Snow mapping and hydrological forecasting by airborne Y-ray spectrometry in northern Sweden

Transcription

1 Hydrological Applications of Remote Sensing and Remote Data Transmission (Proceedings of the Hamburg Symposium, August 1983). IAHS Publ. no Snow mapping and hydrological forecasting by airborne Y-ray spectrometry in northern Sweden STEN BERGSTRQM & MAJA BRANDT Swedish Meteorological and Hydrological Institute, Box 923, S-60119,Norrk'6ping, Sweden ABSTRACT The Kultsjon basin in northern Sweden has been the subject for detailed studies of the potential of airborne Y-ray spectrometry for snow mapping and hydrological forecasting since the spring of A brief introduction to the theory behind this technique is given. Results from the three melt seasons are presented, and problems and uncertainties are discussed. A verification against ground "truth" based on conventional snow courses is shown together with examples of the variability in the snow-accumulation pattern. Finally, it is shown how the data can be used for updating and improving more conventional forecasting models. Cartographie de la couche de neige et prévision hydrologiques a l'aide de la spectrométrie rayons-y, par voie aérienne RESUME Depuis le printemps 1980, le bassin du lac Kultsjon est minutieusement étudié pour connaître le potentiel de la spectrométrie rayons-y en vue de mesurer la couche de neige et prévoir les conditions hydrologiques. Une brève introduction décrivant la théorie de cette méthode et les résultats obtenus au cours des trois saisons printanières sont présentés et les problèmes et les incertitudes de la méthode sont discutés. En plus l'exactitude de cette méthode est prouvée à l'aide de la méthode traditionnelle, et les exemples prouvant les variations de la couche de neige sont donnés. Finalement, il est montré coent les données peuvent être employées pour améliorer les méthodes traditionnelles de prévisions hydrologiques. BACKGROUND Natural radioactive elements in the ground are the sources of gaa radiation from the surface. The attenuation of this radiation due to the water equivalent of the snowpack is the foundation of the use of Y-ray spectrometry for hydrological forecasting. Normally the twoflight method is used, i.e. one flight shortly before the beginning of snow accumulation to measure ground activity, and one when the snowpack is at its maximum. The theory and the technical aspects of Y-ray spectrometry were discussed in great detail at the Workshop on Remote Sensing of Snow and Soil Moisture by Nuclear Techniques arranged by the WMO in

2 422 Sten Bergstr'ôm & Maja Brandt (WMO, 1979), where a number of applications were also shown. The major problems, excluding the instrumentation, can be suarized as follows : (a) Navigation precision: particularly in areas with variable geology and radiation it is important to keep to the original flight lines, where ground activity under snow-free conditions is measured. (b) Wet areas: these reduce the ground activity and have to be excluded in the analysis. (c) Contribution from radon and cosmic radiation: the radon problem is not as important in a climate with good ventilation. (d) Low ground activity in relation to the snowpack: if the ground activity is low, the instrument will register very few counts, which will strongly affect the precision or make the analysis impossible. (e) Irregular snowpack: due to nonlinearity of the attenuation caused by snow, irregularities in the snowpack and bare patches will cause an underestimation of the snowpack. (f) Weather conditions: these may make flying at low altitudes hazardous and therefore delay the measurements. They may also affect the navigation precision and flight elevation. APPLICATION OF THE TECHNIQUE TO THE KULTSJÔN BASIN Since the spring of 1980, Y-ray spectrometry has been applied to the Kultsjon basin in upper Âselealven in northern Sweden in order to explore its potential for hydrological forecasting. The work has been carried out on contract from the Swedish Association of River Regulation Enterprises with support from Studsvik Energy Techniques. The Swedish Geological Survey has been responsible for instrumentation and flights, and the Swedish Meteorological and Hydrological Institute has acted as project leader. The instrument is the same as the one used for geological mapping and prospecting by the Swedish Geological Survey. It is installed in an Aero Coander 680 E airplane. Flying height is approximately 50 m above ground level. The energy band kev is used for the snow measurements. The project started with 14 flight lines covering the basin in a pattern that was possible to fly at very low altitude. These lines are shown in Fig.l. The total length of the lines is approximately km, and the basin area is 1109 km. The number of lines has been reduced, as some of them proved to be unsuitable as the project proceeded. The water equivalent of the snowpack was calculated by a moving average technique based on the number of pulses registered over a given distance. The dates for the flights were 27, 25 and 29 March for 1980, 1981 and 1982 respectively. VERIFICATION AGAINST SNOW COURSES In 1980 the airborne measurements were verified against a total of

3 Snow mapping and hydro logical forecasting 423 FIG.l The original set of flight lines covering the Kultsjôn basin. 17 km of conventional snow courses spread over eight short stretches with snow samples being taken every 50 m using snowtubes. The result in Fig.2, which was first presented by Nilsson et al. (1980), shows good agreement between the two methods except for one stretch (10b). This deviation is caused by a very irregular snowpack with rocks frequently penetrating the snow. Thus, it causes an underestimation of the total snowpack as discussed earlier. Gaa-measurement water equ Snow course water equ. FIG.2 Comparison between conventional snow courses and results from airborne Y-ray spectrometry. Each point represents approximately a 2 km flight line with one sample every 50 m (the numbers refer to the lines in Fig.l). REPRODUCTION OF A REGISTRATION In 1980 line number 1 was flown twice. The total snowpack estimates of water equivalent were 359 and 389, respectively, which is a deviation of 8%. Plottings of the registrations with 1000 m integration intervals and 500 m overlap are shown in Fig.3. As can be seen, local deviations are sometimes considerable. Some of the

4 424 Sten Bergstx'ôm S Maja Brandt 140- I Kultsjon Saxnds FIG.3 Example of reproduction of the snow registration on line 1 when flown twice in peaks are caused by weak ground activity due to wet areas. will automatically be excluded in future work. These SNOW ACCUMULATION AND ELEVATION The relation between snow accumulation, as recorded from the airborne measurements, and elevation above sea level is analysed in a joint plotting for lines 2, 8, and 16 north of Lake Kultsjon (Fig.4) and individually for line 1 (Fig.5) and line 9 (Fig.6) south of Lake Kultsjon. The results show that it is possible to determine a gradient or increase of snow accumulation below the timber line, while there is no significant relation between snow accumulation and elevation above this line. The latter is of course an effect of the strong redistribution of the snowpack above the timber line. When analysing Figs 4-6 it is important to bear in mind that there is also a west to east gradient in the snow accumulation pattern, which is superimposed on the effect of elevation. This is discussed in the following section. LARGE SCALE VARIATIONS OF THE SNOWPACK The west-east gradient in the snow accumulation pattern was analysed by registrations from lines 7 and 3. They are plotted separately for the three years in Figs 7 and 8, where the topography of the landscape is also shown schematically. The figures clearly show a strong west-east gradient for all three years, which is normal under the climatic conditions in northern Sweden, although the gradient is less pronounced in If data from all the lines are used, it is possible to map the snow conditions in the major part of the basin. This is shown in Fig.9 for the three years. The years differ considerably in regard

5 Snow mapping and hydrological forecasting 425 Water equivalent m as I t -H m a s I FIG.4 Relation between ground elevation above sea level and snowpack as registered on flight lines 2, 8 and 16 in 1981 and to total snow storage and its geographical distribution. In Table 1 the total estimated basin snow storage measured by airborne Y-spectrometry is compared to the snow storage calculated by a conceptual runoff model (the HBV-model, Bergstrbm, 1976), which is used regularly for forecasts of remaining inflow to Lake Kultsjon. Water equivalent 700 i m a s I FIG.5 Relation between ground elevation above sea level and snowpack as registered on flight line 1 in 1981.

6 426 Sten Bergstrôm & Maja Brandt Wafer equivalent 700 i A , m a s I FIG.6 Relation between ground elevation above sea level and snowpack as registered on flight line 9 in Water equivalent km FIG.7 The west-east gradients on the snow accumulation as registered on flight line 7. The table also contains the modelling errors when the model was run over the melt seasons with actual precipitation and temperature data. Most of these errors are likely to have their origin in the snow accumulation calculations of the model, which are based on precipitation observations at the three stations shown in Fig.l. The interpretation of Table 1 is that the two methods are TABLE 1 Comparison between basin snow storage measured by y-ray spectrometry and the HBV conceptual model in relation to the modelling error (expressed in water equivalent) HBV-model 350 y-ray spectrometry 353 Difference -3 Modelling error for the melt period

7 Snow mapping and hydrological forecasting 427 surprisingly close for 1980 and 1981, and that the disagreement for 1982 is of the same order of magnitude as the modelling error. This means that the use of the Y-method to support the conceptual model Water equivalent FIG.8 The west-east gradients of the snow accumulation as registered on flight line 3. Note that the 1982 plot is separated from 1980 and FIG.9 Schematic snow mapping in the Kultsjbn basin based on airborne Y-ray spectrometry. would have improved the forecasts. The reason for the improvement is simply that three precipitation stations (shown in Fig.l) are insufficient to cover the biased snow distribution in 1982 (see Fig.9). FUTURE WORK The project is planned to continue over two more melt periods before

8 428 Sten Bergstrom S Maja Brandt a final evaluation is made. In the spring of 1983 only half of the lines will be flown, but every line will be flown twice. Better and quicker routines for data processing will be developed to shorten the timelapse between measurement and forecast, and parts of lines, which have been shown to confuse the results, will be excluded. Efforts will be made to find the maximum snow-water equivalent, which can be observed with any certainty. REFERENCES Bergstrom, S. (1976) Development and application of a conceptual runoff model for Scandinavian catchments. Report no. RHO 7, The Swedish Meteorological and Hydrological Institute, Norrkôping. Nilsson, J., Landstrom, 0., Linden, A. & Melander, A. (1980) Matningar med flygburen gaastraleutrustning for avrinningsprognosverksamhet (Measurements by airborne Y-ray spectrometry for runoff forecasting). In: Problems in Water Power Exploitation (Proc. Nordic Hydrologie Conference 1980). UNGI Report no. 53, Uppsala University, Uppsala. WMO (1979) Workshop on Remote Sensing of Snow and Soil Moisture by Nuclear Techniques. Technical papers presented at the workshop convened by WMO and organized in cooperation with IAHS and the Norwegian National Coittee for Hydrology, Voss, Norway.