Weather and Climate (1991) 11: 43-47 4 3 MODELLING PRESENT AND PAST SNOWLINE ALTITUDE AND SNOWFALLS ON THE REMARKABLES Introduction James R. F. Barringer Division of Land and Soil Sciences, DSIR A computer model simulating snowlines and snowfalls from readily available daily meteorological data has been used to obtain estimates of snowlines over the last 6 years on the Remarkables (Barringer, 1986). The model, based on extrapolation of temperature and precipitation from a nearby low altitude site (ie, Queenstown Airport or Queenstown), uses daily calculations of snow accumulation and ablation to estimate the mountain snowpack over a range of altitudes. Snow ablation is calculated using a simple degree-day melt equation, and accumulation is modelled using a complex rain/snow threshold calculation to give sensitivity to estimates of snowline altitude. The model relies upon the use of four temperature lapse rates for extrapolating temperatures to higher altitudes. Selection of temperature lapse rate type is made using a discriminant analysis of meteorological data from the low altitude base site. This offers a substantial improvement over temperature extrapolation techniques in other studies (eg. Moore & Owens, 1984a) and is crucial to the models ability to accurately simulate snowline altitude. Details of this modelling approach are outlined in Barringer (1986 and in press). The model was calibrated using photographs of the daily winter snowline in the study area during 1984 and 1985. The model was optimised to a best fit of the daily winter snowline data for each year, and this calibration tested against snowline data for the other year. Goodness of fit of modelled to observed data was variable, but in all cases acceptable. Scenarios Because the model can be used to estimate snowline and snowfalls from low altitude daily meteorological records, it is possible to use it in conjunction with scenario data (ie. real data modified to approximate a different climate) to assess the sensitivity of snowline altitude and snowfalls to changes in climate. The use of scenarios to assess the impacts of a climate different from the present does not constitute a prediction. Instead, scenarios "represent a systematic process that uses available theory, facts and judgments to explore the implications of hypothesised conditions" (Lave & Epple, 1985). Clearly the scenarios used in this study are considered to be plausible since there is considerable evidence that our climate is not static, and may be warming (Salinger, 1988), but they are entirely artificial and are subject to considerable uncertainty. This is particularly the case because most scenarios for changing climate are derived from Global Circulation Models (GCMs) which are difficult to apply at a regional or local scale. None of the scenarios used can be said to be any more probable than the others. In this case the implications for snowline altitude of a climate scenario (Salinger, pers. comm.) of cooler temperatures in the 185s (n 1 C cooler than at present) are considered. Cooler temperatures in the 193s are also hypothesised, being assumed to be.5 C cooler than present, and snowline altitude under these intermediate conditions estimated. The estimated snowlines for these scenarios are then compared with snowlines for warm climate scenarios.
44 S n o w l i n e Altitude Snowlines in the 185s and Today By using meteorological records from Queenstown Airport for the period 1972 to 1981 and altering the daily temperature data to match the 1 C cooler scenario for the 185s, the model indicates that the mean winter snowline would be approximately 133 m. Values ranging from 116 m to 142 m are estimated for individual years in this scenario "decade". These figures can be compared with a value for mean winter snowline of 1524 m calculated for the unaltered 1972 to 1981 period, indicating that snowline altitude may have been lower by as much as 19 m if the scenario for the 185s is correct. With respect to temperature this suggests that snowline altitude may change by as much as 19 m per 'C change in mean winter temperature. These figures can be compared with those obtained from similar analyses using other scenarios for climate differing from that being experienced today (Figure 1). These estimates of mean winter snowline altitude under a range of conditions provide us with the following values for change in snowline altitude relative to change in mean temperature. 1. Temperature and rainfall changing at same rate as for the period since 193 (ie. +.6 C by 25 with +1% precipitation) snowline altitude rises by E 1 in/"c rise in mean winter temperature. Calculated using modified 1972-1981 data. 2. +1.5 C by 25, no change in rainfall (Salinger & Hicks Scenario 1) - snowline altitude rises by 113 WC rise in mean winter temperature, calculated using modified 1972-1981 data. 3. +3. C by 25, no change in rainfall (Salinger & Hicks Scenario 2) - snowline altitude rises by -a 86 m/"c rise in mean winter temperature, calculated using modified 1972-1981 data. This figure is an underestimate because of the effect of undefined snowline where the snowline altitude is above the highest point on the range and hence defaults to the maximum altitude of 23 m. SnowlineAltitude(metres) 18-16 - 14-12 - 18 19 Fig. 1. A comparison of rates of mean winter snowline retreat for the 185 3 scenario( D ), the 193 A/ s c e n a r i o ( ), a / - 2 2 5 scenario( a ), / a n d a number of other scenarios for changed co c l i m a t e ( ). Also shown are snowlines estimated by daily rainfall and temperature data from Queenstown 193-1985 (4). 21 Years
Snowline Altitude 4 5 4. A s for 1. but calculated using 193-85 (ie. 55 years) data to estimate altitude gives the same estimate of rate of change but higher snowline altitudes. The analysis in this case is of poorer quality being based on temperature and rainfall data only. These analyses suggest a range of values from a minimum of 86 m to a maximum of 19 m rise in snowline altitude per degree centigrade rise in mean winter temperature. It is notable however, that all the analyses involving estimates for snowline altitude for conditions warmer than at present suggest snowline altitude would rise by about 1 m per degree centigrade, while the analysis of past climate scenarios gives the higher figure of 19 m per degree centigrade. This may well be a function of the temperature lapse rate profiles at this site which, because temperature inversions are common in winter, leads to lower average temperature lapse rates at altitudes below 12 m. With lower temperature lapse rates at lower altitudes a small change in temperature will lead to a large change in snowline altitude. Summer and Winter Snowline Altitudes The rates of snowline retreat relative to temperature increase indicated by the scenario based analysis above can be compared with figures derived by modelling present snowlines on a monthly basis from summer to winter. This allows us to consider snowline altitudes over a range of temperatures far greater than those anticipated by climate change scenarios for the next 5 years. This method also includes an attempt to account for changes in temperature lapse rate type and frequency in warmer conditions, since the scenario approach assumes similarity of temperature lapse rate under warmer conditions with those of present winters, while for the seasonal analysis, lapse profiles are calculated on a seasonal basis, and display significant differences between seasons. This analysis does assume a similarity of mean winter snowlines for possible warmer (or cooler) winters with mean monthly snowlines without allowing for the antecedent effects of snowline altitude in previous months. This effect is clearly seen in Figure 2 where a hysteresis effect is evident in the plotted data, with spring snowlines (ie. winter antecedent conditions) being lower than autumn snowlines (ie. summer antecedent conditions). Figure 2 shows data calculated for the 1972 to 1981 decade with a maximum snowline altitude of 35 m (cf. 23 m for the top of the Remarkables). The analysis has been repeated using a longer times series (ie. 5 years 193-1984), but of precipitation and temperature data only for Queenstown (Figure 3). The results of these two analyses are remarkably similar. With 1 years of full meteorological record from Frankton the rate of snowline change with respect to mean monthly temperature is calculated to be 11 mrc, and for the 5 year Queenstown record 16 mpc. Both these results are also remarkably similar to the estimates obtained from the scenario analyses in the previous section, which apart from the cooler scenario, gave rates of snowline change ranging from 9 mic to 113 mrc. Temperature and Precipitation - Snowline and Snowfalls When analysing the results of the model, it is interesting to correlate model outputs of mean winter snowline altitude and mean winter snow accumulation against mean winter temperatures and precipitation, to see whether simple relationships exist between these parameters and model output. There is a great deal of variability present in the input and output data,which means correlations are not always good. Nonetheless, there is generally a significant relationship between the amount of snow accumulation, particularly at higher altitudes, and
46 S n o w l i n e Altitude Fig. 2. A scatter plot of mean monthly snowline altitude and mean monthly temperature for the decade 1972 to 1981 (ie. using full meteorological records as model input data) indicates a good correlation between temperature and mean monthly snowline altitude (r2=.696). The regression line suggests that snowline altitude changes by 11 mrc. 4 o Summer (,) tu o 1:1 E 3.ot.- 3 CA Autumn Winter o Spring Ea a 3 ) D C l 4Po 4),a 1: o -1 Mean Monthly Temperature (degrees C) Queenstown Airport 1 2 Fig. 3 A scatter plot of mean monthly snowline altitude and mean monthly temperature based on model results for the period 193 to 1985 (le. using maximum and minimum temperature, and precipitation only as model inputs) indicates a good correlation between temperature and snowline altitude (r2=.717). The slope of the regression line suggests that snowline altitude changes by 16 mrc. 4 - (1),E *ma = 3 - - 1 2 Mean Monthly Temperature (degrees C) Queenstown
Snowline Altitude 4 7 precipitation. This relationship is obvious given that at higher altitudes temperatures do not exceed freezing very often, so that most precipitation falls as snow and temperature is a secondary effect. However, nearer the snowline this relationship is less clear and temperature more important. Just as we might expect precipitation to be the primary control of snow accumulation, mean winter temperature should be the primary control on mean winter snowline altitude, with precipitation a secondary factor. However, when these data are correlated the result is not significant. This is at least in part due to the complex relationship between temperature and precipitation that leads to any given snowline altitude, but it is also due to the relatively small range of mean 'winter temperatures (ie. -1-1.5 C). When we consider a wider range of temperature conditions, like those shown for mean monthly temperatures in Figures 2 and 3 (ie. -1-7.5 C), the temperature signal controlling snowline altitude shows clearly through the noise created by natural variability. Conclusions 1. Snowline altitude is complex to model, and displays a high degree of natural variability. 2. Temperature is the first order control on snowline altitude, but within the range of mean winter temperatures experienced over the last 3 to 5 years (ie. 4-1.5 C), variability of snowline caused by the interaction of temperature, precipitation and other factors has masked any trend of rising snowline. 3. Rates of change of snowline altitude with respect to temperature are about 1 mrc change in mean winter temperature, but may be greater under cooler conditions than the present, when cooler valley temperatures lead to more frequent and more intense valley inversions. References Barringer, JAR, 1986: Soil Erosion in relation to Snowline in the Remarkables, Central Otago. Unpub. MSc thesis, Geography Department, University of Otago, New Zealand. Barringer, J.R.F., in press: A Variable Lapse Rate Snowline Model for the Remarkables, Central Otago, New Zealand. N.Z. J. Hydro'. Lave, B.L. & Epple, D., 1985: Scenario Analysis In Kates, RM., Ausubel, J.H. & Berberian, M. (Eds), Climate Impact Assessment, John Wiley & Sons Ltd., 511-528. Moore, R.D. & Owens, LE, 1984a: Modelling Alpine Snow Accumulation and Ablation Using Daily Climate Observations. N.Z. J. Hydrol., 23(2), 73-83. Salinger, M.J., 1988: New Zealand Climate Change: Past and Present. In Climate Change: The New Zealand Response, Proceedings of a workshop held in Wellington, March 29-3, 1988, Ministry for the Environment, 17-24. Salinger, M.J. & Hicks, DAC, 1989: Regional Climate Change Scenarios. Unpublished working scenarios prepared for members of the Impacts Working Group, New Zealand Climate Change Programme, Ministry for the Environment.