Astronomical observatories are ideally constructed and operated in locations that offer excellent observing conditions. In mm- and sub-mm astronomy, the limiting factor is the absorption of the astronomical signal in Erth's atmosphere due to broad wings of atmospheric absorption lines, mainly oxygen and, even more, water. As a result, mm- and sub-mm observatories are usually located at remote locations, in dry deserts and/or high-altitude mountain ranges. One of the best locations in this respect is the Chajnantor plateau in the Atacama desert in northern Chile, not without reason the home of the ALMA observatory, currently under construction.
The Atacama Pathfinder EXperiment (APEX)[1] is a modified ALMA prototype antenna, with 12 m diameter, used for frequencies between 200 GHz and 1.5 THz. It is located at an altitude of 5100 m in the Chilean Altiplano, close to the desert village San Pedro de Atacama. This remote location requires some thoughts about how observations are to be performed. This does not only include economic concerns (e.g. air fares), but also safety issues. Even at the APEX base camp, at an altitude of 2400 meters, health problems can occur.
The classical observing mode is the ``visitor mode'', where the observer travels to the observatory at the time when his/her project is scheduled, to observe exactly this project. This mode is cost intensive and time consuming, and doesn't allow for flexible scheduling based on the atmospheric conditions required for a given project, but has the advantage that the visiting observer (usually the project principal investigator (PI)) has full control over the observations, and - assuming that he is well experienced - can optimize the observing strategy in real-time in order to maximize the scientific value of the observations. In terms of observing management, ``remote observing'' is very similar, except that the observer is connected to the observatory remotely, which reduces travel costs with only a slight reduction in flexibility and efficiency. ``Pool observing'' takes a different approach, here an observer usually travels to the observatory, but observes projects which are selected, at the time of the observations, from a project pool, depending on the LST range and instrumental and atmospheric requirements. For all the above cases, the observer is usually an astronomer from outside. The efficiency of the observing depends on his experience in observational astronomy, and he may not be up to date with the latest developments at the telescope he is observing at, like instrument changes or software upgrades.
In service observing, the observations are performed by observatory staff (astronomers and/or operators) who are working at the telescope a significant percentage of their working time, and therefore (ideally) know all the details about the hard- and software which is used for the observations. In addition, this observing mode allows flexible scheduling, since the astronomer on duty can decide in real-time which project to observe based on weather conditions or telescope problems. It therefore can be the most efficient observing mode possible.
In service observing mode, the PI of a project is not present at the time of the observations. Thus measures have to be taken to make sure that the project is observed in a way that the scientific goals are reached. This includes the preparation of the project, the actual observations (including real-time decisions about issues which define the data quality, like pointing sources, integration time per scan, etc.), and the progress monitoring of a project (in order to define a project as ongoing or finished). The most important of these measures is an efficient information transfer from the PI to the observatory staff about how the project should be observed.
At APEX, service observing is the standard observing mode. Observing management tools had to be developed to use service observing successfully and with a high degree of efficiency. In Section 2 I describe the project submission facility, i.e. how the project information is transferred from the PI to the observer. In Section 3 I outline the actual observing steps, while Section 4 explains how the APEX staff keeps track of the progress of any observing program, and Section 5 covers the handling of the resulting scientific data. In Section 6 I evaluate the service observing concept at APEX, outlining its advantages, and give an outlook on future activities.