Preliminary discussion of results from Dec 1999 run

• Fig 4 shows the stability of the blackbody reference measurements. Top left: plot of variation in detector signal due to ambient blackbody over one week period. The radiance from this blackbody follows the ambient temperature and so variation in this reading is expected. Bottom left: same for LN₂ blackbody. Each day the detector was allowed to warm up and then refilled with LN₂. The residual variation in this signal is likely due to a slight repositioning error of the LN2 blackbody. Typically 20 scans were obtained for each point in the left plot; the standard deviations of those 20 scans is shown in the right plots. The scan to scan repeatability is around 5mV for the LN2 data throughout the week. It is higher for the ambient data but this reflects the slow change in ambient temperature during the 10 minute scans. These data show that the system is intrinsically stable. Detailed analysis of the system responsivity (calculated from the blackbody temperatures and measured signals) on time scales ranging from 30 secs (a single sky-dip) to a week has shown no evidence of variability, as is expected.
• Fig 5 shows three sky-dips observed during a 50 scan sequence. The 1, 25 and 50 scans are shown from which it can be seen that the atmosphere was drying out. The top graph is given in terms of zenith angle (steps 0.18$^{\circ}$), the lower plot in terms of airmass. Each full sky dip takes around 30 seconds.
• Fig 6 is similar to Fig 5 but on a dry and stable night. Since each sky-dip is a measurement of atmospheric radiance over a range in airmass from 1 to 3, then in principle it should be possible to build up an experimental curve of growth (equivalent to the theoretical model (Fig 3)) by combining sky-dip curves taken under different water vapour pwv conditions (since water vapour is the only source of emission in the band). Furthermore, by having an independent measure of the water vapour content of the atmosphere (eg radiosonde) this curve of growth can be calibrated. Unfortunately the weather conditions in Dec 99 were generally poor and often varying and our most stable sky-dips were several hours displaced from the radiosonde launches. Nevertheless, we have developed an algorithm to combine the averages of the most stable sky-dip runs assuming that the atmosphere varied only in water vapour content. This algorithm scales individual sky-dip measurements in terms of airmass (or equivalently pwv amount) minimizing the $\chi^2$ of the overlap regions of individual sky-dip sequences. This analysis is still in progress but our early results are shown in Fig 7 (the x-axis scale is left in terms of airmass but once calibrated would be given in terms of mm pwv). The lower plot shows the result of applying this algortihm to 4 sky-dip sequences (the second and third offset by + and - .1 respectively for clarity); the upper plot shows the 4 traces overlaid. The agreement is remarkable and if we had independent pwv estimates from radiosonde data for the start of each of the sky-dips we would be able to calibrate this curve for direct comparison with our theoretical model.
• During the Dec 99 run Prof. Richard Hills was at the JCMT and kindly showed us how to operate the 183GHz water vapour monitor. On the last night we operated this system while we were pointing at Jupiter. Simultaneously we operated the IWVM with the zenith angle set to that of Jupiter. However, because the IWVM is mounted in front of the windscreen (a part of the building that rotates with the dome, but separate from the telescope) we were not aligned at the same azimuth as the telescope. Nevertheless, the same atmosphere swept through the IWVM beam several minutes later. The results are shown in Fig 8. The lower trace is the 183GHz line-of-sight water vapour. The next three traces are IWVM results (each displaced upwards by .2mm pwv): the first is the raw data (0.1 sec samples), the next is smoothed to 1 sec, the last to 10 sec. Since we have not yet calibrated the IWVM we have used our best estimate model atmosphere for the infrared filter and then scaled our vertical axis. The agreement between the 2 systems is remarkable. Many of the features are evident in both systems while some appear to have shifted slightly during the elapsed time.