SNOWPACK ESTIMATION AND ITS IMPACT ON RIVER FLOW REGIME

Abstract

In the Himalayan region, snow and glaciers are apex natural water resource reservoirs which release large quantity of freshwater round the year. The water yield from high Himalayan basin is considered as a dependable source of water supply for drinking, irrigation, hydroelectric power generation and for other miscellaneous purposes like recreation etc. That is why in northern India, snowmelt water of the Himalayan glaciers is considered as the lifeline for all the living beings. The runoff pattern in terms of timing and intensity in these river basins is governed by the quantity and the spatio-temporal distribution of snowpack. The snowmelt in these river basins is highly dependent on snow physical parameters, i.e., Snow cover area (SCA), Snow depth (SD) and Snow water equivalent (SWE). The long-term global, regional and local changes in these parameters will ultimately affect the functioning of the ecosystem, human use of snow resources and the climate itself through the feedback mechanisms like that of snow albedo. Thus, it is important to study the snow parameters, such as SCA, SD and SWE. Further, the sediment transport in the Himalayan basins is another big problem from the water resources development point of view. The capacity of the reservoirs in the downstream areas is getting reduced at an alarming rate due to the sediment deposition. Thus, the sediment yield assessment from the catchment is an important aspect so as to suggest suitable measures to reduce it. Keeping these issues in mind, the present study focusses on the assessment of snowpack parameters, simulation of snowmelt runoff and sediment yield assessment. Since the climate change is going to be the important decisive factor on the snowmelt, the study also focusses the impact of climate change on snowmelt based streamflow. The present research has been elaborated through the case study of upper part of the Satluj River Basin up to Rampur (Indian part) in Western Himalaya. The catchment area is about 16,650 km2. It is a highly snow-fed mountainous basin whose about 85% area is covered by snow during peak winter season. The hydropower potential of Satluj basin is about 8634 MW, which is equal to almost 50% potential of the entire state of Himachal Pradesh. In order to assess the SCA, SD and SWE and its long term inter-annual variability in space and time, two approaches, namely, (i) remote sensing method and (ii) hydrological modelling method using Variable Infiltration Capacity (VIC) Macroscale hydrological model, have been suggested. In remote sensing approach, the MODIS data is used for SCA and AMSR-E passive microwave remote sensing data is used for the SWE purpose. vi The second part of the study focusses on the simulation of the stream flow from this highly snowfed mountainous basin, where snowmelt runoff contribution is predominant. To find out the contribution from the snow and glacier in the river discharge, three models namely, VIC, SWAT and SNOWMOD have been used. The VIC snow model is based on the energy balance approach; the SWAT snow model uses a temperature index approach and the SNOWMOD uses the two combined approaches i.e., the degree-day approach coupled with elevation band approach to simulate the snow processes. The study describes the SWAT, VIC and SNOWMOD models and process of model set up (calibration and validation) in hydrological simulation components. It is followed by the results and discussions where changes in stream flow and water balance of the basin is predicted for projected climate scenarios. For the purpose of water budget analysis, a period of 21 years (1985 to 2005) is considered. For the purpose of sediment yield assessment, GIS based sediment modelling approach and relationship between suspended sediment load and discharge method is adopted. The input parameters used in these models are daily hydro-metrological data (precipitation, maximum-minimum temperature and wind speed), soil type, Landuse/cover and topographical data. The hydro meteorological and sediment data used in this study were taken from different sources. For example, to run the VIC model, the precipitation data at 0.25°x0.25° resolution received from IMD, while the daily minimum-maximum temperature and wind speed data at 0.125°x0.125° were downloaded from ERA INTERIM websites. On the other hand, snowmelt runoff simulation is carried out using SWAT and SNOWMOD models with daily maximumminimum temperature, rainfall and relative humidity data of five stations, namely, Rampur, Kalpa, Rakcham, Namgia and Kaza taken from Bhakra Beas Management Board (BBMB). Further, the observed daily discharge and sediment data at Rampur and Namgia were also taken from BBMB. The discharge and sediment data at Namgia site is used for point source input from the Tibet part of the basin as the study area is confined up to Indian national boundary. The Landuse/ Land cover map was prepared by using IRS P6 LISS-III satellite images at 23.5m spatial resolution. The NBSSLUP’s soil map and its Meta data were used to obtain the information of soil parameters such as soil type, its texture, hydrological group, initial soil moisture content, soil layer depth, soil particle density, bulk density, etc. To define the topographic settings of the basin and to prepare area elevation curves, the ASTER DEM with 30m spatial resolution is used, which is further classified into 10 homogenous elevation zones ranging from 900 to 6752 m. The Snow cover area needed in snowmelt computation was obtained using remote sensing data (MODIS data). vii In order to assess the SCA, VIC model is applied and results are verified by the limited remotely sensed snow data, i.e. MODIS snow cover product (MOD10A2) from 2001 to 2014. The results of the study show that SCA is highly variable in space and time. Its variability depends on various factors such as terrain features (elevation, slope and aspect) and meteorological parameters such as temperature, wind velocity and sunshine hours etc. It has been observed that the VIC simulated SCA is significantly correlated with MODIS SCA (R2 > 0.9). It was observed that the accumulation of the snow takes place from September to March, whereas depletion takes place from April to August in the region. The monthly average SCA was found to be maximum in the month of March (about 90% of total area of 16,650 km2), whereas, it was minimum in the month of August (about 12% of the total area), i.e. permanent snow or glaciers. The mean annual SCA of an individual year found to be maximum (about 74%) in the year 1989 and minimum (about 35%) in the year 1999. The study has suggested the accumulation and depletion curves and their best fit equations which can help in assessing the SCA in a particular month in the region. It is observed that variation in elevation has a major influence on the snow accumulation/ablation properties in the study area. It is found that SCA was maximum in the elevation range of 5100– 5700 m, which is about 67.3% of the total area of about 4463 km2 in this elevation range. Slope class of 20° to 30° was found to be associated with maximum SCA% while it was minimum in the slope class of 70°–80°. NE and E facing slope exhibit maximum SCA while SW and S directions exhibit minimum. This is due to the fact that south-facing aspects receive more solar radiations. In order to assess the SWE and SD, remote sensing data analysis and hydrological simulation using the VIC model have been used. In remote sensing approach, the SWE was assessed using AMSR-E/ Aqua Daily L3 Global SWE data for the period of 2002 to 2009. In hydrological modelling approach, the SWE and SD were assessed using the VIC hydrological model for the period of 1985 to 2014 on daily basis. The results of VIC model have been quantitatively compared and verified by AMSR-E SWE data and in situ measured snowfall data. It has been observed that SWE using VIC model overestimates but trend remains similar as that of the AMSR-E SWE. The underestimation of SWE values from AMSR-E is perhaps due to the penetration limit of passive microwave data and numerous uncertainties in SWE values retrieval from microwave radiometers, such as topographic characteristics and snow properties. The study also examined the SWE and its inter-annual and seasonal variation. The maximum mean annual SWE was observed to be about 474 mm in the year 2002, while it was minimum (164 mm) in year 2004. The monthly average SWE was found to be maximum in the month of viii March (440 mm) and minimum in the month of August (224 mm). The third order polynomial regression equation was suggested for depletion period and a linear regression equation for accumulation period for assessing the SWE in the region. Further, the study has identified the snow surplus and deficit years on the basis of mean annual SWE and its variation over the various years during the period of 1985 to 2014. Five years (1989, 1996, 2002, 2005 and 2013) were identified as excess snow years, whereas three years (1985, 1988 and 2001) as deficit snow years. The large-scale variability of snow parameters was observed from the seasonal to inter-annual time scales. The Mann–Kendall non-parametric trend analysis of snow parameters (SCA, SWE & SD) is performed for 30 years period and a significant decrease from 1985 to 2014 at 95% confidence level was observed. The variability in SD shows that depletion starts from March till August while it rises from September and continues till March next year. In March, SD is maximum (about 1000 mm), whereas, it is minimum in the month of August (about 326 mm). The mean annual SD of an individual year found to be maximum (about 1061 mm) in year 2002 and minimum (about 370 mm) in year 2012. The study has again suggested the accumulation and depletion curves and their best fit equations which can help in assessing the SD in a particular month in the region. The spatial and temporal variability in SCA, SWE and SD clearly reflect the impact of temperature variation in the study area. At the same time, the geological setup of the basin and effects of terrain attributes (Elevation, Slope and Aspect) on snow characteristics cannot be ignored. It is felt that VIC hydrological model can be an alternative for SCA, SWE and SD assessment, when remote sensing observation is not available. The average contribution of snow and glacier melt in the annual flow of Satluj River at Rampur was found to be about 70% and remaining 30% was from the rain. On the upstream of Rampur, the contribution of snow melt runoff in total runoff is higher than 70% due to higher percentage of snow covered area. In order to examine the climate change impacts on stream flow of Satluj river at Rampur gauge station, data of RCP4.5 and RCP8.5 are used. Both the scenarios (RCP4.5 and 8.5) indicate an increase in temperature of about 3.04 and 5.71 °C towards the end of this century (2099). A 14 to 21% increase in annual precipitation is predicted towards the end of the century. The flow change towards 2099 will predict to be increased by 11%–19% from the current average annual flow of about 333 m3/s for both the RCPs. Although the future flow projections shown here may not be accurate due to uncertainties in climate change scenarios. ix In order to study the sediment yield from the catchment, a discharge-sediment relationship is developed using the observed data. In the second approach, a GIS based SWAT model is applied to estimate the sediment yield from the catchment. The data from 1985 to 1997 is used for calibration while data from 1997 to 2005 is used for validation purposes. The mean annual sediment yield at Rampur site was estimated as 9.57 t/ha/yr against the mean annual observed sediment yield of 11.10 t/ha/yr.