Memorandum. Höfundur: Halldór Björnsson, Nikolai Nawri, Guðrún Elín Jóhannsdóttir and Davíð Egilson.

Transcription

1 EBV Memorandum Date: Title: Estimation of evaporation and precipitation in Iceland Höfundur: Halldór Björnsson, Nikolai Nawri, Guðrún Elín Jóhannsdóttir and Davíð Egilson. Ref: Introduction The Icelandic Meteorological Office (IMO) has at the request of Statistics Iceland (SI) estimated evaporation and precipitation for Iceland. The purpose to obtain a comprehensive overview of the water cycle in Iceland as the previous precipitation data is based on a statistical/dynamical model which was terminated in 2007, and evaporation data is very limited. The calculations were done using the Harmonie weather forecasting model and the results apply for all of Iceland, but were calculated explicitly for 74 municpalites, 50 densely populated urban areas and 32 water extraction points whith flow rates of at least 50 l/s. The results of the model calculations were compared against a 39 year time series of evaporation measured in Reykjavík and it was concluded that the evaporation estimated by the model is acceptable. This document has the following structure: In the next section the model, calculations and results are described. Following this, the analysis of the measured time series are discussed. In the penultimate section we compare the two and discuss whether the model calculations are an acceptable representation of reality. Finally, there is a section with a discussion of the results and concluding remarks. Model estimation The model used for the estimation of evaporation and precipitation is the Harmonie weather forcasting model1. This model is used by 23 (mostly European) meteorological services to provide regional weather forecasts. The model is used by the IMO to provide weather forecasts for Iceland and neighbouring oceanic areas. It is run with a grid resolution of 2.5 km, which is sufficient to resolve most, but not all, terrain features that affect the development of weather. The model version used is 38h1.2 which is the same as is used for IMO operational forecasting See See also Nawri, N. (2015) Impact of snow and cloud cover on the surface energy budget over Iceland based on HARMONIE model simulations Icelandic Meteorlogical Office, report /9

2 The regional model setup requires information about weather outside of the prediction domain. For Harmonie, IMO uses the results from the European Centre for Medium-Range Weather Forecasting (ECMWF). For the calculations done in this project the boundary data came from the ERAINTERIM reanalysis, which covers the time period 1979 to the present. The resolution of the boundary data is approximately 80 km which is far lower than the high resolution of the Harmonie model runs. The boundary date is provided to the model every 6 hours, but additionally, observations of temperature are assimilated into the model initial conditions, to ensure fidelity of the temperature field. The model data covers the years 1990 to SI was provided with both evaporation and precipitation data for the purpose of compiling relevant statistical tables. Furthermore, 2 m air temperature, relative humidity and10 m windspeed, total precipitation together with rain and snow separately, were provided. Figures 1 and 2 show maps of the annual evaporation and precipitation for the years and figures 3 and 4 show the annual cycle of evaporation and precipitation. Figure 1: Annual evaporation in Iceland for the period Units in mm. The map of annual evaporation (Figure 1) shows higher values in the lowlands on the periphery of Iceland. Lowest values are over glaciers and on mountaintops in the highlands. This is in stark contrast to the annual precipitation (Figure 2) where far higher values are observed over mountains 2/9

3 and glaciers. The precipitation map agrees quite well with earlier mapping results3, both in terms of total values and in terms of spatial distribution. As this is quite satisfactory, further comparison of the Harmonie precipitation field with observations and other data will not be needed here. Instead the focus will be on comparing the estimated evaporation with pan-evaporation mesurements. It should be noted, however, that when comparing figures 1 and 2 it is clear that the annual precipitation values far exceed the evaporation values. Figure 3 shows that in Iceland evaporation is far less during October to April than during other months. Values above 50 mm are hardly seen during the winter season, but during peak summer (June and July) the evaporation exceeds 60 mm over most of the low lying areas in Iceland. The seasonal cycle of precipitation (Figure 4) is quite different from that of evaporation. There is considerable precipitation during all months of the year, but substantially more during fall and winter, but the driest part of the year is in May and June. Figure 2: Annual precipitation in Iceland for the period Units in mm. 3 Crochet, P., T. Jóhannesson, T. Jónsson, O. Sigurðsson, H. Björnsson, F. Pálsson and I. Barstad (2007): Estimating the spatial distribution of precipitation in Iceland using a linear model of orographic precipitation. J. of Hydrometeorol., Vol. 8 (6), See also: 3/9

4 Figure 3: Annual cycle of evaporation in Iceland during the period Units are in mm. Figure 4: Annual cycle of precipitation in Iceland from Units in mm. 4/9

5 Measurements of pan evaporation Evaporation has been measured in Reykjavik during the warmest part the year (~May to October) from These data were logged by an observer once a day until 2007 when an automatic gauge was set up. Since the character of the automatic data is quite different than from the manual measurements the analysis here focused on the 39 year period 1968 to Precipitation is also mesured at the same location as the evaporation pan, allowing for corrections to evaporation on days with precipitation. Figure 5 shows the evaporation as measured during 4592 days with high quality evaporation data. It is apparent that the evaporation from the pan is strongly dependent on the strength of sunlight, peaking at summer solstice and diminishing throughout the summer from that point in time. Daily evaporation values range from about 1-6 mm with the highest values obtained during peak summer. The evaporation data was subjected to considerable analysis and it was determined that three factors were dominant in explaining the evaporation, the number of sunlight hours, the ambient humidity and cloudiness. Combined these factors explained more then 60% of the variance in the evaporation data4. 4 Guðrún Elín Jóhannsdóttir, Þórður Arason og Halldór Björnsson. (2016) Uppgufun í Reykjavík Veðurstofa Íslands, Reykjavík. 5/9

6 Figure 5: Daily average evaporation as measured from 1968 to 2006 in Reykjavik. The blue line show a smoothed curve through the data (GAM smoother with 95% confidence interval shaded) and the red line shows the annual cycle of sunlight in Reykjavik. The green vertical bar marks the position of the summer solstice (June 21st) but the evaporation peaks at that time. 6/9

7 Comparison of model data and pan evaporation The purpose of the analysis of the pan evaporation data was to be able to compare the data with the Harmonie model results. This comparison is complicated by a few factors. The evaporation pan is supposed to simulate evaporation from open water, but it has solid side walls that can heat up and enhance the evaporation. As a consequence, when comparing with open water surfaces, the pan data needs to be multiplied by a factor of 0.75 to correct for this5. Furthermore, at the location of the pan, the Harmonie model gridpoints are characterized as a city, surrounded by a typical heath-land and with the closest significant water body at some distance. For comparison the heath-land region was chosen, since this represents a common surface type in Iceland. Figure 6 shows the evaporation at this location. The daily values of evaporation produced by the Harmonie model at Hólmsheiði range from 0 to 4 mm. This is lower than the 1 to 6 mm daily evaporation values observed in the Reykjavík evaporation pan. However, the monthly median evaporation in the pan data for the month of June is about 2.5 mm, and when this is corrected for side wall heating the value obtained is only slightly above the median value for June at Hólmsheiði. This shows that the Harmonie evaporation is in a range that is similar to that expected from observations. A further comparison can be made between pan evaporation data and the model data, by examining the seasonal cycle of both. Figure 7 shows the surface energy fluxes at this location and the evaporation. By comparing the strength of the solar radiation and the latent heat flux it is easy to see that the latter peaks at summer solstice when the solar radiation is at maximum. The latent heat flux and evaporation are strongly coupled, as can be seen by comparison with the lower panel of the figure. As noted in the previous section, the pan evaporation data peaked at summer solstice and we can see similar behaviour in the Harmonie data. It should be noted that Harmonie can estimate evaporation when the ground is snow covered. This is marked as sublimation in the lower panel of figure 7, and reaches a peak during winter time. Analysis of the pan evaporation data revealed little correlation with wind strenght. This is puzzling since evaporation is in general held to be dependent on windspeed. In the analysis of the panevaporation data it was suggested that different results might be obtained during winter season when the sun was lower on the horizon. Indeed, in the Harmonie results the correlation between evaporation and wind during the warm season (April - September) is only 0.2, but it rises to 0.7 during the cold season. Thus the model results capture realistic effects that have not been measured, due to the fact that the evaporation pan is taken indoors as soon as night temperature become so low that the water in the pan freezes over. 5 Linacre, E. T. (1994) Estimating US class A pan-evaporation from few climate data. Water International, 19, /9

8 Figure 6: A boxplot showing the seasonal cycle of daily evaporation in Harmonie at a heath land location just outside Reykjavik (Hólmsheiði). The red horizontal lines show the median values for each month, the solid green line interpolates the monthly average values. Units are in mm. 8/9

9 Figure 7: Energy fluxes at the Hólmsheiði location in Harmonie (upper panel) and evaporation and sublimation (lower panel). For the energy fluxes the solar radiation (SW) and thermal radiation (LW) are taken as positive downwards, but the sensible and latent heat fluxes are positive upwards. Summary and conclusions The Harmonie model was used to estimate evaporation and precipitation over Iceland. The precipitation field obtained was similar to that previously calculated by other methods, but the evaporation field had not previously been estimated to this degree of detail. Comparison with pan evaporation data showed that the evaporation calculated by the model was in the correct range, and the seasonal cycle of evaporation estimated by the model agreed with observations. Overall, the model-data comparison is satisfactory and the model results can be used as a basis for freshwater statistics in Iceland. 9/9