ELECTRONIC SUPPORZTING MATERIAL
Migration by soaring or flapping: numerical atmospheric simulations reveal that turbulence kinetic energy dictates bee-eater flight mode
Nir Sapir, Nir Horvitz, Martin Wikelski, Roni Avissar, Yitzhak Mahrer,and Ran Nathan
RAMS VALIDATION
RAMS simulation output has been tested against measured data in different parts of the world, at different biomes and at various temporal and spatial scales [S1–S4]. Here, we tested the match between RAMS-simulated data and ground measurements from seven meteorological stations of the Israeli Meteorological Service (IMS) distributed throughout southern Israel, in relation to temperature, wind velocity and wind direction. We examined the relationships between the 10 min temperature and wind velocity data registered by the IMS and RAMS output using linear regressions after the data has been tested and found to be normally distributed. These analyses were made using SPSS 15.0 [S5]. We used the Mardia-Watson-Wheeler test to examine if wind direction measured by IMS and simulated by RAMS differed from each other. We also used the V-test to examine if the average mean angle of the difference between the datasets is different than 0° [S6]. Oriana 2.02e was used for circular statistical analyses [S7].
We illustrate in supporting figure 1 data from Hatzeva station collected during eight consecutive days in May 2005. RAMS-simulated temperature fitted the observational data exceptionally well (figure 1a; simulated temperature = 0.88 × measured temperature + 3.9, R2 = 0.83, p < 0.001). RAMS-simulated data overestimated temperatures below 32.5 °C, and underestimated temperatures above this value, yet the absolute difference between simulated and measured data in the range of temperatures measured during the study period was smaller than 2.2 °C. Wind velocity showed a higher degree of deviation from observed data (supporting figure 1b; measured wind velocity = 0.58 × observed wind velocity + 0.96, R2 = 0.52, p < 0.001). RAMS-simulated data overestimated wind velocity below 2.3 m s-1, and underestimated wind velocity above this value so that under extremely strong wind conditions of 10 m s-1, RAMS-simulated data were substantially lower (6.8 m s-1) than measured data. Measured and RAMS-simulated wind direction (Supporting figure 1c) had a nearly identical mean direction (288° and 290° for measured and RAMS-simulated data, respectively). Yet, these data differed statistically when the data were paired in time (Mardia-Watson-Wheeler test, n = 1102, W = 136.2, p < 0.001). The average difference between measured and RAMS-simulated data was 335°; hence RAMS simulated wind direction with an average of 25° counter clockwise deviation. We found that the average angle of the difference in wind direction was statistically different from 0° (V-test, V = 0.45, u = 21.0, p < 0.001). No TKE field measurements were available to validate RAMS-simulated TKE, yet RAMS has been shown to predict TKE very well in other studies (e.g., Chan [S8]).
SUPPORTING REFERENCES
1. Avissar, R., Eloranta, E. W., Gurer, K. & Tripoli, G. J. 1998 An evaluation of the large-eddy simulation option of the regional atmospheric modeling system in simulating a convective boundary layer: A FIFE case study. J. Atmos. Sci. 55, 1109-1130.
2. Avissar, R. & Pan, H. 2000 Simulations of the summer hydrometeorological processes of Lake Kinneret. J. Hydrometeorol. 1, 95-109.
3. Ter Maat, H. W., Hutjes, R. W. A., Ohba, R., Ueda, H., Bisselink, B. & Bauer, T. 2006 Meteorological impact assessment of possible large scale irrigation in Southwest Saudi Arabia. Glob. Planet. Change 54, 183-201.
4. Zhong, S. Y. & Whiteman, C. D. 2008 Downslope flows on a low-angle slope and their interactions with valley inversions. Part II: Numerical modeling. J. App. Meteorol. Climatol. 47, 2039-2057.
5. SPSS Inc. 2006 SPSS for Windows, release 15.0.1. Chicago: SPSS Inc.
6. Batschelet E. 1981 Circular statistics in biology. London: Academic Press.
7. Kovach Computing Services. 2007 Oriana, version 2.02e. Wales: Kovach Computing Services.
8. Chan, P. W. 2009 Atmospheric turbulence in complex terrain: verifying numerical model results with observations by remote-sensing instruments. Meteorol. Atmos. Physics 103, 145-157.
Supporting Table 1. Pearson correlation coefficients1 between TKE, temperature and tailwind speed at 100 and 500 m altitude, and at variable altitude matching bird altitude
TKE – temperature / 0.67*** / 0.62*** / 0.64***
TKE – tailwind / 0.07NS / 0.24*** / 0.47**
Temperature – tailwind / 0.30*** / 0.34*** / 0.27NS
1NS = not significance (p > 0.05), * = 0.05 > p > 0.01, ** = 0.01 > p > 0.001, *** = p < 0.001
Supporting Table 2. Results of multinominal logistic regressions at 500 m altitude with comparisons between flight modes for each variable1,2, after weighting the data negatively proportional to the sample size of each bird in the dataset
Factor / AIC / DAIC / Nagelkerke pseudo-R2 / p F-MTKE / p F-S
TKE / p M-S
TKE / p F-M
Temp. / p F-S
Temp. / p M-S
Temp. / p F-M
TWS / p F-S
TWS / p M-S
TWS / p overall model
TKE / 447.7 / 0.0 / 0.521 / *** / *** / NS / - / - / - / - / - / - / ***
TKE, Temp., TWS / 464.8 / 17.1 / 0.585 / *** / *** / NS / *** / NS / ** / ** / *** / NS / ***
TKE, Temp. / 475.1 / 27.4 / 0.560 / *** / *** / NS / *** / NS / ** / - / - / - / ***
TKE, TWS / 480.7 / 33.0 / 0.549 / *** / *** / NS / - / - / - / NS / ** / ** / ***
Temp., TWS / 615.6 / 167.9 / 0.242 / - / - / - / *** / *** / ** / NS / ** / NS / ***
Temp. / 621.1 / 173.4 / 0.215 / - / - / - / *** / *** / *** / - / - / - / ***
TWS / 653.4 / 205.7 / 0.117 / - / - / - / - / - / - / ** / *** / ** / ***
1TKE = Turbulence kinetic energy, Temp. = Temperature, TWS = Tailwind speed, F = Flapping, M = Mixed flight, S = Soaring-gliding
2NS = not significance (p > 0.05), * = 0.05 > p > 0.01, ** = 0.01 > p > 0.001, *** = p < 0.001
Supporting Table 3. Results of multinominal logistic regressions at 100 m altitude with comparisons between flight modes for each variable1,2
Factor / AIC / DAIC / Nagelkerke pseudo-R2 / p F-MTKE / p F-S
TKE / p M-S
TKE / p F-M
Temp. / p F-S
Temp. / p M-S
Temp. / p F-M
TWS / p F-S
TWS / p M-S
TWS / p overall model
TKE, Temp., TWS / 519.5 / 0.0 / 0.517 / *** / *** / NS / NS / NS / ** / * / ** / NS / ***
TKE, Temp. / 523.5 / 4.0 / 0.501 / *** / *** / NS / NS / * / *** / - / - / - / ***
TKE, TWS / 524.7 / 5.2 / 0.499 / *** / *** / *** / - / - / - / * / ** / * / ***
TKE / 534.1 / 14.6 / 0.472 / *** / *** / *** / - / - / - / - / - / - / ***
Temp. / 602.7 / 83.2 / 0.318 / - / - / - / *** / *** / *** / - / - / - / ***
Temp., TWS / 606.4 / 86.9 / 0.319 / - / - / - / *** / *** / *** / NS / NS / NS / ***
TWS / 646.3 / 126.6 / 0.046 / - / - / - / - / - / - / NS / *** / * / **
1TKE = Turbulence kinetic energy, Temp. = Temperature, TWS = Tailwind speed, F = Flapping, M = Mixed flight, S = Soaring-gliding
2NS = not significance (p > 0.05), * = 0.05 > p > 0.01, ** = 0.01 > p > 0.001, *** = p < 0.001
Supporting Table 4. Results of multinominal logistic regressions of data with altitudinal information with comparisons between flight modes for each variable1,2
TKE / p F-S
TKE / p M-S
TKE / p F-M
Temp. / p F-S
Temp. / p M-S
Temp. / p F-M
TWS / p F-S
TWS / p M-S
TWS / p overall model
TKE, TWS / 54.3 / 0.0 / 0.590 / NS / * / * / - / - / - / ** / NS / NS / ***
TKE, Temp., TWS / 57.3 / 3.0 / 0.607 / NS / NS / * / NS / NS / NS / ** / NS / NS / **
Temp., TWS / 59.9 / 5.6 / 0.481 / - / - / - / NS / NS / NS / * / NS / NS / **
TWS / 60.3 / 6.0 / 0.38 / - / - / - / - / - / - / ** / * / NS / **
TKE / 62.1 / 7.8 / 0.336 / NS / ** / * / - / - / - / - / - / - / **
TKE, Temp. / 66.0 / 11.7 / 0.336 / NS / * / NS / NS / NS / NS / - / - / - / *
Temp. / 68.7 / 14.4 / 0.143 / - / - / - / NS / NS / NS / - / - / - / NS
1TKE = Turbulence kinetic energy, Temp. = Temperature, TWS = Tailwind speed, F = Flapping, M = Mixed flight, S = Soaring-gliding
2NS = not significance (p > 0.05), * = 0.05 > p > 0.01, ** = 0.01 > p > 0.001, *** = p < 0.001
Supporting Figure 1. An example of RAMS-generated data (filled circles) in relation to meteorological measurements (line) in the Arava Valley (Hatzeva station 30°47'N 35°15'E) during eight days in spring 2005, including (a) temperature, (b) wind velocity and (c) wind direction data (left panels). For temperature and wind velocity, measured and RAMS-simulated data are presented with a linear fit (mid panels), and for all three variables a histogram of absolute deviations between measured and RAMS-simulated data is presented (right panels)
Supporting Figure 2. The relationships between meteorological variables simulated at 100 m above ground and bird flight mode. (a) TKE, (b) Temperature, (c) Tailwind. For statistical details see Supporting Table 3.
Supporting Figure 3. The relationships between meteorological variables bird flight mode with matching bird and atmospheric altitudinal data. (a) TKE, (b) Temperature, (c) Tailwind. For statistical details see Supporting Table 4.