High-Altitude Wind Velocity at Sierra Negra and San Pedro Mártir

Transcription

1 Publications of the Astronomical Society of the Pacific, 117: , 005 January 004. The Astronomical Society of the Pacific. All rights reserved. Printed in U.S.A. High-Altitude Wind Velocity at Sierra Negra and San Pedro Mártir Esperanza Carrasco Instituto Nacional de Astrofísica, Óptica, y Electrónica, Luis Enrique Erro 1, Tonantzintla, Puebla 7840, Mexico; bec@inaoep.mx Remy Avila 1 Instituto de Astronomía, UNAM, México D.F., Mexico; r.avila@astrosmo.unam.mx and Alberto Carramiñana Instituto Nacional de Astrofísica, Óptica, y Electrónica, Luis Enrique Erro 1, Tonantzintla, Puebla 7840, Mexico; alberto@inaoep.mx Received 004 November 8; accepted 004 November 6; published 005 January 1 ABSTRACT. It has been proposed that the global circulation of atmospheric winds at 00 mbar can be used as a criterion to establish the suitability of a site for the development of adaptive optics techniques such as slow wave front corrugation correction. By using the NOAA NCEP/NCAR reanalysis database, we analyze the monthly average wind velocity at 00 mbar for a 16 yr period for two sites in Mexico: Sierra Negra and San Pedro Mártir. We compare the results with those obtained for Mauna Kea, Paranal, and La Silla, and with Maidanak in Uzbekistan and Gamsberg in Namibia. We show that for all the sites studied, there is a yearly wind speed modulation, and we model that modulation. Our results show that Sierra Negra and San Pedro Mártir are comparable to the best observatory sites, such as Mauna Kea, and are among the best sites to apply adaptive optics techniques. 1. INTRODUCTION 1.1. The Relevance of the High-Altitude Winds One of the peculiarities of the next generation of telescopes, also called extremely large telescopes, is their reliance on adaptive optics. Therefore, the selection of sites for new-generation telescopes requires reliable studies of the vertical turbulence profiles, represented by the refractive index structure constant CN and the velocity of the turbulence layers, V(h). The average velocity of the turbulence V 0, defined as function of CN and V(h), is a key parameter because other quantities relevant to adaptive optics are directly related to V 0 (Roddier et al. 198). Thus, if it is possible to obtain V 0 by independent and reliable means, the rest of the adaptive optics quantities can be estimated from V 0. Sarazin & Tokovinin (00) have shown that for two sites in the north of Chile Paranal, the site of the Very Large Telescope, and Cerro Pachón, the site of Gemini South the average velocity of the turbulence, V 0, is clearly related to the wind velocity at a 00 mbar pressure level. Assuming a standard atmosphere, 00 mbar corresponds to about 1 km above sea level. For their analysis, they include data from 35 balloon flights. They compare the V 0 obtained from the in situ measurements with those obtained at 00 mbar from the NOAA (National Oceanic and Atmospheric Administration) Global Gridded Up- 1 On leave from Centro de Radioastronomía y Astrofísica, UNAM, Morelia, Mexico. per Air database (GGUAS). The result is the linear relation V, with an m s 1 0 p max (v ground, 0.4v 00 mbar) rms p.6, and where v ground is the wind velocity at ground level. The authors conclude that with moderate wind at ground level and an equal seeing quality, existing and potential astronomical sites can be ranked in terms of suitability for adaptive optics simply by looking at the wind speed at 00 mbar. Hence, the first evaluation of a site under consideration for developing adaptive optics techniques could be the statistics of the wind speed at 00 mbar. Moreover, the use of meteorological databases for astronomical characterization offers the possibility of carrying out studies, under the same conditions, of different sites. Here we present monthly statistics of the 00 mbar wind velocity variation for the Mexican sites Sierra Negra and San Pedro Mártir. We compare the results with those obtained from Mauna Kea in Hawaii, Paranal and La Silla in Chile, Gamsberg in Namibia, and Maidanak in Uzbekistan. A detailed study about the Canary Islands observatories can be found in Chueca et al. (004). It must be emphasized that we are not presenting a full characterization of the sites. Nevertheless, we observe the quality of these sites with respect to high-altitude wind speeds as a very important parameter for adaptive optics. 1.. Sierra Negra and San Pedro Mártir s Sierra Negra is the site of the Large Millimeter Telescope (LMT/GTM), a 50 m antenna optimized for 1 3 mm obser- 104

2 WIND VELOCITY AT 00 mbar 105 TABLE 1 Geographic Coordinates Coordinates Latitude Longitude Closest Grid Points Latitude Longitude Elevation (m) Costa Rica Sierra Negra San Pedro Mártir Mauna Kea Paranal La Silla Gamsberg Maidanak vations, now under construction. It is located at north, west, at an altitude of 4580 m above sea level, in the state of Puebla, Mexico. Sierra Negra is in Pico de Orizaba National Park, about 100 km west of the coast of Veracruz on the Gulf of Mexico, and 300 km from the Pacific coast. The LMT is a binational project of Mexico and the United States, led by the Instituto Nacional de Astrofísica, Óptica, y Electrónica (INAOE) in Mexico and the University of Massachusetts at Amherst. With the development of the Sierra Negra site, INAOE started to characterize site quality for observations in other wavelengths. Because of its altitude and location, the site is intrinsically very dry (the opacity at 40 GHz goes down to 0.0; Meza et al. 003); therefore, conditions are likely to be favorable for near/mid-infrared observations. A median seeing of 0.73 was obtained by using a differential image motion monitor (DIMM) during 17 nights. The data span 000 February and 003 November (Carrasco et al 003a; Avilés 004). The meteorological conditions are quite mild for such a high altitude, and snowfall is generally light throughout the year. The global median temperature is 1. C, the diurnal cycle is typically only C, and the average temperature varies seasonally about 3 C. The global median wind velocity is 4 m s 1, the first and third quartile are. and 5.8 m s 1, respectively, with a diurnal median wind speed of typically 3.6 m s 1. The average wind speed varies seasonally less than m s 1 (Carrasco et al. 003b). San Pedro Mártir (SPM) is the site of the Observatorio Astronómico Nacional, located at north, west, at an altitude of 830 m above sea level, in the state of Baja California, Mexico. It is located about 60 km from the Pacific and 53 km from the Gulf of Cortés. The site, which is administrated by the Universidad Nacional Autónoma de México (UNAM), is within a national park. It has three optical telescopes that are, 1.5, and 0.84 m in diameter, respectively. A study carried out between 198 and 00 shows that the fraction of photometric nights was 64%, while the fraction of spectroscopic nights was 81% (Tapia 003). Furthermore, Tapia remarks that the fractional number of totally clear photometric nights increased considerably, from 54.% prior to 1996 to 74.6% in the next 7 yr. The median visual extinction is ky p 0.13 mag air mass 1 (Schuster et al. 00; Parrao & Schuster 003). Atmospheric turbulence profiles above the observatory have been obtained during several campaigns (Avila et al. 003, 004, 005 in preparation). A median seeing of 0.60 was obtained with a DIMM during 13 observing nights between 000 and 003 (Michel et al. 003a). The median wind speed is 3.9 m s 1 during the day and 5.3 m s 1 at night (Michel et al. 003b). In fact, the site characterization carried out over the years has proven that SPM is indeed a very good site for optical/nir observations, and is probably one of the best (Cruz-González et al. 003).. THE DATA The NOAA runs several climatological projects. Within the NOAA, the National Climatic Data Center (NCDC) is in charge of managing the collection of global in situ and remotely sensed climatological data and information. Atmospheric data are obtained using instrument packages that include radiosondes and rawinsondes carried by weather balloons that transmit the data back to a receiving station on the ground. The upper air data consist of temperature, relative humidity, atmospheric pressure, and wind speed. In a similar way, the goal of NOAA s Climate Diagnostics Center (CDC) is to identify the nature and causes of climate variations, on timescales ranging from a month to centuries. The Global Reanalysis Project at the CDC NCEP/NCAR (National Center for Environmental Prediction/National Center for Atmospheric Research) uses a state-of-the-art analysis and forecasting system to perform data assimilation using data from 1948 to the present. The NCEP s method is to use current and fixed (1995 January) versions of data assimilation and operational forecast models. The task of NCAR is to collect and organize many of the land and marine surface data archives, to provide these to NCEP, along with many observed satellite and aircraft observations, and to receive and store the output archives. The CDC s Derived NCEP Reanalysis products include over 80 different variables and several different coordinates systems at 0, 6, 1, and 18 forecasted values. In 005 PASP, 117:

3 106 CARRASCO, AVILA, & CARRAMIÑANA TABLE Monthly Average Wind Velocity at 00 mbar ( ) Costa Rica Sierra Negra SPM Mauna Kea Paranal Month Avg. rms Avg. rms Avg. rms Avg. rms Avg. rms Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec Average Note. Data from the NCEP/NCAR reanalysis database. Wind velocity is in units of m s 1. particular, the Derived NCEP Pressure Level product provides the monthly wind speed on a.5 global grid at 17 pressure levels, with about a 00 km horizontal resolution (Kistler et al. 001). We use the monthly wind velocity at 00 mbar on the basis of four records per day between 1980 and For our analysis, we used the NCEP/NCAR reanalysis database, because it provides a more accurate determination of the monthly average wind speed than the GGUAS data (used by Sarazin & Tokovinin [00]), and the fluctuations are more than a factor of larger for GGUAS than for the NCEP data (Carrasco & Sarazin 003). 3. RESULTS The coordinates of the sites in this study are shown in Table 1. Costa Rica (San José) is included as a tropical location that has no jet stream. Two sets of coordinates are given for each site. The first are the geographic coordinates used as input to the database; the next two columns correspond to the grid points closest to the geographical coordinates. The elevation in meters above sea level is included in the last column. In Tables and 3, the wind velocities obtained from the NCEP/NCAR reanalysis database are shown. The first column indicates the month. For each, site two values are reported: first, the wind speed obtained by averaging the monthly values over the 16 yr period, and second, the instantaneous rms fluctuations around the average. The annual average wind speed and the quadratic average of the monthly rms values are included at the bottom of the tables. The latter measure represents a typical fluctuation of the wind speed. Figures 1 and show plots of the monthly average wind speed at 00 mbar for the site indicated; the error bars are the rms fluctuations. The plots are ordered in increasing absolute value of the latitude and are grouped according to hemisphere: TABLE 3 Monthly Average Wind Velocity at 00 mbar ( ) La Silla Gamsberg Maidanak Month Avg. rms Avg. rms Avg. rms Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec Average Note. Data from the NCEP/NCAR reanalysis database. Wind velocity is in units of m s 1. Costa Rica, Sierra Negra, Mauna Kea, SPM, and Maidanak in the Northern hemisphere, and Gamsberg, Paranal, and La Silla in the Southern hemisphere. Costa Rica, which has no jet stream, is shown as a reference. For the northern sites, as the latitude increases, so does the average wind speed. Therefore, Sierra Negra presents the lowest average wind at 00 mbar of all the sites analyzed; next is Mauna Kea, SPM, and Maidanak. Analogously, in the Southern hemisphere, as the absolute value of the latitude increases, so does the average wind speed. Among the Southern sites, Gamsberg has the lowest annual average wind speed at 00 mbar. A seasonal trend of slower winds in the summer is seen for the Northern sites of Sierra Negra, Mauna Kea, and SPM, and to a lesser extent for Maidanak. Similarly, for the sites in the Southern hemisphere, the months with lower wind velocity at 00 mbar correspond to the Southern summer. Table 4 shows a summary of the annual average wind velocities. Here we include the uncertainty in the annual averages, given by the annual rms fluctuations divided by N, where N p 1. Sierra Negra is 3.5, 5.7, 1.6, 6.9, and 7.5 j below Mauna Kea, Maidanak, Gamsberg, Paranal, and La Silla, respectively. The average annual wind velocity for SPM is 1., 1.6, and.6 j below Maidanak, Paranal, and La Silla, respectively. On the other hand, SPM is 1 and 1. j, above Mauna Kea and Gamsberg, respectively. For Mauna Kea, the annual average wind velocity is.4,.6, and 4.1 j below Maidanak, Paranal, and La Silla, respectively. In contrast, Mauna Kea is 1.6 j above Gamsberg, giving a statistically tangible site ranking. Although annual averages have been computed for all sites, it can be shown that the wind speed is not a constant function of time. Considering that error-weighted averages minimize x, we tested the hypothesis of a constant wind speed v(t) p C c. The values obtained are shown in Table 5, where is the C c 005 PASP, 117:

4 WIND VELOCITY AT 00 mbar 107 Fig. 1. Monthly wind speed for the sites considered. Averages and rms deviations are indicated by the points and corresponding error bars. The solid lines indicate the best fit of a yearly modulated wind speed, v(t) p Cs B cos [p(t t 0)/1 yr]. Dotted lines indicate a constant wind speed fit, v(t) p Cc, shown only for the data that do not reject such a hypothesis. error-weighted average wind speed. The P 11 hypothesis rejection probabilities are indicated for each site. The hypothesis of a constant wind speed is rejected at the 99.9% confidence level for Sierra Negra, San Pedro Mártir, Mauna Kea, Paranal, and Gamsberg. We also tested the alternative hypothesis of a wind speed following v(t) p C B cos [p(t t )/1 yr], (1) s PASP, 117:

5 108 CARRASCO, AVILA, & CARRAMIÑANA Fig.. Monthly wind speed for the sites considered. Averages and rms deviations are indicated by the points and corresponding error bars. The solid lines indicate the best fit of a yearly modulated wind speed, v(t) p Cs B cos [p(t t 0)/1 yr]. Dotted lines indicate a constant wind speed fit, v(t) p Cc, shown only for the data that do not reject such hypothesis. with three fitting parameters: C s, the average wind speed; B, the amplitude of the wind speed modulation; and t 0, the time of the year of minimum wind speed. The values of these three parameters, determined through simultaneous x minimization, are shown in Table 6. The corresponding x probabilities are all compatible with the hypothesis, which in most cases provides a very good fit to the data. Minimum wind speed is reached between June and August in the northern sites, and six months earlier for the three southern sites (Paranal, La Silla, and Gamsberg). Even for the sites where a constant wind speed 005 PASP, 117:

6 WIND VELOCITY AT 00 mbar 109 TABLE 4 Annual Average Wind Speed at 00 mbar ( ) Velocity (m s 1 ) Costa Rica Sierra Negra SPM Mauna Kea Paranal La Silla Gamsberg Maidanak cannot be rejected (Costa Rica, La Silla, and Maidanak), a yearly modulation provides fits with lower x probabilities. Average 00 mbar wind speeds are only a broad indication of the site properties; the yearly modulation should be taken into account, as they can be as large as the average wind speed. We note that B C s for Sierra Negra and San Pedro Mártir. The fits are shown in Figures 1 and. 4. COMPARISON WITH GENERALIZED SCIDAR MEASUREMENTS AT SPM In 000 May, measurements of the turbulence profiles C N(h) and the velocity of the turbulent layers V(h) t were performed during 16 nights at SPM using the generalized scidar (GS) of the Laboratoire Universitaire d Astrophysique de Nice, France. A description of the measurements and the results can be found in Avila et al. (003, 004, 005 in preparation). The interest here is twofold: first, the comparison of the values of v 00 mbar at SPM for May, obtained from the NCEP/ NCAR reanalysis database, with the values of the velocities measured with the GS between 10 and 15 km above sea level ( v ). Second, the comparison of the values of V 0 obtained from the empirical relation proposed by Sarazin & Tokovinin (00) with the actual values calculated from the CN and Vt profiles measured at SPM ( V 0, GS ). It is worth mentioning that the GS profiles used in this calculation do not include turbulence inside the telescope dome. As shown in Table, for May, the mean and rms values of v are 8.7 and 7.1 m s 1 00 mbar. The corresponding values of v are 6.7 and 1.6 m s The number of v10 15 values used for these estimations is The agreement of both estimates is remarkable. For each simultaneous measurement of C N(h) and V(h) t, one value of is calculated: V 0, GS 5/3 3/5 dhfv(h)f C (h)] t N dh C N(h) V p. () 0, GS [ This computation was performed for the 3016 measured profiles. The mean and rms values of are 1. and 5.7 m s 1, which V 0, GS TABLE 5 Error-Weighted Average Wind Speeds and s Under Consideration C c (m s 1 ) x x for P 11( x ) Costa Rica Sierra Negra * San Pedro Mártir * Mauna Kea * Paranal * La Silla Gamsberg * Maidanak Note. P refers to the x 11 with an 11 degrees-offreedom (1 points, 1 fit parameter) cumulative probability function. Asterisks indicate the rejection of a constant wind speed hypothesis above a 99% confidence level. agrees perfectly with V m s 1 0 p 0.4v00 mbar p This simple comparison supports the empirical relation proposed by Sarazin & Tokovinin (00). 5. CONCLUSIONS We have analyzed seasonal variations of the monthly average wind velocity at 00 mbar for the period between 1980 and 1995 for Sierra Negra, SPM, Mauna Kea, Paranal, La Silla, Gamsberg, Maidanak, and Costa Rica (the latter used as a reference). We used the monthly average data, based on four records per day, provided by the NOAA NCEP/NCAR reanalysis data set. We find that Sierra Negra has the lowest yearly average wind velocity of the sites analyzed. The wind speed at 00 mbar at SPM is comparable to that of Mauna Kea, and in some months probably lower. SPM annual average velocity is lower than for Paranal, La Silla, and Maidanak. Following the criterion that the average velocity of turbulence V 0 is linearly related to the wind velocity at a 00 mbar pressure level, we conclude that Sierra Negra and SPM are among the best ob- TABLE 6 Error-Weighted Average Wind Speeds and Under Consideration C s (m s 1 ) B (m s 1 ) x t 0 (month) for s x P 9( x ) Costa Rica Sierra Negra San Pedro Mártir Mauna Kea Paranal La Silla Gamsberg Maidanak Note. P refers to the x 9 with a 9 degrees-of-freedom (1 points, 3 fit parameter) cumulative probability function. All data conform the hypothesis of a yearly modulated wind speed. 005 PASP, 117:

7 110 CARRASCO, AVILA, & CARRAMIÑANA servatory sites for adaptive optics techniques. Moreover, we also find that for all the sites analyzed, there is an annual wind speed modulation, and we model that modulation. In particular, the amplitude of the wind speed modulation is comparable to the mean wind speed for Sierra Negra and SPM. We consider that this yearly wind modulation should be be taken into account when comparing the quality of different sites. In the case of SPM, by analyzing V0 and v10 15 data obtained with a GS during a 16 day campaign, we found the same relation V0 p 0.4v00 mbar proposed by Sarazin & Tokovinin (00) for Paranal and Cerro Pachón. The measured wind velocity v10 15 is consistent with the v00 mbar value from the NCEP/ NCAR reanalysis database. NCEP/NCAR reanalysis data provided by the NOAA-CI- RES Climate Diagnostics Center, Boulder, Colorado, from their Web site. R. Avila acknowledges the support of DGAPA, UNAM, through grant IN The NOAA-CIRES Climate Diagnostics Center Web page is cdc.noaa.gov. REFERENCES Avila, R., Masciadri, E., Vernin, J., & Sánchez, L. J. 004, PASP, 116, 68 Avila, R., et al. 003, in Rev. Mexicana Astron. Astrofis. Ser. Conf. 19, San Pedro Mártir: Astronomical Evaluation, ed. I. Cruz- González, R. Avila, & M. Tapia (Mexico City: UNAM), 11 Avilés, J. L. 004, M.Sc. thesis, Instituto Nacional de Astrofisica, Optica, y Electronica Carrasco, E., Carramiñana, A., Avilés, J. L., & Yam, O. 003a, PASP, 115, 879 Carrasco, E., Carramiñana, A., Avilés, J. L., Yam, O., & Luna, F. 003b, INAOE Tech. Rep. (RT0 548; Puebla, Mexico: INAOE) Carrasco, E., & Sarazin, M. 003, in Rev. Mexicana Astron. Astrofis. Ser. Conf. 19, San Pedro Mártir: Astronomical Evaluation, ed. I. Cruz-González, R. Avila, & M. Tapia (Mexico City: UNAM), 103 Chueca, S., García-Lorenzo, B., Muñoz-Tuñon, C., & Fuensalida, J. J. 004, MNRAS, 349, 67 Cruz-González, I., Avila, R., & Tapia, M., ed. 003, in Rev. Mexicana Astron. Astrofis. Ser. Conf. 19, San Pedro Mártir: Astronomical Evaluation, ed. I. Cruz-González, R. Avila, & M. Tapia (Mexico City: UNAM) Kistler, R., et al. 001 Bull. Am. Meteor. Soc., 8, 68 Meza, J., Torres, A., Aguilar, L., & Gutiérrez, C. GTM/LMT Tech. Rep. Michel, R., Echevarría, J., Costero, R., Harris, O., Magallón, J., & Escalante, K. 003a, Rev. Mex. AA, 39, 91 Michel, R., Hiriart, D., & Chapela, A. 003b, in Rev. Mexicana Astron. Astrofis. Ser. Conf. 19, San Pedro Mártir: Astronomical Evaluation, ed. I. Cruz-González, R. Avila, & M. Tapia (Mexico City: UNAM), 99 Parrao L., & Schuster, W. J. 003, in Rev. Mexicana Astron. Astrofis. Ser. Conf. 19, San Pedro Mártir: Astronomical Evaluation, ed. I. Cruz-González, R. Avila, & M. Tapia (Mexico City: UNAM), 81 Sarazin, M., & Tokovinin, A. 00, in ESO Conf. Proc. 58, Beyond Conventional Adaptive Optics, eds. R. Raggazoni, N. Hubin, & S. Esposito (Garching: ESO), 31 Schuster, W. J., Parrao, L., & Guichard, J. 00, J. Astron. Data, 8 Roddier, F., Gilli, J. M., & Lund, G., 198, J. Opt., 13, 63 Tapia, M. 003, in Rev. Mexicana Astron. Astrofis. Ser. Conf. 19, San Pedro Mártir: Astronomical Evaluation, ed. I. Cruz-González, R. Avila, & M. Tapia (Mexico City: UNAM), PASP, 117: