Home Articles Snow melt runoff simulation and seasonal snowmelt volume forecast using Remote Sensing...

Snow melt runoff simulation and seasonal snowmelt volume forecast using Remote Sensing and Geographic Information System

A.Jeyaram, Y.V.N. Krishna Murthy, D.S. Pandit, S.Adiga*
Regional Remote Sensing Service Centre, ISRO
Department of Space, Amravati road, Nagpur – 440 010
* RRSSC-NNRMS, ISRO Headquarters, Bangalore

The Martinec – Rango snowmelt runoff model has been successfully used in the United States, Japan, Poland and French Alps. In addition to snow cover data, the model requires hydro-meteorological parameters in the higher altitudes of watershed. But these parameters are not available in the higher altitudes of Himalayan catchments. Precipitation and temperature data are generally measured in the lower altitudes of the Himalayan catchments. In view of the above fact, the model has been modified wherever necessary, to suite the Himalayan watershed conditions. Beas catchment above Manali, Himachal pradesh has been taken for the present investigation. A package, SRM-GIS has been developed using C – language and AML programming in GIS environments for daily snowmelt simulation and seasonal snowmelt volume estimation. SRM-GIS is semi-automated package and performs snow cover estimation using satellite data, calculation of coefficients, degree – day factor, recession coefficients, snow/rain separation, orographic increase in precipitation, model calibration, regression models, goodness of fit and plotting of hydrographs, mass curves etc. The model has been successfully applied in Beas watershed above Manali for six years period. The capability of the model has been demonstrated through the package SRM-GIS to provide reasonable simulation of snowmelt hydrograph and runoff volume estimation.

A vast amount of snow is deposited on the Himalayan slopes during the winter months forming biggest water resources in the northern parts of India which continuously melt and feed the north Indian rivers thus making them perennial. The estimation of snowmelt runoff and resulting stream flow from Himalayan catchments is of considerable use in planning and operation of river valley projects. Different models have been developed in various parts of the world and the temperature index model is among the notable ones developed by Martinec & Rango (1979) and is used by various organizations in many parts of the world. The snowmelt process is influenced by a number of factors and taking all the factors into consideration is practically impossible. In temperature index degree – day approach, maximum atmospheric temperature is taken as an index which is the representative of all the melting factors. The degree day for melt computation is normally the positive departure of the maximum temperature above base temperature of 0 0C. Thus, this is a empirical measure of the amount of energy available to melt snow based on atmospheric temperature alone. The data in most of the Himalayan catchments are available only in the lower altitudes thus making them unrepresentative in the higher altitudes. The suitable adjustments are made to suite the Himalayan catchments and snowmelt runoff volume has been calculated with reasonable accuracy.

Experimental Watershed
Beas watershed above Manali covers an area of about 344.09 sq. km and lies in Himachal pradesh of North India.. The altitude ranges from 1900 m to 5932 m above msl (HANUMAN TIBBA) which is representative of a typical high rise Himalayan basin and there for has been considered suitable for developing and testing methodologies appropriate to the Himalayas. The watershed experience heavy snowfall covering 80-90 percent of the area during winter and with minimum temperature if the order of -12 C at base station Manali and about 90 percent of the are is snow free during summer. The hydrometeorological data on routine observation are being carried out by Bakra-Beas Management Board (BBMB), Mountaineering Institute and snow and Avalanche Study Establishment (SASE) for their own purpose. The catchment is divided into 20 elevation zones each of 200m attitude. The study indicates thick vegetation in the lower altitudes of the watershed and bushes in higher altitude.

The Martinec – Rango model has been successfully used in the United States, Japan, Poland and French Alps. In addition to snow cover data, the model requires temperature and precipitation as inputs. These input parameters are necessary in each of the elevation zones. Precipitation and temperature are generally measured in the lower altitudes of the Himalayan basins. These parameters are not representing the higher altitudes of the basin. The degree – day factor is obtained from an empirical formula as a = 1. 1 pslpw where ps is density of snow and pw is density of water. The large variation of degree – day factor is experienced in Beas basin above Manali due to large difference in elevations as well as it is not possible to calculate in each of the elevation zone. The model accepts satellite based snow cover inputs in each of the elevation zones. The snow cover area estimation in each of the elevation ranges- could not be carried out accurately due to large difference in elevations. In view of the above fact, the model has been modified wherever necessary, to suite the. Himalayan watershed

The snowmelt on any day may be expressed as the ordinate of normal recession curve together with additional discharge due to snowmelt and rain in the catchment on that day, thus the discharge on the n+lth day is given by

where Q – average daily discharge in m3 sec-1
SM – Snowmelt volume, M 3
RI – Rain input volume, M3
K – recession coefficient
a – degree – day factor, mm. oC . d-1
Tsub>max – maximum atmospheric temperature, oC
A – snowcover area, in sq.km
p – precipitation at base station at Manali in mm
g – orographic increase in precipitation
x – percentage of snowfall
i – elevation zone
n – sequence of days during the discharge computation period.

Snowmelt runoff model based on Rango & Martinec (1979) with appropriate modification has been developed in “C”. These routines are invoked in ARC/INFO – GIS environment through systematic sequences (Through AML) as customised application interface called SRM-GIS. Image processing software package EASI/PACE functions available in developers toolkit are used to determinate snow line altitude and to, calculate snow cover area, percent snow cover area using area – elevation data. The package is a semi-automated with graphical functions available in IBM, RISC 6000 workstation.

Structure of SRM-GIS
The hydrometeorological data of measured discharge, temperature (minimum and maximum), precipitation data ( rain and snow ), digital elevation data of the watershed and the satellite data form inputs for SRM-GIS package and hydro-meteorological data are kept as .dat filesand satellite & DEM are kept as .pix files. The package is designed and developed with a set of pull-down menus.

The main menu in this package consists

Image Processing Module

Snowcover area Estimation invokes EASI/PACE image processing environment, open display window and loads satellite data (FCC). The user has to enter the date of pass (Date of acquire) file name of image. By clicking, along the snow line it automatically calculates the average snow line altitude and calculate snow cover area, percent snow cover area by area-elevation method.

  • Satellite data
  • Elevation data
  • Snowcover depletion curves
  • Area – elevation curves
  • Hydro-meteorological data
  • Quit
  • Data Base Module
    The database module invokes data menu which contain the input data used for simulation purpose which can be displayed and viewed.

    Model Calibration module
    This module calculates recession coefficient, degree-day factors and snow/rain coefficients. It calculates recession coefficients for given period starting form the given year. If given period is more than the years in data file it calculate till the end of data base year. It calculate degree-day factors for the year of simulation. It calculate Snow/Rain coefficients for observed temperature and percent snowfall. All these coefficients displayed on monitor in a window. licking on model calibration invokes model menu which calculates different coefficients.

  • Recession Coefficients
  • Degree-Day factor
  • Snow/Rain coefficients
  • Quit
  • The module calculates various coefficients like recession coefficients, degree day factor, snow/rain separation coefficients, plotting of these graphical models and use these coefficients for final snowmelt calculations.

  • Model Run
  • Nash Sutcliff goodness of fit
  • Hydrograph plot
  • Mass curve
  • Quit
  • Snowmelt Runoff Simulation module
    The module would perform calculation of snowmelt runoff on daily basis, calculates goodness fit between measured and estimated snowmelt, plotting of measured and estimated snowmelt as hydrograph and plotting of cumulative measured and estimated snowmelt along with correlation coefficients, R2 .

    Model run accepts the period of estimation in terms number of days and calculates the snowmelt discharge estimation for the said period using the data from the previous modules and data files. It calculate daily discharge for the given period in M3 /s. Clicking on Hydrograph-Plot, plots the hydrograph of observed and calculated discharge for user entered period. Mass curve Plot the mass curve of observed and calculated discharge for the specified period.

    Seasonal Snow Melt Volume Forecast
    Seasonal snowmelt runoff volume forecast is basically the total amount snowmelt during the effective melt period i.e. from April to June. This is being calculated based on the correlation between snowcover area and historic snowmelt runoff by linear and non-linear methods.

    Through image invokes image processing environment, the satellite data is loaded and the snowline altitudes are obtained as in the image processing module. The snowline altitude obtained either from image processing module or through key board form direct input for the seasonal snow melt runoff volume forecast.

    Snow cover area – runoff method estimates the discharge (From 1st April to end of June.) by linear correlation and estimated seasonal snowmelt runoff volume is given in M 3 . Regression method estimates the discharge for melt season by non-linear regression and estimated seasonal snowmelt runoff volume is given in M3 .

    Summary and Conclusions
    Customised application interface SRM-GIS has been developed to provide complete solution for daily snow melt runoff simulation and seasonal snow melt forecast. The mapping of snow cover area from the satellite data using digital image processing has been integrated with SRM-GIS with minimal user interaction as an important input for daily snowmelt runoff simulation and seasonal snow melt forecast procedures. The package calculates snow cover area and is stored as one of the input along with precipitation, temperature (Max, Min) and observed discharge data. SRM-GIS calculates various coefficients of recession, degree-day factor, snow/rain separation etc to fine tune the main snow melt model. These coefficients are used for daily snow melt runoff simulation and Nash and Sutcliffe’s goodness of fit (R2) is also calculated for the observed and calculated discharges. The final out put of hydrograph and mass curves also displayed on the monitor.

    The SRM-GIS package can be used for any of the snow cover catchments similar to watershed above Manali in Himalayas. ‘The adaptability of this package for any other watershed is simpler and very user friendly.


    • Bagchi, A. K. , Ranga Raju, K. G. , and Tiwari, R. S. , 1981, Studies concerning precipitation in Beas catchment using satellite images, IAHR seminar communications, New Delhi, February 1-7, 1981.
    • Jeyaram, A. and Bagchi, A.K., 1982, Snow cover area estimation in Tos basin using Landsat images, International symposium on hydrological aspects of mountainous watersheds, U.O.R, Roorkee, November, 1982.
    • Jeyaram, A., Tiwari, R.S., and Sharma, K.P., 1983, Snowmelt runoff using Landsat images, First national symposium on seasonal snow cover, Vigyan Bhavan, New Delhi, April, 1983.
    • Kuz’min, P. P. , 1972, melting of snow cover, UOS, department of commerce, national technical information service, translated from Russian by the Israel programme of scientific translation, Jerusalem, 1972.
    • Martinec, J. , 1960, The degree – day factor for snowmelt runoff forecasting, proceedings of general assembly of Helsinki commission on surface waters, IASH publication no.51, 1960. Martinec, J. and Rango, A. , 1979, Discharge forecast in mountain basins based on satellite snow cover mapping, proceedings of operational applications of satellite snow cover observation, NASA conference publication 2116, 1979, pp 223 – 238.
    • Shafer B.A. and Jones E.V. and Frick D.M., 1981, snowmelt runoff simulation using Martinec-Rango model on the South Fork Rio Grande and Conejos river in Colorado, AGRISTARS, conservation and pollution technical report NASA, 1981.