astro-ph0304229/ms.tex
1: %%%%%%%%%%%%%%%%%%%%%%%%% PREAMBLE %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
2: %\documentclass[manuscript]{aastex}
3: %\documentclass[preprint2]{aastex}
4: \documentclass[12pt,preprint]{aastex}
5: \shorttitle{Seeing at Sierra Negra}
6: \shortauthors{Carrasco et al.}
7: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
8: \begin{document}
9: \title{\bf Optical Seeing at Sierra Negra}
10: 
11: \author{Esperanza Carrasco, Alberto Carrami\~nana,\\
12: Jos\'e Luis Avil\'es \& Omar Yam}
13: \affil{Instituto Nacional de  Astrof\'{\i}sica, \'Optica y Electr\'onica,\\ 
14: Luis Enrique Erro 1, Tonantzintla, Puebla 72840, M\'exico}
15: \email{bec@inaoep.mx}
16: 
17: \keywords{site-testing, atmospheric effects}
18: 
19: \slugcomment{February 25, 2003}
20: 
21: %---------------%
22: \begin{abstract}
23: %---------------%
24: Optical seeing measurements carried out at Sierra Negra, the site of the Large Millimeter
25: Telescope, are reported. The site, one of the highest peaks of Central Mexico, offers good 
26: coverage of Northern and Southern hemispheres and we have undertaken several campaigns to 
27: investigate the astronomical potential of the site in the optical. Here we report on our 
28: campaign aimed at establishing the seeing quality of the site. We present data of the first 
29: three campaigns of optical seeing monitoring covering from February 2000 to May 2002, carried 
30: out with a Differential Image Motion Monitor. The results clearly indicate a sub-arcsec seeing, 
31: better statistics during the dry season and no dependence with the time of night. We find
32: no dependence of our results with the integration time used.
33: %---------------%
34: \end{abstract}
35: %---------------%
36: 
37: %%%%%%%%%%%%%%%%%%%%%%
38: \section{Introduction}
39: %%%%%%%%%%%%%%%%%%%%%%
40: Sierra Negra is an extinct volcano in the State of Puebla, Mexico, located at 
41: 18$^{\circ}$~59$^{'}06^{"}$~N latitude, 97$^{\circ}$~18$^{'}53^{"}$~W longitude and 
42: an altitude of 4580~m above sea level. The mountain is next to Citlaltepetl, the highest 
43: peak in Mexico, with just 8~km separating both summits. Sierra Negra is about 100~kms West 
44: from the coast of Veracruz in the Gulf of Mexico and 300~kms from the Pacific Coast. The 
45: site, administrated by the Instituto Nacional de Astrof\'{\i}sica, \'Optica y Electr\'onica 
46: (INAOE), is inside the Pico de Orizaba National Park. Easy access is available via 100~km 
47: motorway from the city of Puebla followed by a 20~km  access road to the summit. The 
48: journey from Puebla city takes about two hours. 
49: 
50: In February 1996  Sierra Negra was selected, among more than twenty potential sites, as the 
51: site of the Large Millimeter Telescope (LMT/GTM), now  under construction. The decision was 
52: based on its low atmospheric water vapour content, with registered opacities at 240~GHz down
53: to $\la 0.02$. The LMT/GTM is a 50~m antenna optimised for 1-3~mm observations. First light 
54: and  science operations are planned for  2005. The LMT/GTM is a bi-national project between 
55: Mexico and the United States, leaded by INAOE in Mexico and the University of Massachusetts, 
56: at Amherst, in the USA. 
57: 
58: With the development of the Sierra Negra site, INAOE planned to measure its quality for 
59: optical observations. Because of its altitude and location the site is intrinsically very 
60: dry, therefore the conditions are likely to be favorable for near infrared and optical
61: observations. An unknown property of the site is its optical seeing, a key parameter to 
62: determine how good an astronomical site is. In the last few decades great efforts have been 
63: dedicated to the development of 8-10~m class diameter telescopes and for instrumentation 
64: which requires very precise site characterization. Furthermore, the new generation of 
65: Extremely Large Telescopes will require selecting sites with very good seeing conditions. 
66: 
67: INAOE decided to undertake a first optical seeing and weather measurements campaign starting 
68: in February 2000 without basic facilities available. A temporary set-up was prepared for the 
69: February campaign, which was in fact the first astronomical night-time work performed at the 
70: site. The second campaign started in October 2000, with better facilities such as 5~m tower, 
71: a container, a suitable power supply and a place to rest -at about 3000~m above sea level- 
72: after observing. From  May 2001 we started a routine measurements regime.  Here we present 
73: the results obtained from February 2000 to May 2002.  Weather data have been taken almost 
74: continuously from November 2000 up to date. The weather analysis will be reported elsewhere. 
75: 
76: %%%%%%%%%%%%%%%%%%%%%%%%%
77: \section{The Instrument}
78: %%%%%%%%%%%%%%%%%%%%%%%%%
79: The data for this work were taken with a Differential Image Motion Monitor  (DA/IAC DIMM) 
80: developed  by Vernin \& Mu\~noz-Tu\~non (1995), based on the same physical principle 
81: as the ESO DIMM (Sarazin \& Roddier 1990) but commercialized by the French company LHESA.
82: The DIMM principle is to produce twin images of a star with the same telescope via 
83: two entrance pupils and a wedge. The instrument consists of a 20~cm  Celestron telescope, on 
84: a very robust equatorial mount, with an intensified CCD camera coupled via an optical fiber 
85: bundle to the CCD, a Matrox frame grabber board  and a PC. The two $D = 60\rm mm$ apertures, 
86: separated by a distance $d = 140\rm mm$, are located on a mask attached to the 
87: telescope entrance pupil. A precisely cut wedge placed over one of the pupils 
88: deviates the incoming light separating the two star images by approximately 30~arcsec.
89: The intensified CCD camera and the frame grabber register the relative  position of both 
90: stellar images after computing the centroid position of each. A statistical seeing value is 
91: assessed based on the variance of the differential image motion after 200 images are taken. 
92: The measurement corresponds roughly to a wavelength $\lambda = 0.5~\mu$m, as dictated by the 
93: response of the system.
94: 
95: %%%% begins comment 1:
96: 
97: 
98: Because it is a differential method the technique is, in principle, insensitive to erratic 
99: motions of the telescope introduced by wind or ground vibrations. Sarazin \& Roddier (1990) 
100: showed that, assuming a Kolmogorov power-law spectrum for the turbulent cells, the 
101: longitudinal and transverse variances of the differential motion between 
102: the images, $\sigma_{l}$ and $\sigma_{t}$, are related to the Fried parameter $r_0$ as:
103: 
104: 
105: %
106: \begin{eqnarray} 
107: \sigma{_{l}}^{2} &=& 2\lambda^{2}{r_o}^{-5/3}[0.179D^{-1/3}-0.097 d^{-1/3}]\\
108: \sigma{_{t}}^{2} &=& 2\lambda^{2}{r_o}^{-5/3}[0.179D^{-1/3}-0.145 d^{-1/3}] .
109: \end{eqnarray}
110: %
111: Two independent $r_0$ values are obtained which, in principle, should have the same value.
112: The parameter $r_0$ can be imagined as the telescope diameter that would produce a 
113: diffraction spot of the same size as that produced by the atmospheric turbulence.  
114: The seeing is given by $s_{FWHM}= 0.98~(\lambda/r_0)$. These computations are carried 
115: out internally by the instrument, providing measures of $s \equiv s_{FWHM}$ derived from 
116: the longitudinal and transverse estimates. The DA/IAC DIMM achieves an accuracy better 
117: than 0.1" for stars brighter than fourth magnitude with a 10~ms time exposure. 
118: A reliable seeing measurement is attained within twenty seconds.
119: 
120: 
121: %%%%  Ends comment 1
122: 
123: 
124: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
125: \section{Observations and Data}
126: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
127: The seeing monitor location  at the summit is shown in figure \ref{layout}. The LMT is 
128: located to the  left. The circle corresponding to the LMT track is marked and it has 40 
129: meters of diameter. The approximate location of the seeing monitor and weather station is 
130: marked by the black square, to the North-East of the LMT. The  DIMM and the weather station 
131: are on a five meter height tower  to avoid the surface layer (Vernin \& Mu\~noz-Tu\~non 1994). 
132: The DIMM is on a tower independent of the platform where the observers move freely without 
133: affecting the seeing measurement. The tower is near a sharp edge to face directly the 
134: incoming winds. Thermal equilibrium is ensured by the absence of an enclosure.
135: 
136: The data available cover 85 nights, grouped in three sets:
137: 4 nights between February 22$^{nd}$  and  April 7$^{th}$, 2000,  corresponding to the first 
138: campaign; a second one with 10 observing nights between October 23$^{rd}$ and December 
139: 13$^{th}$ 2001; and the third campaign which consisted of 71 nights from May 24$^{th}$, 2001 
140: to May 3$^{rd}$, 2002. We observed bright stars, almost always $m_{V}\lesssim 2.5$.
141: 
142: The high-altitude and precarious initial development of the site made the initial runs a 
143: real challenge. Nevertheless we successfully carried out measurements,
144: for the first time, during the night at the summit.  
145: To compare all measurements, we present  statistics giving equal weight to 
146: every data point.  
147: For the analysis we only consider data files with at least 20 points acquired
148: close to the zenith (airmass$\leq 1.15$) with non-saturated images (DIMM parameter 
149: pixmax $\leq 255$) to ensure a reliable stellar centroid determination.
150: 
151: \subsection{Results on seeing statistics}
152: %------------------------------------------%
153: Figure~\ref{s_daily} is the daily plot of all the measurements. The dots are the median and 
154: the error bars go from the first to the third  quartile. The central dotted line denotes the
155: seeing median for the all data set, corresponding to 0.78". The top dotted line is
156: the distribution third quartile, 1.05" and the bottom dotted line is  the first quartile, 
157: corresponding to 0.62". The histogram and the cumulative distribution of the 
158: same data are shown in figure \ref{s_distr_tot}. It can be appreciated a
159: sub-arcsec seeing 75\% of the time.
160: 
161: To investigate the seeing seasonal behaviour we define the dry season from November to
162: April and the wet season from May to October. The histogram and cumulative distribution
163: for the dry and the wet seasons are shown in figures ~\ref{dry_season} and 
164: \ref{wet_season} respectively. While the seeing median during the dry season is 0.75",
165: for the wet season it raises 0.92". However it  should be noticed that during August 2001 
166: the seeing was specially bad (with a median of 1.49"; see table~\ref{table1}). To study 
167: the August contribution to the seeing we made the exercise of calculating statistics 
168: for the wet season without August 2001, the seeing median becoming 0.78", significantly 
169: closer to the dry season seeing median.
170: 
171: To compare the monthly seeing behaviour we present, in table~\ref{table1}, statistics giving 
172: equal weight to each individual seeing measurement for the complete data set: the first column 
173: is the month, the next
174: three columns give the details of the data acquired {\em i.e.} number of nights, number of
175: observing  hours during those nights and number of points. The next columns give the data 
176: statistics: mean, standard deviation, minumuum value, first quartile, median and third 
177: quartile. The global statistics shows that for 85 observing nights spanning from 
178: February 2000 to May 2002, the  seeing median is 0.78" with a standard deviation of 0.44". 
179: 
180: In figure~\ref{shourly} the distribution of seeing values as a function of  UT is shown for 
181: the 85 observing nights. The dots represents the median for each hour and the error bars go 
182: from the first to the third  quartile. The histogram is shown in the upper panel. It must the 
183: noted that the first bin, corresponding to 7-8 PM local time, has only a few points so the 
184: high seeing value might be due to low number statistics rather than a intrinsically higher 
185: seeing at the beginning of the night. We conclude that our data do not show any systematic 
186: trend along the night. \cite{mvv97} observed that there is no general trend in the seeing 
187: evolution for the Roque de los Muchachos Observatory. In contrast~\cite{gio01}, point out 
188: %% begins comment 2:
189: that for high altitude cordillera sites, in northern Chile,  the seeing  tends to be of lower
190: quality at the beginning of  the evening.
191: %% ends comment 2:
192: 
193: 
194: \subsection{The seeing integration time}
195: %------------------------------------------%
196: It has been discussed by several authors that the temporal averaging of the variance of the 
197: differential motion with a finite exposure time depends on the average velocity and the 
198: direction of displacement of
199: the wavefront corrugation with respect to the DIMM aperture~\citep{ma87,zt97}.  
200: \citet{gio01}, on high altitude  cordillera sites in Chile, measure the seeing 
201: alternating 10~msec and 20~msec exposures. 
202: They obtain seeing estimates derived from 10~msec exposures, from 20~msec exposures and extrapolations to 
203: ``zero exposure", obtained by multiplying the 10~msec seeing by the ratio of the 10 and  20~msec measurements. 
204: They find that the median values of the 0 ms seeing vary between 0.66" and 0.76", those of the 10~msec seeing 
205: between 0.56" and 0.65" while those of the 20~msec between 0.48" and 0.56". According to these
206: authors, the seeing for the 10~msec series is statistically worse than that for the 20~msec 
207: series, as the latter smears the image motion somewhat.   
208: 
209: 
210: 
211: 
212: Our DIMM has a default integration time of 20~msec. The  first data set that spans from 
213: February 2000 to August 2001 were taken using that integration time. 
214: During the next 12 observing nights we alternated measurements with 10 and 20~msec 
215: integration times. The camera control allows the user to alternate between the default time 
216: and a mechanically selected integration time. As the selection is manual we decided to take 
217: 15~integrations at each integration time. The results are shown in Figure \ref{alternados_1} 
218: where the seeing as a function of time is shown for each night. The filled circles 
219: corresponds to the medians of the 10~msec integration samples while the open circles to 
220: the medians of the 20~msec samples. Qualitatively, there is no significant difference between 
221: the trends of both data sets. We compare quantitatively the samples by plotting the 10~msec 
222: seeing medians {\em vs.} 20~msec seeing medians where each data set has been interpolated
223: through a spline fit to overlap in time the other data set, such that at each time we
224: have one data point (either from the 10~msec or 20~msec sample) and one spline interpolation
225: (from the 20~msec or 10~msec sample) which can be compared. The comparison is shown in 
226: figure~\ref{alternados2}, where the best fit to the data is the dotted line and the full 
227: line represents $s_{10}=s_{20}$.  The best fit slope $(0.89\pm 0.20)$ and intercept 
228: $(0.10\pm 0.15)$ are compatible with $s_{10}=s_{20}$.
229: 
230: We also compared the complete distributions of 10~ms and 20~ms integration times seeing 
231: values, shown in figure~\ref{alternados_distr}. A $\chi^{2}$ comparison test between both distribution
232: gives $\chi^{2} = 64.8$ for 46 degrees of freedom, that is a 3.5\% probability that both 
233: distributions are the same. However, if we compare both distributions with the common
234: distribution, derived from putting together the samples, then the respective $\chi^{2}$
235: values are 10.1 and 12.1 for the same number of degrees of freedom, giving respective 
236: probabilities of $1-(4\times 10^{-9})$ and $1-(10^{-7})$ that each distribution can be derived
237: from the same parent distribution. As it can be appreciated in the two lower panels of 
238: figure~\ref{alternados_distr}, the two distributions of seeing values are compatible with a 
239: single parent distribution. The total seeing median including both integration times is 0.77".
240: 
241: %%%  Begins comment 3
242: 
243: Following a suggestion by Marc Sarazin (private communication), we studied the presence of 
244: temporal averaging effects on our seeing measurements by comparing the seeing median {\em vs.} the wind 
245: velocity at 200~mb using the NOAA Global Gridded Upper Air data base, for each night of our 20~msec sample. 
246: The results are shown in the upper panel of Figure \ref{seeing_200mb}. The seeing error bars  
247: go from the first to the third quartile. The wind data are daily  average of four measurements available 
248: on the NOAA database. 
249: The dotted line represents the best least square linear fit, consistent with slope equal to zero 
250: within 1.1$\sigma$. 
251: The correlation coefficient is equal to $-0.227$ and the rms dispersion is 0.359". The errors
252: were obtained using a bootstrap technique. The data show that the seeing does not drop at high speed 
253: as it would be expected in the presence of temporal averaging effects. In the lower panel of the same 
254: figure, the 10~ms seeing data are shown. In this case the best fit is also consistent with zero slope, 
255: within 1.5$\sigma$, and the correlation coefficient is $-0.274$.
256: 
257: To study the presence of any correlation between  seeing and wind velocity at ground level, we
258: compare the seeing and  wind velocity daily medians at the site.
259: The wind velocity was measured with a meteorological station located on the seeing monitor tower.  
260: Figure~\ref{seeing_ground} shows the seeing median as a function of the wind velocity median 
261: for those nights that have simultaneous seeing and wind velocity data. 
262: The seeing error  bars  go from the first to the third quartile.  The wind velocity 
263: distribution median was calculated from the data obtained between 8:pm and 6:am local
264: time, the error bars corresponding to the first and third quartile are not included in the plot for clarity. 
265: In the upper panel the results for 17 nights of the 20~msec sample are shown. 
266: The dotted line represents the best fit that is consistent with zero slope within 1.4$\sigma$, a 
267: correlation factor equal to $-0.435$ and a rms dispersion of 0.384".  In the lower panel the data obtained for
268: 50 nights of the 10~ms sample are shown.  The best fit is consistent with slope equal to zero 
269: within 1.1$\sigma$, a correlation factor of  0.281 and a rms dispersion equal to 0.256". 
270: 
271: 
272: 
273: 
274: %%% Ends comment 3
275: 
276: In contrast with 
277: \citet{gio01} results on high altitude cordillera sites in  Chile, the data for Sierra Negra 
278: suggest that, within our statistics, there is no difference between the 10~msec and the
279:  20~msec series. It must 
280: be noted that \cite{vm95} find the seeing bias produced by the difference in exposure
281: time to be highly dependent on the magnitude of the star. As we generally use stars brighter
282: than $m_{V} = 2$, it is possible that this bias is in our data but below measurable error,
283: therefore not influencing our results. Nevertheless from December 2001 onwards all our 
284: measurements are made using 10~msec exposures. 
285: 
286: %%%%%%%%%%%%%%%%%%%%%%
287: \section{Conclusions}
288: %%%%%%%%%%%%%%%%%%%%%%
289: We present the first seeing measurements carried out at Sierra Negra. For 85 observing nights, 
290: seeing values of 0.62" were achieved 25\% of the time, below 0.78" during 50\% of the 
291: time, and sub-arcsec seeing for 70\% of the time.  The comparison  between the dry  
292: and the wet season shows that the seeing median is better during the dry season with a value
293: of 0.73" and sub-arcsec  seeing 77\% of the time. 
294: In contrast for the wet season the seeing 
295: median is 0.92" value is strongly affected  by  August 2001 contribution.
296: %
297: We analysed the dependence of the seeing statistics with time of night without 
298: finding any systematic trend of seeing as a function of time.
299: %
300: We did not find any correlation between the seeing values and the 200~mb wind velocity. A preliminary
301: analysis of the correlation between the seeing and  the wind velocity at ground level was carried out. The
302:  results show that there is not an  obvious correlation between them. 
303: Nevertheless we will continue analysing  the seeing as a function
304: of other meteorological parameters in more detail  to try to find  out  where  most of the turbulence
305: is concentrated. 
306: 
307: 
308: %
309: The results obtained so far show that Sierra Negra is a competitive site for optical 
310: astronomy. We will continue with our seeing and meteorological measurements to characterise 
311: the site on a longer time scale basis. We have carried out an independent analysis of the 
312: seeing temporal structure and the results will be reported in a separated  paper.
313: 
314: 
315: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
316: 
317:  
318: \acknowledgments {We thank R.J. Terlevich his interest and his useful comments and suggestions.  This study was  
319: possible through funds from the Instituto Nacional de Astrof\'{\i}sica, Optica y Electr\'onica. 
320: The 20mb wind velocity data was provided by the NOAA-CIRES Climate Diagnostics Center, Boulder, Colorado,
321: USA, from their Web site at http://www.cdc.noaa.gov/.} 
322: 
323: 
324: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
325: \begin{thebibliography}{}
326: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
327: \bibitem[Giovanelli~et~al.(2001)]{gio01}
328:  Giovanelli, R., Darling, J., Sarazin, M., Yu, J., Harvey, P., Henderson, Ch., 
329:  Hoffman, W., Keller, L., Barry, D., Cordes, J., Eikenberry, S., Gull, G., Harrington, J., 
330:  Smith, J. D., Stacey, G., \& Swain, M. 2001  \pasp\ 113, 789.
331: \bibitem[Martin(1987)]{ma87} Martin, H. M. 1987, \pasp\ 99, 1360.
332: \bibitem[Mu\~noz-Tu\~non, Vernin, \& Varela(1997)]{mvv97} 
333: Mu\~noz-Tu\~non, C., Vernin, J. \& Varela, A.~M., 1997, \aaps\ 125, 183.
334: \bibitem[Sarazin \& Roddier(1990)]{sr90} Sarazin, M. \& Roddier, F., 1990, \aap\ 227, 294.
335: \bibitem[Sarazin \& Tokovinin(2001)]{zt97} Sarazin, M. \& Tokovinin A., 2001,
336: Proceeding of Beyond Conventional Adaptive Optics, Venice 7-10, 2001, 
337: eds. R. Raggazoni, N. Hubin, \&  S. Esposito.
338: \bibitem[Vernin \& Mu\~noz-Tu\~non(1994)]{vm94} Vernin, J. \& Mu\~noz-Tu\~non, C., 1994, \aap\ 284, 311.
339: \bibitem[Vernin \& Mu\~noz-Tu\~non(1995)]{vm95} Vernin, J. \& Mu\~noz-Tu\~non, C., 1995, \pasp\ 107, 265.
340: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
341: \end{thebibliography}
342: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
343:  
344: \clearpage
345: 
346: \begin{figure}
347: \plotone{fig1.ps}
348: \caption{Layout of the Large Millimeter Telescope site. North is up and East to 
349: the right. The LMT track is marked by the 40~m diameter dark circle located at the 
350: middle left.  The approximate location of the seeing monitor and weather station is 
351: marked by the black square, East-North-East of the LMT construction. \label{layout}}
352: \end{figure}
353: 
354: \clearpage
355: 
356: \begin{figure}
357: \plotone{fig2.eps} 
358: \caption{Seeing daily statistics for all data points. The dots are the median and the error
359: bars go from the first to the third quartile. The central dotted line denotes the
360: seeing median for the whole data set, corresponding to {\bf 0.78}". The top dotted line is
361: the distribution third quartile, equal to 1.04", and the bottom dotted line is the
362: first quartile, corresponding to 0.61".\label{s_daily}}
363: \end{figure}
364: 
365: \clearpage
366: 
367: \begin{figure}
368: \plotone{fig3.eps}
369: \caption{Seeing histogram ({\em top}) and cumulative ({\em bottom}) distribution of all data 
370: points. The global 
371: statistics show that for 85 observing nights spanning from February 2000 to May 2002. 
372: Seeing values of 0.61" were achieved  25\% of the time, seeing below 0.78" 
373: during 50\% of the time and sub-arcsec seeing 70\% of the time.\label{s_distr_tot}}
374: \end{figure}
375: 
376: \clearpage
377: 
378: \begin{figure}
379: \plotone{fig4.eps}
380: \caption{Histogram ({\em top}) and cumulative ({\em bottom}) seeing distribution for the {\bf dry season} 
381: {\em i.e.} from November to April. The seeing median, indicated by the central 
382: dotted line, is {\bf 0.75}". The distribution first quartile, equal to 0.59", 
383: and the third quartile value, equal to 0.96", are indicated by the bottom and 
384: top dotted lines respectively. \label{dry_season}} 
385: \end{figure}
386: 
387: \clearpage
388: 
389: \begin{figure}
390: \plotone{fig5.eps}
391: \caption{Histogram ({\em top}) and cumulative ({\em bottom}) seeing distribution for 
392: the {\bf wet season}, 
393: {\it i.e.} from May to October. The seeing median, indicated by the central dotted 
394: line is {\bf 0.92}". The bottom dotted line corresponds to the first quartile value, 
395: equal to 0.69", and the third quartile value, equal to 1.34", is indicated by the top 
396: dotted line.\label{wet_season} }  
397: \end{figure}
398: 
399: \clearpage
400: 
401: \begin{figure}
402: \plotone{fig6.eps}
403: \caption{Distribution of seeing values for different hours of the night
404: (in UTC). The dots mark the median values while the error bars go from the
405: first to the third quartile. The upper panel shows the number of data points
406: for each hour interval. Note that the 1-2 hrs UTC bin (7-8:pm local) has
407: a much lower coverage than the middle of the night, so the deviation of 
408: uniformity for that bin might be apparent and due to low number statistics
409: ({\em i.e.} dominated by one or two nights with bad seeing). The data are basically 
410: consistent with no systematic trend during the night.\label{shourly}}
411: \end{figure}
412: 
413: \clearpage
414: 
415: \begin{figure}
416: \plotone{fig7.eps}
417: \caption{Seeing time profiles for 12 nights with samples of 15 seeing measurements
418: taken alternating with 10~msec and 20~msec integration times. Each point in the figure 
419: represents the median over each sample, 
420: distinguishing between 10~msec samples (filled circles) and 20~msec 
421: samples (open circles). There is no clear qualitative systematic difference
422: between both data, all taken in 2001.\label{alternados_1}}
423: \end{figure}
424: 
425: \clearpage
426: 
427: \begin{figure}
428: \plotone{fig8.eps}
429: \caption{Comparison of seeing data taken with integration times of 10 and 20~msec.
430: Each data set has been extended through a spline fit to overlap in time
431: the other data set, so that each point represents one sample (either 10~msec
432: or 20~msec) and the reciprocal spline interpolation (from the 20~msec or
433: 10~msec data). The full line represents $s_{10}=s_{20}$ while the dotted
434: line is the best fit to the data: $s_{20} = (0.888\pm 0.205)s_{10} + (0.102\pm 0.152)$,
435: with a correlation of 0.92 and rms=0.08". Errors were determined with the bootstrap method. 
436: \label{alternados2}}
437: \end{figure}
438: 
439: \clearpage
440: 
441: \begin{figure}
442: \plotone{fig9.eps}
443: \caption{Comparison between the distributions of $s_{10}$ and $s_{20}$ values ({\em top});
444: between the common distribution and $s_{10}$ values ({\em middle}); and between the common 
445: distribution and $s_{20}$ values ({\em bottom}).
446: As expected from the figure, the quartile and median values are similar. 
447: In particular $q_{2}(s_{10}) = 0.76"$, $q_{2}(s_{20}) = 0.78"$ and $q_{2}(s_{10}\& s_{20}) = 
448: 0.77"$. Both $s_{10}$ and $s_{20}$ distributions are compatible with the same single parent 
449: distribution. \label{alternados_distr}}
450: \end{figure}
451: 
452: 
453: 
454: \clearpage
455: 
456: \begin{figure}
457: \plotone{fig10.eps}
458: \caption{Seeing median as a function of the wind velocity at 200~mb.  The seeing error bars go 
459: from the first to the third quartile.  The daily average wind velocities were obtained from the 
460: NOAA Global Gridded Upper Air data base. Seeing data for the 20 ms sample ({\em top}): the dotted 
461: line represents the best linear fit, of slope $=-0.009\pm 0.008$, consistent with a zero slope 
462: within 1.1$\sigma$. The correlation coefficient is equal to $-0.274$ and a rms dispersion is 0.260". 
463: Seeing data for the 10~ms sample ({\em bottom}): the best linear fit, slope~$=-0.006\pm0.004$, 
464: is consistent with a slope equal to zero within 1.5$\sigma$. The correlation coefficient is $-0.227$ 
465: and the rms dispersion 0.359". 
466: \label{seeing_200mb}}
467: \end{figure}
468: 
469: 
470: \clearpage
471: 
472: \begin{figure}
473: \plotone{fig11.eps}
474: \caption{Seeing median as a function of the wind velocity median at ground level for those nights that
475: have both seeing and wind velocity data. The seeing error bars go from the first to the third quartile, 
476: while the wind velocity error bars are not included for clarity. Seeing data for 17 nights of the 20~ms 
477: sample ({\em top}): the dotted line represents the best linear fit, slope $=-0.06\pm 0.04$, which is 
478: consistent with a null slope within 1.4$\sigma$.
479: The correlation coefficient is $-0.435$ and a rms dispersion 0.384". Seeing data for 50 nights of 
480: the 10~ms sample ({\em bottom}): the best fit, slope $=0.03\pm 0.03$, is also consistent with a slope $=$ 0 
481: within 1.1$\sigma$. The correlation coefficient is 0.281 and the rms dispersion 0.256". 
482: \label{seeing_ground}}
483: \end{figure}
484: 
485: 
486: \clearpage
487: 
488: 
489: \begin{deluxetable}{lrrrrrrrrr}
490: \tabletypesize{\small}
491: \tablecaption{Global statistics per month considering equal weight per each seeing datum.
492: \label{table1}}
493: \tablewidth{0pt}
494: \tablehead{
495: \colhead{Date}  & \multicolumn{3}{l}{Data acquired} & \multicolumn{6}{l}{Data statistics (")}\\
496: \colhead{Month Year} & \colhead{Nights} & \colhead{Hours} & \colhead{Points} & \colhead{Mean} &
497: \colhead{$\sigma$} & \colhead{Min} & \colhead{$q_{1}$} & \colhead{Median} & 
498: \colhead{$q_{3}$}
499: }\startdata
500:  Feb 2000 &  1 & 0.3  &   42 & 0.756 & 0.087 & 0.606 & 0.689 & {\bf 0.748} & 0.795  \\
501:  Mar 2000 &  1 & 1.2  &  227 & 0.541 & 0.077 & 0.397 & 0.488 & {\bf 0.531} & 0.586 \\ 
502:  Apr 2000 &  2 & 1.5  &  205 & 0.650 & 0.132 & 0.482 & 0.578 & {\bf 0.631} & 0.696 \\
503:  Oct 2000 &  2 & 2.3  &  389 & 1.035 & 0.344 & 0.460 & 0.744 & {\bf 0.958} & 1.259 \\
504:  Nov 2000 &  2 & 3.5  &  427 & 0.701 & 0.204 & 0.290 & 0.553 & {\bf 0.678} & 0.824 \\
505:  Dec 2000 &  6 & 25.5 & 2383 & 0.661 & 0.354 & 0.238 & 0.401 & {\bf 0.506} & 0.894 \\
506:  May 2001 &  2 & 1.7  &  302 & 0.887 & 0.204 & 0.552 & 0.722 & {\bf 0.854} & 1.021 \\
507:  Jun 2001 &  3 & 7.5  & 1178 & 0.910 & 0.468 & 0.466 & 0.749 & {\bf 0.848} & 0.965 \\
508:  Jul 2001 &  4 & 12.0 & 1926 & 0.736 & 0.374 & 0.229 & 0.488 & {\bf 0.670} & 0.870 \\
509:  Aug 2001 &  8 & 26.7 & 4467 & 1.589 & 0.602 & 0.476 & 1.172 & {\bf 1.489} & 1.894 \\
510:  Sep 2001 &  2 & 6.0  & 1063 & 0.783 & 0.307 & 0.329 & 0.576 & {\bf 0.677} & 0.857 \\
511:  Oct 2001 &  3 & 15.2 & 2009 & 0.811 & 0.258 & 0.376 & 0.616 & {\bf 0.789} & 0.962 \\
512:  Nov 2001 &  9 & 37.6 & 7352 & 0.826 & 0.312 & 0.290 & 0.622 & {\bf 0.776} & 0.959 \\
513:  Dec 2001 &  7 & 37.5 & 6380 & 0.835 & 0.326 & 0.343 & 0.614 & {\bf 0.755} & 0.963 \\
514:  Jan 2002 &  4 & 29.6 & 3837 & 0.782 & 0.342 & 0.326 & 0.574 & {\bf 0.679} & 0.840 \\
515:  Feb 2002 &  6 & 36.0 & 4207 & 0.861 & 0.404 & 0.272 & 0.602 & {\bf 0.758} & 1.002 \\
516:  Mar 2002 &  9 & 36.1 & 4867 & 0.980 & 0.431 & 0.293 & 0.669 & {\bf 0.846} & 1.203 \\
517:  Apr 2002 & 10 & 44.1 & 7616 & 0.788 & 0.284 & 0.274 & 0.588 & {\bf 0.742} & 0.913 \\
518:  May 2002 &  4 & 17.9 & 3202 & 0.972 & 0.447 & 0.347 & 0.675 & {\bf 0.834} & 1.100\\
519: \tableline
520: All data  & 85 &342.1& 52079 & 0.898 & 0.441 & 0.229 & 0.615 & {\bf 0.784} & 1.046 \\
521: \enddata
522: %\tablenotetext{}{All statistical values in arcsecs.}
523: \end{deluxetable}
524: 
525: \end{document}
526: