HYDROLOGICAL MODELING OF HIGHLY GLACIERIZED RIVER BASINS Nina Omani, Raghavan Srinivasan, Patricia Smith, Raghupathy Karthikeyan, Gerald North
Problem statement Glaciers help to keep the earth cool High solar radiation reflection with albedo.8 (albedo ranges -1) Glaciers as reservoirs of fresh water Only 2% glacier cover provides more than 3% of total annual discharge in Indus river basin (Immerzeel et al., 21). Glacier compensation effect In European Alps during dry summer 23, glacier contribution was 12 times greater than normal, permitting normal hydroelectric production, whereas power production was significantly reduced elsewhere (Koboltschnig et al., 28).
Background Spatial variation of melt rates is affected by: topography (aspect, slope), ice or snow cover, climate type etc. SWAT model snow routing has been modified (SWAT212), so that the snow melt/accumulation parameters are allowed to be spatially variable in sub basins and elevation band scale Distributed temperature-index, allowing for spatially variable melt estimates (Cazorzi and Fontana, 1996; Hock, 1999; Daly et al., 2). 3
Objectives 1) Evaluation of SWAT s distributed snow algorithm in simulation of monthly runoff in glaciered basins 2) Extending the applied method to other river basins that are vary in climatic condition 4
Major steps 1) Identifying the vulnerable areas 2) Data collection 3) Modeling the glacierized area in SWAT 4) Model calibration and validation for pilot study areas vs. streamflow only vs. streamflow and glacier measurements 5) Extending the methodology to the other regions 5
Study Area
Precipitation/Snow depth (mm) Flow (cms) Temperature (C) Precipitation (mm) Temperature (C) Flow (cms) Precipitation (mm) Flow (cms) Temperature (C) Runoff regime of the basins 18 16 14 12 1 8 6 4 2 Upper Rhone (Switzerland) Precipitation Observed flow Temperature Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec 14 12 1 8 6 4 2-2 -4-6 -8 6 5 4 3 2 1 Narayani (Nepal) Precipitation at 1 m Precipitation at 3 m Observed flow Temperature Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec 14 12 1 8 6 4 2-2 -4 9 8 7 6 5 4 3 2 1 Vakhsh, Mendoza, Chile Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec 25 2 15 1 5-5 -1-15 Precipitation Snow depth 7 Observed flow (high glaciated) Temperature
ELA (m) Setting the climatic glacier boundary The climatic glacier boundary was set at some randomly selected subbasins Glacier climatic boundary is a synthetic line where accumulation = ablation Mass budget (Ba) is zero. Parameter setting: Precipitation laps rate: controls accumulation Temperature laps late: controls melt Snow melt temperature: controls melt y = -.21x + 2864 R² =.98 Gries glacier 29-3 36 34 32 3 28 26 24-3 -2-1 1 Ba (mm w. e.) ablation accumulation
Glacier modeling in SWAT 1 elevation bands with 1m to 25m intervals Glaciers accumulation boundary: Lowest boundary of climatic glaciation Infinite depth (1 km) GLIMS % Modeled Area% Narayani 1 11 Vakhsh 1 12 Upper Rhone 14 14 Mendoza - 4 Central Chile - 4
Model Calibration and Validation A calibrated model can give good runoff predictions for the wrong reasons. On glaciers there are extra possibilities: mass balance and equilibrium-line altitude, besides the discharge can help constrain the values of parameters. 1) streamflow only 2) streamflow and glacier measurements Monthly flow data at 35 gauging stations. Automatic calibration tool: SWAT-CUP (Abbaspour, 22) 1
Effect of melt parameter distribution on simulated flow (Massa-Blatten in Rhone) SMFMX varies between 2 to 8 mm d -1 ºC -1 thorough elevation bands Glacier free bands Below 29 m Glacierized bands Over 29 m 11
Flow (cms) 5 45 4 35 3 25 2 15 1 5 Massa-Blatten in Rhone Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec Average of Observed Average of Simulated Average of Simulated (modified) 12
Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec Flow (cms) Ba (mm w. e.) Rhone glacier (Gauged) Calibration 1: monthly flow (automatic calibration) Calibration 2: monthly flow + annual ELAs (198-1983) (manual parameter adjustments) Specific net mass balance 1 Rhone gauge station 1 Rhone glacier 8 6 Observed 5 4 2 Flow calibration Flow+ELA calibration -5 198 1981 1982 1983-1 13
Jul Aug Sep Oct Nov Dec Jan Feb Mar Apr May Jun Flow (cms) Jan-93 Sep-93 May-94 Jan-95 Sep-95 May-96 Jan-97 Sep-97 May-98 Jan-99 Sep-99 May- Jan-1 Sep-1 May-2 Jan-3 Sep-3 May-4 Jan-5 Sep-5 May-6 Jan-7 Sep-7 Flow (cms) Model accuracy in simulation of seasonal runoff was enhanced by setting the melt parameters for glaciers (high altitude elevation bands) and seasonal snow (low altitude elevation bands) separately. 1 8 6 4 2 Observed Lump model Modified model Mendoza (Colorado-Punta de Vacas) 5 4 3 2 1 14
Rhone Mendoza Narayani Calibration R 2 NSE PBIAS Validation lumped distributed lumped distributed lumped distributed R 2 NSE.81.83.81.83-6.6 +.5 - -.85.83.77.73 +1.5 +25.7 - -.88.88.85.87 +2.1 +8.6.79.78.91.89.69.7 +39.5 +33.8.78.7.76.78.7.77 +22.5 +5..8.71.66.65.32.59-15.8 +3.6.43 -.24.6.62.52.58 +7.2 +7.9.61.59.45.54.28.5-17.9 +5.5.57.46.82.85.76.83-24.7-13.2.86.81.81.81.75.74 +2.7 +5.5.76.61.86.91.85.91-7.3-1.5.86.82.95.95.86.95 +25.9 +2.2.95.89
Conclusions Treating the glacierized and glacier free areas separately, significantly improved the SWAT model performance in simulation of volume and seasonality of runoff in glacierized areas. Significance of meltwater is negligible where the melt season is coincidence with monsoon precipitation, so there was no significance different in the results by distributed model. This is due to dominance of rainfall-runoff model rather than the snow melt model. The results emphasized on the importance of combining even few information about the glaciers along with measured discharge data in model calibration.