAO (1991), the forest area is approximately 2,480,00 ha or 18% of the total land area. Further, arable land and permanent pasture accounted for 9% and 15%, respectively of the country.



2.0 RESULTS

2.1 Land Cover Assessment: 1992-1993

2.1.1 Data Quality

Data recorded by the NOAA satellite system covering the Nepal region from June to October 1992 (monsoon) were frequently cloud covered. Data used were mainly from the winter season with few scenes from early summer time. Scenes from different seasons help differentiate forest species from agricultural lands. However, there are not many data sets which were relatively cloud free. Processing of several scenes did not yield a completely cloud free cover of the region. However, the NDVI could still be used for the identification and categorization of land cover. Thus, mainly scenes from the winter season and a few from early summer (in the 1985-86 data) were used in this study.

Most of the AVHRR scenes obtained for the Nepal region had more than 50 per cent cloud cover due to the season, climate, and elevation range. In spite of the mosaicking procedures adopted, the final data used for the 1985-1986 study have almost 45% of the cloud cover. However, the Terai and Siwaliks areas are well under cloud free zones and the results indicate the land transformations, mainly in these two regions.

Further, near nadir scenes would have less distortion and consequently better processing results. However, all of the scenes actually processed were away from nadir, especially the 1985-86 data.

2.1.2 Snow and Cloud Differentiation

Cloud cover has been a critical issue in processing datasets for the Nepal region. The signature validation to distinguish between clouds and snow had to take into consideration several issues. Even though the overlap of range in the thermal bands make it difficult to distinguish clouds from snow, the differing reflectance behaviour of the thermal bands towards snow and clouds gives a good base for the exercise. Additionally, the brightness index from channels 1 and 2, in combination with channel 4, gave fairly good results in identifying snow and clouds.

2.1.3 Spectral Characterization and Land Cover Mapping

2.1.3.1 The Classification Scheme

Land cover classification usually differs from country to country. Keeping in view that the present study should ultimately build up a common database at the regional level, the following classification system has been followed to study land cover and to observe changes.


i). Forest Areas - major forest types in Nepal are tropical and sub-tropical, warm temperate, cool temperate, sub-alpine and alpine. The natural forest mainly consists of hardwood, mixed and coniferous species. Hardwoods are dominant in the Terai and lower hills while conifers grow at higher elevations. Four major forest subclasses are identified in the classification. Conifer, hardwood, mixed conifer, and hardwood forest and shrub. Further, the hardwood and mixed forests are also identified with a relative crown density as high and low.

ii). Agricultural Areas - even though the classification system in Nepal generalizes the agricultural areas as three major cultivation subclasses, viz. terai, hillslope, and valley cultivation, all the agricultural areas are categorized into one single class as "agriculture".

iii). Grazing Lands - areas used primarily for grazing lack sufficient shrub or tree cover to be included as forest.

iv). Rock - rock outcrops are found throughout the hill areas of the country, with a higher percentage in the High Himal. In High Himal, large areas of gently sloping stony pavements are mapped as rock.

v). Snow - the winter snowfall is heaviest in the western parts of the country; the winter snow line is lower than in the east. Snow also tends to stay longer into spring in the west. Snow cover is quite large through the winter season and is categorized separately.

2.1.3.2 Land Cover Distribution and Assessment

The final observations are based on the two time series data sets in combination with the LRMP datasets produced from field surveys and aerial photography. Table 1 indicates the areas classified under the adopted categorization scheme. The figures do not reflect the total area of each of categories since only about 56% of the total area was under cloud free zone in 1985-1986. The figures under the column head indicating 1992-1993 may be taken as a good indicator of land cover areas because most of the cloud cover area (approximately 14%) for this period is usually snow covered. Thus a comparison of the analytical results does not represent a true quantitative analysis. The qualitative analysis indicates the significant changes of the forest areas in Terai region. (Note: These changes were not verified through field surveys. However, considering the strong signatures on the satellite scenes, the indicated forest degradation in the Terai area by population encroachment for cultivation practices can be assumed and accepted with a high degree of confidence). Maps 1 and 2 show the different land cover for 1992-1993 and 1985-1986 while Table 2 displays the different land cover type distribution for 1992-1993 in various physiographical regimes of the country.

Map 1 Classifed Land Cover Map of Nepal 1992/1993


Map 2 Classifed Land Cover Map of Nepal 1985/1986


Table 1. Nepal Land Cover Areas (Sq.km./%)

Land Cover

1980 ***

1985-1986

1992-1993 **

Coniferous

6.3 %

3,736 (2.6%)
2,722 (1.8%)
Hardwood Forest

16.6 %

14,923 (10.4%)
20,927 (14.6%)
Hardwood Forest
(low crown density)

6.1 %

25,488 (17.7%)
17,655 (12.3%)
Mixed Forest

2.0 %

4,009 (2.8%)
8,037 (5.6%)
Mixed Forest
(low crown density)

6.9 %

86 (-)
4,790 (3.3%)
Agriculture

26.9 %

29,529 (20.6%)
45,135 (31.4%)
Shrub

4.7 %

-
2,180 (1.5%)
Grazing

12.1 %

-
8,105 (5.6%)
Rock

14.6 %

1,729 (1.2%)
10,300 (7.2%)
Snow

3.6 %

-
4,075 (2.8%)
Cloud Cover

- -

64,223 (44.7%)
19,795 (13.8%)

** The figures are not indicative of the total land area because of the cloud cover and hence cannot be directly compared with figures in the 1992-93 column.
*** Based on 1:500,000 scale Land Utilization maps prepared by the His Majesty's Government of Nepal and the International Bank for Reconstruction and Development / The World Bank. Source: The Land Resources Mapping Project (LRMP) map series. Data for the 1:50000 scale LRMP Land utilization Map series (published in 1985) was derived from aerial photography and extensive field work carried out between 1979 and 1985. MMU area was 25 ha. MMU, for 1992-1993, 900 ha.

2.2 Accuracy Assesment

Because of very rough terrain and insufficient reference data in the region under consideration, the derived land cover maps gave an accuracy level of 60%. However, the accuracy level for the Terai and Siwaliks area has been observed to be much above 80%, which is quite satisfactory. Categorization of coniferous forests from mixed forests and hardwoods also was quite significantly difficult in this area.

A comparison of the classified data with the LRMP data derived at 1:50,000 scale indicates a considerable reduction in the coniferous forests. The LRMP data classification indicates the protected forests as a separate entity. Hence, the difference of figures in Table 3 do not necessarily indicate the exact land changes. The signature validation for the coniferous in the High Mountain and Middle Mountains also should be considered.

Table 2. Land Cover Areas by Physiography in Sq.Km. (1992-93)


Land Cover

Terai

Siwaliks

Middle
Mountains

High
Mountains

High
Himal

Coniferous

-
20
688
1312
569

Hardwood Forest

-
4306
11154
3916
436

Hardwood Forest
(low crown density)

90
5188
7671
2387
290

Mixed Forest

1254
143
2694
3956
851

Mixed Forest
(low crown density)

-
327
2465
1575
190

Agriculture

-
7460
12078
3879
1323

Shrub

18183
41
1571
455
-

Grazing

-
12
625
3941
3110

Rock

18
-
195
2293
7298

Snow

-
-
-
584
3287

No data available
(cloud covered)

-
-
369
3437
15019

2.3 Land Cover Monitoring

On the basis of observation of the two sets of AVHRR data on Nepal for 1985-1986 and 1992-1993 along with the LRMP (Land Resource Mapping Project) data of 1980, the changes that occurred in the Terai could be identified. Areas other than the Terai region also indicated changes in the land cover. However, these observations should be carefully considered since the accuracy assessments for these areas were not definite to reflect the significant variability in the High Mountain areas.

During the period under consideration (1985-1993), the forest areas in the Terai seem to have been cleared by the people for cultivating the land. Some of the areas of the Terai, recommended for protection against soil erosion, also seem to have been degraded. Most of the forest degradation of the Terai seem to be due to population pressure and cultivation practices. Areas where forest has degraded quite heavily were also observed in the Terai region.


3.0 CONCLUSIONS

At the national scale, this project showed the usefulness of NOAA AVHRR data in the assessment and monitoring of land cover information for sustainable development planning and management. In addition, the availability as well as the reliability of ancillary data needed for the interpretation of coarse resolution AVHRR data are considered as an indispensable input to generate more precise land cover information. For Nepal, 10 classes were discerned (e.g., coniferous, hardwood forest, hardwood forest(low crown density), mixed forest, mixed forest(low crown density), agriculture, shrub, grazing, rock, and snow) of which four out of 10 were attributed to different forest types .

The use of AVHRR data for monitoring land reforms at regional scale was viewed from two different aspects. The qualitative assessment of the land cover status can make a strong statement towards land reforms. Quantitative assessments of the land reforms from such studies should be carefully regarded because of the coarse resolution of spatial data. Even though these results can still be considered, a high level of accuracy has been assessed. Considering the present scale of temporal and spatial change of land cover in these countries, at least a decade is required to indicate a more comprehensive land cover change monitoring. The predominant shifting cultivation practices signify a gradual but chronic degradation of the landscape in line with the nature of land transformation.























REFERENCES



Burrough, P.A. 1986. Principle of Geographic Information Systems for Land Resources Assessment. Oxford, U.S.A.

Curran, P.J. 1980. Multispectral Remote Sensing of Vegetation Amount. Prog. Phys. Geogr., 4, 315.

Defourny P. et al.,1993. Overview of South-East Asia Land Cover using a NOAA AVHRR one kilometer mosaic, Under review for publishing in Geocarto International.

ERDAS, 1991. ERDAS Field Guide. ERDAS, Inc. Atlanta, GA.

FAO, 1992. Forestry Profiles. Tiger Paper. Vol., 19, No. 4. October-December 1992, Bangkok.

Malingreau, J. P. 1992. Satellite Based Monitoring; A Review of Current Issues. Proceedings World Forest Watch Conference in Brazil. Joint Research Centre, Commission of the European Communities.




Copyright 1995-2002 by:   ESS   Environmental Software and Services GmbH AUSTRIA