3 Data reduction

The pointing accuracy was checked using the raw data of the four strong radio sources Cas A, Cyg A, Tau A, and Vir A. The center of the drift curve of each source was estimated and compared with the corresponding source position. Deviations smaller than $0{{\rlap.}{^{\circ}}}2$, and to the east, were found for each of these sources. It was assumed that all the other pointings were also located along the local meridian plane with the same accuracy. Since the MU radar has no calibration system we used the Maipu 45-MHz survey of the University of Chile (Alvarez et al. 1997) to calibrate the MU radar observations. The MU-radar and the Maipu arrays have important similar characteristics: the Chilean array, consisting of 528 full-wavelength dipoles, has an effective area of 11200 m², comparable to that of the Japanese array; the operating frequency of the Chilean array (45 MHz) is very close to that of the MU radar (46.5 MHz), and the beam width of the Maipu array is $4{{\rlap.}{^{\circ}}}6 \times 2{{\rlap.}{^{\circ}}}4$, also close to that of the MU radar ( $3{{\rlap.}{^{\circ}}}6$). Since the declination coverage of the southern array is $-86^{\circ}$ to $+19^{\circ}$, the instruments can observe a common declination range $+5^{\circ}$ to $+19^{\circ}$.

The data were smoothed by software simulating a normal RC filter in a detector. The integrating process was performed in direct and reversed time to minimize time shifting. The time constant of the filter was 60 s. Previous to this step, cleaning and excising of static bursts or interference was required in some of the observations. Next, the data acquisition rate of about one point every 10 s, was changed to one data point per minute in right ascension in order to make it the same as in the Maipu data. Each data point was centered at the 00 s of each minute and it represents the integration of five or six original data points. The profiles obtained were calibrated against the scan at $4{{\rlap.}{^{\circ}}}85$, obtained from the Maipu survey. This $4{{\rlap.}{^{\circ}}}85$ profile was selected because it was the overlapped position with best data in both surveys.

   

Table 3: Low-frequency surveys below 100 MHz in the northern hemisphere

$\nu$	Coverage ($\delta$)	Resolution ( $\alpha, \delta$)	Instrument	Reference
(MHz)	( $\hbox{$^\circ$ }$)	( $\hbox{$^\circ$ }$)
81	-28 - +82	2 $\times$ 15	Cylindrical paraboloid	Baldwin (1955)
45	+5.0 - +65	$3.6\times 3.6 \,{\rm sec}\, Z$	Filled circular array	This work
38	-25 - +70	8	Dish	Milogradov-Turin & Smith (1973)
38	-15 - +90	0.75	T-array	Williams et al. (1966)
38	-20 - +70	$\sim 2.5$	Moving-T	Blythe (1957)
34.5	-36 - +64	$26\hbox{$^\prime$ }\times 42\hbox{$^\prime$ }\ {\rm sec}\, Z$	T-array	Dwarakanath & Udaya Shankar (1990)
22	-28 - +80	$1.1\times 1.7 \,{\rm sec}\, Z$	T-array	Roger et al. (1999)
10	- 5.0 - +71	2	T-array	Caswell (1976)

A plot of temperature from the Maipu Survey versus the relative intensity from Shigaraki data, for every point of a profile at a given declination, is defined as a T-T plot in what follows. By means of an iterative T-T plot analysis the northern relative intensity profile at $4{{\rlap.}{^{\circ}}}85$ was calibrated to the corresponding temperature profile from the 45 MHz southern survey. Even though the overall correlation was good (about 0.998), the $4{{\rlap.}{^{\circ}}}85$northern profile thus calibrated, showed differences along the day with respect to the $4{{\rlap.}{^{\circ}}}85$calibrator profile of the Southern Survey. We have attributed these differences to temperature gain variations of the MU radar receiver. To correct for these fluctuations we produced a curve formed by ratios of corresponding points in RA, in both profiles. Next we fitted a polynomial to this curve in order to have a continuous and smooth curve of ratios (factors). This curve of factors was applied to the calibrated profile and a new T-T plot analysis was done with the profile thus modified. This is an iterative process that continued until the curve of factors was practically flat at 1. The convergence was fast and no more than three iterations were necessary. Then the final curve of factors, resulting from the product of the intermediate curves of factors, and the parameters derived from the T-T plots were applied to the rest of the northern profiles. Since the measured temperature is an average of the brightness distribution within the beam, and independent of the antenna effective area, we did not apply any Z-dependent correction. This process was applied separately to each of the data sets. We have assumed that, in each of the declination positions of the stepping beam, the gain of the system remained the same.

Although the original northern sky data were taken at 46.5 MHz, the calibration process just described brings the data to 45 MHz. The basic assumption in this process is that at a given declination the shape of the profile is the same at both frequencies. This seems reasonable since the frequencies are very close.

In combining the 1988 and the 1998 data, the two coverages have been added by using an algorithm similar to PLAIT (Emerson & Gräve 1988), with a double weight for the 1988 data. A further quality increase has been achieved by applying the so-called method of unsharp masking (Sofue & Reich 1979).