Simulations under contemporary conditions

Figure 18.3 shows observed and calculated hydrographs for one hydrological year in the Abramov region, where the best results have been achieved. For the Tuyuksu area, the quality of measured discharge is insufficient (Hagg 2003), and at Glacier No. 1, the short dataset of only four hydrological years does not allow intercomparison through time.

Annual terms of the water balance in the three investigation areas, calculated by the HBV-ETH model, are shown in Table 18.3.

Abramov 1974/75

Measured

Simulated

Figure 18.3 Comparison between measured and simulated daily discharge in the Abramov region

Table 18.3 Annual terms of the water balance in the three investigation areas as determined by the HBV-ETH-model (Q = runoff, P = basin precipitation, ET = basin evapotranspiration, S = storage changes) in mm/a

Q

P

ET

Sglacier

Ssnow

Sground

Tuyuksu

968

904

155

-239

6

-15

(1981/82-1984/85)

Abramov

1348

1146

167

-460

57

34

(1968/69-1987/88)

Glacier No. 1

443

551

231

-135

39

(1986/87-1989/90)

Numerical parameters for the goodness of fit (Table 18.4) show satisfactory values. Especially in the Abramov region, the model efficiency criterion Reff (Q) following Nash & Sutcliffe (1970) indicates good results. At this test site, the data series is long enough to separate a calibration period, for which the parameter values are optimised, from a validation period, in which this parameter set is tested.

Worldwide testing of conceptual models (Rango 1992) has shown that Reff (Q)-values higher than 0.8 are above average for runoff modelling in glaciated catchments. Generally, comparison of Reff(Q)-values between different basins has to be regarded carefully, because this statistical item is strongly influenced by runoff variability, which may explain the relatively low values at Glacier No. 1, where runoff variability is highest, due to the small size of the catchment. Furthermore, the summer maximum of precipitation introduces uncertainty in the discrimination between rain and snow on this glacier. For the investigated catchments in the Alps, the threshold temperature is a critical model parameter only in relatively few cases in spring, autumn, and cool periods in summer, but the main input of snow occurs during winter, when nearly all precipitation is solid.

To evaluate the amount of glacial meltwater and the significance of glaciation for runoff variability, all the modelling was also carried out with the glacierised parts of the catchment areas set to zero. All other parameters were left unchanged so that these model runs represent current climate conditions but without glaciers as water storages. The fractions of glacier melt in total river runoff are shown in Table 18.5 and are discussed later. Annual runoff for current conditions and no-glacier-scenarios are given in Figure 18.4. As the data series available for Glacier No. 1 are too short, this test site was omitted from the runoff variability analysis.

A main characteristic of glacier runoff is the so-called compensating effect described by Rothlisberger & Lang (1987). Ice ablation is highest in hot, dry periods and lower during wet conditions, when clouds reduce radiation, and snowfall at higher elevations increases the albedo. Glacier melt and precipitation have a negative correlation, and therefore glaciers reduce the year-to-year variation of runoff, which is of great importance for water resources management. In this context, it can be observed that annual runoff of glaciated and glacier-free catchments shows a contrary behaviour (Rothlisberger & Lang 1987). Glaciated catchments reach their highest discharge values in hot summers, when glaciers show large areas of bare ice and deliver a high basic load of meltwater. Basins without glacier cover show maximum runoff volumes in wet and cool summers. This contrasting behaviour cannot be shown in Figure 18.4, the curves for the simulations with and without glacier cover run more or less parallel.

One should bear in mind that the simulations without glaciers assume a current climate in which glaciers still exist, although they are retreating. Therefore, the ratio between snow accumulation in winter and ablation in summer is higher than in effectively unglaciated, lower basins. This means that the model builds up an

Table 18.4 Goodness of fit for the three investigation areas (Qdiff = accumulated difference between measured and simulated discharge in mm/a [percentage of the measured value], Reff (Q) = Nash-Sutcliffe criterion)

Qdiffmean

Qdiffmin

Qdiffmax

mean

8-

max

Tuyuksu

77 [7.6%]

10 [1%]

185 [18.3%]

0.81

0.80

0.85

(1981/82-1984/85)

Glacier No. 1

113 [21.1%]

31 [7.8%]

191 [31.6%]

0.76

0.73

0.78

(1986/87-1989/90)

Abramov

226 [14.5%]

117 [8.3%]

539 [25.0%]

0.85

0.77

0.91

(1968/69-1977/78) calibration period Abramov

(1978/79-1987/88) validation period

0.83

0.70

0.91

(1968/69-1977/78) calibration period Abramov

(1978/79-1987/88) validation period

0.83

0.70

0.91

Annual runoff in the Tuyuksu region

Annual runoff in the Tuyuksu region

With glaciers Without glaciers

Annual runoff in the Abramov region

With glaciers Without glaciers

Annual runoff of Rofenache

With glaciers Without glaciers

Annual runoff of Vernagtbach

Annual runoff of Vernagtbach

With glaciers Without glaciers

Figure 18.4 Simulated annual runoff for present climate, with and without present-day glacier cover

With glaciers Without glaciers

Figure 18.4 Simulated annual runoff for present climate, with and without present-day glacier cover oversized snow pack that persists long into the ablation season and partly takes over the seasonal storage function of glaciers, which may explain why the curves are parallel instead of showing the expected contrary behaviour. In all cases, annual runoff decreases with the absence of glaciers, owing to the negative mass balances in the investigated periods. For comparison, the year-to-year variation of runoff should be related to the mean. This is accomplished by the coefficient of vari-standard deviation ation CV =- , which measures the mean relative dispersion.

If only July and August, the months with most intense ablation, are considered, the year-to-year variation of runoff is higher for the model runs without glaciers. Only at Vernagtbach, the coefficient of variation becomes somewhat smaller, if glaciation drops from 78 to 0%, but is relatively high in both cases. This is in good agreement with the theory of compensation, which states that the balancing effect is highest in basins with a moderate glacier cover, while runoff variation rises towards heavily glaciated and unglaciated catchments. According to Kasser (1959), the maximum compensation effect in the Alps is observed in basins with a glaciation of 30-40%. For the main ablation season, minimum variations of runoff are found in catchments with 30 -60% glacier cover (Rothlisberger & Lang 1987).

In Figure 18.5, the CV-values of the period July-August are plotted against glaciation, including the simulations without glaciers. The calculated trend shows a minimum year-to-year variation at degrees of glaciation between 20 and 50%.

0 0

Post a comment