RESUMEN

Se presentan estadísticas del uso del telescopio de 2.1m del Observatorio Astronómico Nacional en San Pedro Mártir, Baja California, Mexico. También se incluyen parámetros climáticos básicos como la temperatura y humedad relative al nivel del telescopio. La fracción de noches perdidas debido a mal clima fue de 27.4% en el periodo julio 1982 a junio 1992. De enero 1984 a diciembre 1991, el 56.7% de las noches fueron ``fotométricas'' mientras que el 80.4% de las noches fueron ``espectroscópicas''. La humedad relativa promedio fue de 54%. La mejor época, en término de noches despejadas y baja humedad, es en primavera y otoño, ya que en el invierno se tiene la influencia de las colas de algunas tormentas del Pacífico norte, mientras que tormentas locales afectan algunas noches de verano. El porcentagunas noches de verano. El porcentaje total de noches nubladas en San Pedro Mártir es similar al de Cerro Tololo y La Silla, un poco mejor que en Mauna Kea y considerablemente mejor que en Mount Graham, especialmente en verano.

ABSTRACT

Ten year statistics are presented of the use of the 2.1m telescope of the Observatorio Astronómico Nacional at San Pedro Mártir, Baja California, Mexico. Also included are some basic climatic and meteorological parameters such as temperature and relative humidity as well as the amount of cloud cover. The fraction of nights lost due to bad weather was 27.4% in the period July 1982 to June 1992. From January 1984 to December 1991, 56.7% of the nights were of ``photometric'' quality and 80.4% were of ``spectroscopic'' quality. The average relative humidity near the ground was 54%. Spring and Autumn are the best seasons in terms of cloudless and low humidity nights while winter is affected by the tails of North Pacific storms and mid summer is characterized by a mild monsoon season. The total cloud cover percentage in San Pedro Mártir is similar to that of Cerro Tololo and La Silla, slightly better than Mauna Kea and considerably better that Mount Graham, especially in summer.

1. Introduction

In the late 1960's, a search was made in order to find the best site for astronomical observations in Mexico. This was done by analysing two years (1969 - 1971) of mewo years (1969 - 1971) of meteorological satellite photographs (Mendoza et al. 1972) and several in situ tests in the most prominent candidate sites. The summit of the Sierra of San Pedro Mártir in Baja California, Mexico was chosen to locate the Observatorio Astronómico Nacional, which has been engaged in continuous astronomical research programs since 1971. Several studies of the climatological properties of this site have been reported, mainly based on data covering short periods during the first years of operation of the observatory (Mendoza 1971, 1973, Mendoza et al. 1972, Walker 1971, 1983, Westphal 1974, Alvarez 1982, 1984 and Alvarez and Maisterrena, 1977).

The Observatorio Astronómico Nacional de México at San Pedro Mártir is located 31°02'40''N and 115°28'00''W, some 100 km east of the Baja Californian West Coast. It has has three Ritchie-Chrétien telescopes of diameters 0.84m, 1.5m and 2.1m in operation. The latter started continuous operation in 1981. It is expected that in the near future, the Mexican astronomical community engages in the construction and operation of a larger (4 to 8 met larger (4 to 8 meter class telescope) and San Pedro Mártir is a candidate site for this instrument.

The present work describes the data gathered in a consistent manner during the last ten years of operation of the 2.1m telescope concerning the nocturnal basic climatological parameters, such as temperature and relative humidity as well as cloud cover and observing statistics for this telescope. This is the first long-term report on the weather charateristics of San Pedro Mártir and the data covers from 1st July 1982 to 30th June 1992 (3653 calendar nights) and, unless specifically stated, the results given here refer to that period.

Finally, a comparison is made with similar results published for several other observatory sites, including Mauna Kea (Bely 1987, Merrill and Forbes 1987), Mount Graham (Merrill and Forbes 1987), Kitt Peak (Crawford 1987), La Silla (Sarazin 1990) and Cerro Tololo (Osmer and Wood 1983).

2. Data archive and analysis

The 2.1m telescope is the only one at San Pedro Mártir that requires night assistant for its operation. Since July 1982, it is one of his duties to daily fill in an entry on the log book of the telescope with the following information: Minimum and maximum temperature, wind speed and relative humidity readings at the level of the telescope at the beginning and end of each night; sky conditions in terms of the percentage of cloud cover during the of cloud cover during the night; estimate of seeing conditions; name of observers and auxiliary instrument used; times of openning and closure of the dome and notes on the reasons for early ending or late start of the observing night.

Unfortunately, due to several failures with the wind meters and lack of a consistent method for estimating the size of the seeing disc, these two parameters are left out of this compilation. All other parameters were accurately recorded on the log book every night except when no observer was scheduled on the telescope. This occur mainly because of holidays between Christmas and a few days after New Year, telescope maintenance and engineering runs and scattered periods of undersubscription of telescope use. Figure 1 shows the monthly average (ten years) distribution (in percentage over the total number of calendar nights each month) of the time scheduled for observation. Apart from the months of January and December, and maybe some periods in summer when eventual major engineering runs are scheduled, there is no clear explanation for this distribution. It is important to note that for all months, a statistically representative number of nights were recorded and, as will be seen later, this pattern has no relation with patterns of ``good'' or ``bad'' weather seasons. A total of 2533 nights were scheduled for observation from July 1982 to June 1992. The statistics of the use of specifi of the use of specific auxiliary instruments on the 2.1m telescope lies outside the scope of the present work, but it is important to note that 42% of the scheduled nights were used for work requiring clear nights (``photometric'' conditions) and 58% for programs which could be carried out during clear and partially cloudy nights (``spectroscopic'' conditions).

For the analysis of the data and the presentation of the results, the following definitions were made:

The units used were half nights, meaning a period of five hours during the night. Except for the results of the number of nights of high, medium and low relative humidity.

Photometric sky: When the percentage of cloud cover was less or equal to 15%. In the great majority of the cases, if clouds were present, they were reported seen near the horizon or present anywhere in the sky for no more than thirty minutes in a five hour period.

Spectroscopic sky: When the percentage of cloud cover was greater than 65%. This applied either to percentage of the sky covered by clouds or to percentage of time with the sky covered.

< the sky covered.

Time observed: When actual observations took place, meaning that they were not impeded by bad weather (including high winds, very high humidity and very bad seeing) or other causes (including telescope or instrument failure).

Night of high relative humidity: When anytime during the night the relative humidity was higher than 70%.

Night of low relative humidity: When at no time during the night the relative humidity was higher than 45%.

Night of medium relative humidity: When at all times during the night the relative humidity was between 45% and 70%.

3. Results

Nocturnal air temperature variations in San Pedro Mártir are illustrated in Figure 2. Monthly absolute minima and maxima and average (10 years) minima and maxima are shown. Temperature variations within a single night are relatively small. In the period 1985--1991, the maximum minus minimum nocturnal temperature DeltaT was less than 1.5°K in 61% of the recorded nights (1261 in total), while DeltaT < 2.5°K in 81% and DeltaT < 3.5°K in 89% of the nights. Unfortunately, no information is available for the daytime temperatures and therefore no day-night temperature gradients can be estimated. Figure 3 shows the monthly average relative humidity from all nights of observty from all nights of observation and those nights when the observations were stopped or cancelled because of very high humidity (usually with relative humidity larger than 90%) for the period January 1984 to June 1992. The average value is 54%, being May and June the dryest months and August and September the most humid. Note that these are night-time readings and the relatively low values for the summer months are surprising since during the monsoon season (from late July to early September) it is quite common to have rainstorms in the afternoon which in most cases dissapear at sunset. Probably more useful for our purposes is Figure 4 which shows the average monthly value of the percentage of nights with high, low and medium relative humidity as defined in Section 2.

The ten-year observing statistics for the 2.1m telescope is given in Table 1. All entries are self explanatory. Table 2 gives the monthly percentage of nights lost due to bad weather in the 2.1m telescope over the almost ten years considered. In this category are included the (half) nights lost due to cloud cover, high humidity, high winds and very bad seeing, but no difference is made between time lost for photometric or spectroscopic/imaging programs. Clearly, the percentages are calculated over the total number of scheduled nights, as no weather information io weather information is recorded for the nights not scheduled for observation.

From the nightly data on the percentage of cloud cover, the average number of ``photometric'' and ``spectroscopic'' nights (as defined in Section 2) was computed for the January 1984 to December 1991 period. Previous to 1984, the log book did not contain detailed cloud cover information. In order to include only full years, the 1992 data was omitted. The results are shown in Tables 3 and 4 and displayed graphically in Figure 5 (monthly averages) and Figure 6 (yearly averages). In the latter, note the dip in the percentage of observed nights in late 1982 and 1983. This was caused by the well known El Niño phenomenon. It is interesting that, in spite of reports of the re-appearance in 1992 of a similar phenomenon, it has only barely been reflected in the amount of cloudy nights in San Pedro Mártir up to the end of June 1992. Figure 6 also shows the yearly fraction of nights of actual observation with the 2.1m telescope (excluding tescope (excluding the time lost due to telescope/instrument failure) from July 1982 to June 1992.

From January 1984 to June 1992, the fraction of photometric and spectroscopic nights in San Pedro Mártir were 56.6% and 79.9% while astronomical observations were actually carried out on the 2.1m telescope (1984 - 1992) 69.7% of the scheduled nights, 25.4% of the nights were lost due to poor weather and 9.1% due to telescope/instrument failures and other causes. From July 1982 to June 1992, 27.4% of the scheduled nights were lost due to bad weather (Table 1).

4. Comparison with other sites

From the first studies of its climatological properties, San Pedro Mártir has been recognized as one of the potentially best sites for optical/infrared observations in the Northern Hemisphere (Mendoza et al. 1972; Walker 1971; Alvarez and Maisterrena 1977). It is only after ten years of continuous monitoring of the metereologial characteristics of this site, that proper comparison can be made with other existing (or well studied) observatory sites like Mauna Kea, Hawaii, Kitt Peak and Mt. Graham, Arizona, Cerro Tololo and La Silla, Chile. Graphic comparisons of the monthly percentage of the cloud cover will be made using plots on the same scale of published data and those reported in Section 3.

Geographically speaking, San Pedro Mártir is the Northern Hemisphere co the Northern Hemisphere counterpart of Cerro Tololo or La Silla. They are at roughly the same latitude and at similar distance to the Pacific coast, though San Perdro Mártir is at an elevation some 500m higher. The data on the monthly fraction of photometric nights considered in this work for the Chilean sites is the weighted (by number of nights) average of those published by Osmer and Wood (1984) for Cerro Tololo in the period 1973 - 1982 and by Sarazin (1990) for La Silla in the periods 1965 - 1972 and 1983 - 1990. The fraction of spectroscopic nights, defined as ``one with at least 50% [of the sky] free from clouds or at most 20% absorption over a major part'', refer to the period 1984 - 1986 in La Silla only (Sarazin 1990). These values are plotted in Figure 7a. The upper continuous line connects the points corresponding to the spectroscopic sky and the lower line, the photometric sky. Note that for direct comparison with the Northern sites, the abscisa has been shifted six months. The average values for San Pedro Mártir are shown connected by broken lines. The average fraction of photometric nights in the Chilean sites is 59.8% (Sarazin 1990; Osmer and Wood 1983) and there seems to be no appreciable difference between Cerro Tololo and La Silla; the fraction of spectroscopic nights at La Silla is 82.8% (Sarazin 1990).

The Mauna Kea Observatory in the Big Island of Hawaii is the largest observats the largest observatory in the Northern Hemisphere (in terms of collecting area of existing and planned telescopes). It is also the highest and dryest site in continuous operation. Mt. Graham, Arizona, has also been studied extensively as a possible site for some of the largest telescopes planned. Merrill and Forbes (1987) made a comparison study of the astronomical site quality between Mauna Kea and Mt. Graham and concluded that the former has superior astronomical qualities. This study was based on similar simultaneous (February 1984 to March 1986) tests on the basic parameters which affect the optical and infrared astronomical observations. For the purpose of this work, only the cloud cover and relative humidity on the ground will be considered.

Figures 7b and 7d show the monthly average fraction of photometric (lower continuous curves) and spectroscopic (SKYOK; upper continuous curves) nights at both sites as reported by Merrill and Forbes (1987). The definition by these authors of SKYOK nights (defined as ``times when the cloud cover was <= 60% and itermitent'') is analogous to that adopted here for ``specd here for ``spectroscopic'' nights and therefore, these data are directly comparable. Merrill and Forbes (1987) state that their results are representative of the long term weather behaviour in each place, based on comparison with similar values recorded for longer periods at Kitt Peak, some 200 km SW of Mt. Graham and the log book of the University of Hawaii 2.2m telescope on Mauna Kea. To illustrate this, also included in Figures 7b and 7d are the reported (dotted lines) values of the fraction of observed time with the 2.1m and 4m telescopes on Kitt Peak (Crawford 1987) and the U. of Hawaii 2.2m telescope. Finally, Figure 7c shows that for San Pedro Mártir, the weather in the period March 1984 to February 1986 (continuous lines) was not very different from the long term weather pattern observed for 1984 - 1992. In all panels of Figure 7, the San Pedro Mártir monthly average fraction of photometric (lower broken lines) and spectroscopic (upper broken lines) are presented for comparison.

5. Conclusions:

The results reported in this work and those published for Mt. Graham, Mauna Kea and Cerro Tololo and La Silla are summarized in Table 5. From this and the plots in Figure 7, the following conclusions can be reached concerning the amount of cloud cng the amount of cloud cover and relative humidity at the ground:

1. San Pedro Mártir has proved to be a site with a very low fraction of nocturnal cloud cover in the long term. The figures presented here for the period 1982 to 1992 are in agreement with the observations made over shorter periods (up to three contionous years) in the early 1970's. Winter is the worst season, spring and a autumn have a larger number of clear nights. Local afternoon rainstorms during the summer months cause a moderate increase in the number of cloudy nights in July and August, combined with an increase in relative humidity. Apart from this, no seasonal pattern is observed in the nocturnal readings of relative humidity near the ground.

2. The percentage of clear (photometric) and partially clear (spectroscopic) nights in San Pedro Mártir is slightly lower than in the Chilean sites. The main difference occur in the months of summer, due to a moderate monsoon season in the Baja Californian site. The rest of the year they show similar statistics, though in winter it seem to be slightly less cloudy in San Pedro Mártir. The ESO definition of ``spectroscopic'' nights (cf. Sections 4) is slightly more relaxed than the ones adopted for San Perro Mártir.

3. The fraction of spectroscopic nights in Mauna Kea is slightly lower that in San Pedro Mártir though there is a marked difference in the number of photometric nhe number of photometric nights.

4. Mount Graham seems to be the site with the smallest fraction of clear and partially clear nights. This is most dramatic during the period May to October, when the rather long monsoon season (caused by winds from the Gulf of Mexico) make the chances of having cloudless, or even partially cloudy, nights rather slim. In the winter and early spring, the weather in Mt. Graham and San Pedro Mártir seems to be similar, affected by winter storms from the North Pacific. From late spring to late autumn, San Pedro Mártir has a much larger percentage (almost 30% difference) of clear and astronomically usable nights than Mt. Graham.

5. The average relative humidity near the ground at San Pedro Mártir is 54%, implying that it is a dryer site than Mt. Graham (73%; Merrill and Forbes 1987) but a more humid site than Mauna Kea (25%; Merrill and Forbes 1987) and La Silla (~ 30%; Sarazin 1990).

Clearly, many other factors apart from those discussed here should be considered for the qualification of a site for optical and infrared observations. Of prime importance is the mean size, or rather, the distribution er, the distribution of the size of the seeing disk over a single night and throughout the year. Only very few proper measurements of the seeing in San Pedro Mártir have been made by Walker (1971; 1983) in 52 nights from April to July 1970. He found that the seeing was less than 1" in 17% of the nights and around 60% of the nights had seeing better than 2". A continuous monitoring of the seeing is now being carried out in San Pedro Mártir in collaboration with Steward Observatory. Results from a very limited number of nights suggest that the seeing is extremely good (J. Echevarría, private communication). These data are being supplemented with a second seeing measuring system on loan from Las Campanas observatory.

Sky brightness over San Pedro Mártir, as in all other observatories, has varied considerably in the years following the eruption of the Chichón and Pinatubo volcanoes. The 1970 observations reported by Walker (1971) show that the sky brightness (V magnitude) is 21.9 at the zenith while it is 21.7 at zenital distance 45° in the direction of the nearest large cities (Mexicali, Tijuana, San Diego, Imperial Valley and Ensenada), all located at distances greater than 150 km. These sky brightness values are similar to those for Mauna Kea and Cerro Tololo and better than Mt. Graham (Garstang 1989) for the same epoch.

Finally, the amount of precipitable water in the atmosphee water in the atmosphere, critical to evaluate the quality of the sky for infrared observations, is also being monitored continuously from January 1992 in San Pedro Mártir with a radiometer in collaboration with the University of Massachusetts. Up to the end of May, the observed distribution of the amount of precipitable water in the atmosphere above San Pedro Mártir (L. Salas et al.; in preparation) is similar to those reported for La Silla for winter and spring (Sarazin 1990).

6. Acknowledgements

This compilation would not have been possible without the cooperation of the 2.1m telescope night assistants: Javier Ojeda, Jaime López, Octavio Sánchez, Antonio Chaidez, Marco García, Gabriel García, Juan Ruiz, Celso Valle, Eduardo López, Ricardo Flores, Alvaro Lepe, Salvador Monrroy, Celso Valle, Francisco Martínez, Alfonso Paredes, Joel Rodríguez, Felipe Montalvo and Roberto Velazco. Thanks also to Joaquín Bohigas, Rafael Costero, Alberto López and Miguel Roth for help and comments and to Luis Salas and Juan Echevarría for showing me preliminary results of the seeing and water vapour content monitoring.

REFERENCES

Alvarez, M. 1982, Reporte Técnico, Instituto de Astronomía, UNAM, No. 5.

Alvarez, M. 1984, Contribución, Instituto de Astronomía, UNAM, No. 123.

AlvareAM, No. 123.

Alvarez, M. and Maisterrena, J. 1977, Rev. Mex. Astron. Astrof., 2, 43.

Bely, P.Y. 1987, PASP, 99, 560.

Crawford, D.L. 1987, Kitt Peak National Observatory Facilities Book, NOAO.

Garstang, R.H. 1989, PASP, 101, 306.

Mendoza, E.E. 1971, Bol. Obs. Tonantzintla Tacubaya, 6, 95.

Mendoza, E.E. 1973, Mercury, 2, 9.

Mendoza, E.E., Luna, J. & Gómez, T. 1972, Bol. Obs. Tonantzintla Tacubaya, 6, 215.

Merrill, K.M. and Forbes, F.F. 1987, Comparison Study of Astronomical Site Quality of Mount Graham and Mauna Kea, NNTT Technology Developement Program Report No. 10, NOAO.

Osmer, P.S. and Wood, H.J. 1984, in Site Testing for Future Large Telescopes, eds. A. Ardeberg and L. Woltjer, ESO, p95.

Sarazin, M. 1990, VLT Site Selection Working Group Final Report, ESO-VLT Report No. 62.

Walker, M.F. 1971, PASP, 83, 401.

Walker, M.F. 1983, in Site Testing for Future Large Telescopes, ESO, p3.

Westphal 1974, Infrared Sky Noestphal 1974, Infrared Sky Noise Survey, Final Report NASA-NGR-05-002-185, California Institute of Tecnology.

FIGURE CAPTIONS