USE OF LANDSAT DATA FOR SNOW COVER ESTIMATION AND RELATED RIVER RUN-OFF IN BASIN IN

Siddiqui, Z.A., Farooq, M., Khan, P.A. and Shah, S.S.

Remote Sensing Applications Division, P.O. Box 8402, SUPARCO, Karachi-75270, Pakistan.

ABSTRACT

More than two third of the annual flow in the Jhelum river basin is derived from the melting of snow in the catchments areas. For optimal operation of canals, dam reservoirs, hydel power stations, headworks etc., it is vitally important to obtain repetitive information on the areal extent of snow cover in this basin. In view of the vastness and remoteness of the mountainous area, satellites are the only source for obtaining information on snowcover on routine basis. Landsat Thematic Mapper (TM) bands 3 and 4 data have been used in this study for estimation of the areal extent of snowcover in Jhelum river basin comprising four sub basins viz; Kunhar river at Naran, Kunhar river at Ghari Habibullah, Kishanganga river at Muzafferabad and Jhelum river at Kohala. Correlation between snowcover extent at the beginning of snowmelt and the subsequent river run-off for different years has been attempted. A linear regression equation was determined. This equation, however need to be tested further so that the expected snowmelt river run-off can readily be determined if snowcover data of the basin is available. The expected snowmelt run-off data can therefore be provided to the end user agencies.

1. INTRODUCTION

It is estimated that over 70% of the annual flow through the rivers of the Indus basin are derived from the melting of snow and glaciers in the mountainous catchment areas of Northern Pakistan. Information on the extent of snowcover in the upper Indus basin in Northern Pakistan is, therefore, of vital importance for the economy of Pakistan. The regulation and operation of dam reservoirs, barrages, Headworks, canals as well as hydel power plants can be optimized if the volume of water expected to flow through the rivers as a result of snowmelt could be known in advance. This would obviously lead to more efficient utilization and management of available water resources for agriculture as well as for other industrial and municipal need. This, of course, can only be possible if reliable empirical relationship between snowcover extent and river run-off has been made and to relate this with the river run-off have been derived. It is therefore, highly pertinent for countries like Pakistan to apply satellite remote sensing data for the estimation of the areal extent of snowcover and to relate this with the induced snow melt river run off for optimal management of its water resources.

The relevance of satellite remote sensing to snow surveying on routine basis is well established. Orbiting satellites are particularly well suited for monitoring and mapping of extent of snowcover over large inaccessible areas like the mountain ranges. For such hazardous mountainous areas, conventional ground and aerial surveying cannot always provide the desired areal coverage or the desired observational frequency, in view of the obvious hazard and high cost involved. Satellites are thus; the only mean whereby the snowcover over such large inaccessible areas can be studied repetitively. Space-acquired imagery has been extensively used for studying and mapping snowcover, with reference to its potential applications in river run-off forecasting.

2. STUDY AREA

The study area covers the whole of Jhelum River drainage system. It lies between 33:20 N to 35o: 20/ N and 73o: 10/ E to 75o: 40/ E. The location of study area is shown in Figure 1. Jhelum River basin/watershed comprised the following sub-basins: • Kunhar river at Naran • Kunhar river at • Kishanganga river at • Jhelum river at Kohala.

Proceedings of the 21st International Cartographic Conference (ICC) Durban, South Africa, 10 – 16 August 2003 ‘Cartographic Renaissance’ Hosted by The International Cartographic Association (ICA) ISBN: 0-958-46093-0 Produced by: Document Transformation Technologies Figure 1. Snow survey study area (Jhelum river basin)

The mean elevation of the area is very high ranging from 8000-12000 feet above sea level. The study area has a number of streams that join the river at different points. Stream gauge stations for the measurements of river discharge are located at appropriate points.

3. SATELLITE REMOTE SENSING DATA ANALYSIS AND SNOWCOVER ESTIMATION

The most suitable band for identification and delineation of snowcover is Band-3 (0.6-0.7 microns) since it provides the best contrast between snowcover and snow free areas. The period 1-15 April can be considered to be the time at which in general, substantial melting of the snow pack begins in the area. From the point of view of driving empirical relationship between snowcover and the run off during the spring/summer-melting season, data of the first half of April is therefore significant. However, snow-covered images (ID No. 150-36 and 149-36), which cover the study area pertaining to the date 29-3-1998, were available in the archive. Further, imagery prior to early March is normally avoided on account of the shadow effect produced by low sun elevation angle of about 25o-35 o in February and early March over a high mountain relief. By late March-April, sun elevation angle is generally over 45o and the shadow effect has disappeared.

For computing the areal extent of snowcover, mosaics as shown in Figures 2 and 3 were made of Landsat TM imagery prints at the scale of 1:1,000, 000. Squared millimeter grids of ordinary squared millimeter graphs were overlaid on the mosaic for direct counting of number of squares falling over the snow-covered area in each sub-basin. Since 1:1,000, 000 scale prints were used; the number of such squares directly gives the areal extent of snowcover in square kilometers. The areal extent of snowcover was thus estimated for each constituent sub-basin/watershed of the Jhelum River as tabulated in Table 1. In this table, estimates related to snowcover and snowmelt run-off during the year 1979 have been incorporated for reference purpose. Percentage decrease (depletion) in snowcover and related snowmelt run off are also shown in the Table 1. Figure 2. Snowcovered area of Jhelum river basin (Landsat TM Image Band-3 March, 1998)

Figure 3. Snowcovered area of Jhelum river basin (Landsat TM Image Band-4 March, 1998) Table 1. Sub-Basin-wise Snowcover and Snowmelt Run-off Estimates of Jhelum River Basin Basin/Watershed Snow covered Area (sq.km) Snowmelt Difference in %age Area (km2) and % area of basin shown run-off Runoff Depletion in brackets (106 m3) Of 1979 and of Snow 1998 1979 1998 1979 1998 Kunhar River at 991 995 1282 1282 0 0% Naran (1036) (95.6%) (96.04%) Kunhar river at Ghari 1522 770 2301 1164 1137 50.58% Habibullah (2383) (63.86%) (32.31%) Kishanganga River at 4725 2900 7273 4464 2809 61.37% Muzaferabad (7278) (64.92%) (39.84%)

Jhelum River at 10056 1350 12884 1729 11155 86.6% Kohala (24890) (40.40%) (5.42%) Total 17294 6015 23740 8639 15101 65.2%

4. CORRELATION BETWEEN SNOWCOVER ESTIMATION AND RIVER RUN-OFF

Snowcover estimates used for correlation with river run-off are based on visual interpretation of Landsat imagery. The areal extent of snowcover in square kilometer (sq.km) measured along the X-axis as the independent variable was plotted against the corresponding total run-off during a specific time period (as recorded by WAPDA's stream gauge stations) in million cubic meter (106m3) measured along the Y-axis as the dependent variable were plotted through the MS Excel to relate snowcover to total run-off separately for each sub-basin It was inferred that run-off depends on the snowcover in linear fashion which could be described by a relationship of the type Y=mX+C, where X and Y represent snowcover and run-off respectively. A statistical linear regression analysis was therefore performed which yielded a data set valued of constants a and b, for a linear relation of the type Y= aX + b (which of course is the same as Y= mX+C) where X and Y represent snowcover and run off respectively. The resultant equation with co-efficient of correlation (r) is given as mentioned below: Y= 656+ 1.14X Y= 656+ 1.14(6015) = 7513 Run-off figure for 1998 was 8639 million cubic meters (WAPDA’s estimates) while from the equation it is 7513 million cubic meters. This shows a difference of 15% in river run off between WAPDA's and figure determined by analysis. This shows that there exists a closed correspondence between the WAPDA’s estimates and those obtained from regression analysis. This also infers that if snowcover estimates are available then the expected snowmelt river run-off can be easily determined through this analysis. The difference in snowcover depletion during 1979 and 1998 is as follows: 1979: 17294 km2 1998: 6015 km2 Percentage difference: 34.8%, which shows 100-34.8 = 65.2% depletion of snow during the period 1979-1998.

5. CONCLUSION

Satellites are the only means of obtaining repetitive information on snowcover in vast, inaccessible and hazardous areas like the Upper Indus Basin and adjoining areas in Northern Pakistan. Such type of information on snowcover is very valuable and can be used as input for the estimation of river run-off during the melting season. River run-off forecasting can be useful for an agrarian country like Pakistan for optimizing the operations of dam reservoirs, hydel power plants, canal irrigation system etc. This will certainly lead to more economical utilization of available water resources.

Attempts to correlate the satellite derived estimates of snowcover extent and corresponding river run-off during the melting season indicate that the two are relateable. The resultant linear regression equation appears to hold reasonably well on the whole. More data obviously need to be correlated before definite statements can be made and valid relationship suitable for use in run-off forecasting can be derived. 6. REFERENCES

[1] Rango, A; Salomonson, V.V; Foster, J.L., Employment of Satellite Snowcover Observations for Improving Seasonal Run-off Estimates, Proceedings of the Workshop on Operational Applications of Satellite Snowcover Observations, South Lake Tahoe, California, U.S.A., NASA SP-391, pp 157-174 (1975) [2] Barnes, J.C., Bowley, C. J.: The Evolution of Satellite Snow Mapping With Emphasis on the Use of Landsat Data in the Snow AVST Study Areas, Proceedings of the Workshop on Operational Applications of Satellite Snowcover Observations, Sparks, Nevada, U.S.A., NASA CP-2116, pp 1-19 (1979) [3] Odegaard, H. A., Andersen, T., Ostrem, G.: Application of Satellite Data for Snow Mapping in Norway, Proceedings of the Workshop on Operational Applications of Satellite Snowcover Observations, Sparks, Nevada, U.S.A., NASA CP-2116 pp 93-106 (1979) [4] Andersen, Tom, Odegaard, H. A.,: Application of Satellite Data for Snow Mapping in Norway, Norwegian Committee for Hydrology Report No. 3, Oslo, (1980) [5] Odegaard, H. A., Osterm, Gunnar,: Application of Satellite Data for Snow Mapping in Norway, Vassdrags Direktoratet Hydrologisk Avedeling Report No. 7-77, Oslo, ( 1977) USE OF LANDSAT DATA FOR SNOW COVER ESTIMATION AND RELATED RIVER RUN-OFF IN JHELUM RIVER BASIN IN PAKISTAN

Siddiqui, Z.A., Farooq, M., Khan, P.A. and Shah, S.S.

Remote Sensing Applications Division, P.O. Box 8402, SUPARCO, Karachi-75270, Pakistan.

CURRICULUM VITAE

1. POSITION :Image Analyst

2. NAME : Zamir Ahmad Siddiqui

3. DATE OF BIRTH :19th May 1950

4. NATIONALITY : Pakistani

5. EDUCATION: M.Sc. (Physics) 1972

6. OTHER TRAINING: Course on Remote Sensing Technology and Application, Germany. Course on Microwave Remote Sensing, Bangkok. 7. LANGUAGE & DEGREE OF PROFICIENCY : Urdu Excellent English Excellent 8. MEMBERSHIP OF PROFESSIONAL SOCIETIES : Member Pakistan Physical Society. Member Pakistan Association of Scientists and Scientific Professions Member Karachi Physics Society 9. COUNTRIES OF WORK EXPERIENCE : Pakistan

10. COMPUTER SKILLS : Operating systems used : Window XX Programming languages : FORTRAN.

Software packages used :

Image Processing ERDAS Imagine, ER-Mapper.

Geographic Information System (GIS) Arc/Info, Arc/View

11. EXPERIENCE RECORD

FROM November 1975 : To Date EMPLOYER : Pakistan Space & Upper Atmosphere Research Commission (SUPARCO) POSITION HELD AND : Manager DESCRIPTION OF DUTIES : Conduct and overall supervision of various remote-sensing projects related to water resources management and development of various GIS databases.

Flood plain mapping and monitoring of Indus basin river system In this project carried out mapping and monitoring of flood plain making use of SRS data.

Flood inundation of Indus Basin since 1973 flood Mapping and estimation of flood inundated areas of different floods pertaining to years 1973,76,78,88,92 and 97 was carried out. Reports of these studies were prepared and forwarded to Federal Flood Commission, WAPDA etc.

Snow survey mapping in the Northern Pakistan Snow cover mapping with a view to correlate with snowmelt river runoff was carried out in the Indus basin river system.

Dam reservoir siltation/turbidity mapping Mapping of dam reservoirs like Tarbela, Mangla and Hub was carried out and delineated areas of active silt deposition.

Flood risk zone mapping Mapping of areas prone to floods and their classification was carried out. Identified areas likely to be flooded.

Environmental degradation of lakes Environmental degradation and impact assessment of Kalri and Manchar lakes in Sindh was attempted and proposed recommendation for their preservation.

Development of flood GIS database of Indus rivers system Comprehensive flood GIS database of historic floods in the Indus rivers system was carried out, depicting detailed hydrological and geomorphic information related to Indus basin

12. PUBLICATIONS : 30 scientific research papers/reports of water resources and environmental studies presented and published in different national and international journals, seminar, symposia, conference etc.