Once the telescope is configured, scheduled, aligned, and focused, normal operations can commence. ROTSE-III is designed to be an automated, robotic telescope; most nights, it should take hundreds of images without necessary human intervention. Once the images are recorded, there is a suite of analysis programs to extract useful information from these images. Ultimately, we hope to make a full closed-loop automated analysis program, in which the automatic software processes the results gleaned from images to instruct the daq in the scheduling of future images, as well as to distribute discovery alerts, but for the moment, only some of this process is fully automated. The entire analysis pipeline sequence is diagrammed in Figure 8.1. In this chapter, we describe the automated analysis code and resulting data products, a few steps that must be performed manually after the automated steps are complete, and the methods we have established to monitor system behavior during operation.
We have written several sets of programs to analyze ROTSE-III data as it is recorded to disc. These programs in concert provide calibrated object lists for observed fields. After a GRB alert has been received, sets of object lists for multiple observations of the target field are automatically compiled into ``match structures''. These match structures enable us to track the change in intensity of all objects in the field. They can be easily scanned for highly variable objects, as a GRB is expected to be, and they are a useful tool for weeding out spurious artifacts that only appear in a single image. Although match structures are produced in response to a GRB alert, the primary output of the automated realtime processing is the calibrated object list, so we describe in Section 8.2 how these lists are created, and defer all discussion of match structures to Section 8.3, when the manual manipulation of data products is described.