1 Introduction

Pulsars have been studied for more than five decades (Hewish et al. 1968) and are observed across the whole electromagnetic spectrum, from low radio frequencies (e.g., Pilia et al. 2016; Xue et al. 2017) to very high-energy photons ∼TeV (e.g., H. E. S. S. Collaboration et al. 2023). However, their emission mechanism is still not well understood (e.g., Beskin 2018). Radio observations covering a wide frequency range allow us to measure pulsar wideband spectra and to measure their emission properties evolving as a function of frequency and therefore are particularly important for our understanding of pulsar emission mechanisms. Previously, only a small fraction of pulsars have been examined over a wide frequency range and most existing wideband observations used multiple instruments rather than a simultaneous frequency coverage (e.g. Dai et al. 2015). Such a situation is changed thanks to new instruments like the Ultra-Wideband Low (UWL) receiver system (Hobbs et al. 2020) on the Parkes radio telescope (also known as Murriyang). The UWL receiver system has a continuous frequency range from 704 to 4032 MHz with exceptional sensitivity and polarization properties, which have proven to be extremely powerful in understanding radio emissions from various types of pulsars (e.g., Dai et al. 2019; Johnston et al. 2021; Sobey et al. 2022; Oswald et al. 2023).

Radio pulsars as a population show steeper spectra and their spectra can be roughly modelled as a simple power law (e.g., Lorimer et al. 1995; Maron et al. 2000; Jankowski et al. 2018). Jankowski et al. (2018) examined the spectral features of 441 radio pulsars observed with the Parkes telescope and indicated that around 79\(\%\) of the pulsars had simple steep spectra with a mean spectral index of −1.6. However, some pulsar spectra exhibit a turn-over and deviate from simple power law at low frequencies (e.g., Malofeev 1993; Kuniyoshi et al. 2015; Lee et al. 2022), while others display a break in the spectrum or high-frequency cut-off in the form of a spectrum steepening (Jankowski et al. 2018; Lee et al. 2022).

The observed flux density of pulsars varies on different time scales. One of the main causes of such variations is interstellar scintillation, which is observed to be strong variations of pulsar flux density in both time and frequency. There are two types of interstellar scintillation, the so-called diffractive and refractive scintillation (e.g., Rickett 1990). Diffractive scintillation is caused by small-scale spatial fluctuations in electron density and can be observed as strong flux density variations from several minutes to hours (e.g., Wang et al. 2005; Reardon et al. 2019). On the other hand, refractive scintillation is caused by large-scale inhomogeneities in scattering screens and can be observed as flux density variations from days to several months (e.g., Kumamoto et al. 2021). Sensitive observations can also detect parabolic arc structures in the secondary spectra, which result from the interference of multiple signal pathways at a specific distance between the pulsar and the observer (Stinebring et al. 2001). Walker et al. (2004) developed theoretical descriptions of the parabolic arcs and used Monte Carlo methods to present secondary spectra.

Pulsar emission is frequently strongly polarised (both linear and circular) (Lorimer and Kramer 2012), with average linear polarization fractions of ∼ 20\(\%\) that decreases towards higher frequencies and average circular polarization fractions of ∼ 10\(\%\) that increases towards higher frequencies (e.g., Johnston and Kerr 2018; Sobey et al. 2021). Broadband polarization observations can reveal essential information on pulsar beam geometry, radio emission mechanisms and properties of pulsar magnetosphere. Furthermore, many pulsars exhibit consistent variations in linear polarization position angle (PA) across the pulse phase (e.g., Dai et al. 2015; Spiewak et al. 2022). The rotating vector model (RVM), for example, explains the observed PA swings in several pulsars as an S-shaped curve (Radhakrishnan and Cooke 1969). The circular polarization across the pulse is also observed to show sense reversal, which can be linked to the emission mechanism or propagation effects (e.g., Wang et al. 2010).

While the population of pulsars at high Galactic latitudes is much smaller than that of the Galactic plane, recent all-sky surveys have significantly increased the sample (e.g., Keith et al. 2010; Boyles et al. 2011; Keane et al. 2018; Li et al. 2018; Sanidas et al. 2019; Bhat et al. 2023). Measuring their rotation measures (RMs) and scintillation properties could provide valuable insights into the magnetic field and ISM turbulence at high Galactic latitudes. Understanding the polarization and spectrum of this population of pulsars is also important for establishing the strategies for pulsar searches with next-generation radio continuum surveys, such as Australian Square Kilometre Array Pathfinder (ASKAP), which has the potential to uncover hundreds of pulsars (Dai et al. 2017). For example, the first pulsar discovered with the ASKAP was initially identified as a highly polarised radio source in continuum images (Kaplan et al. 2019). Another pulsar in the Large Magellanic Cloud was discovered as a highly circularly polarised source in ASKAP images (Wang et al. 2022). Two pulsars were discovered as highly polarised sources in the radio images from the Low-Frequency Array Two-metre Sky Survey (LoTSS; Sobey et al. 2022). Six millisecond pulsars and one normal pulsar were discovered by cross-matching steep spectrum radio sources with unidentified \(\gamma \)-ray sources (Frail et al. 2018). Other methods also have been explored and developed to distinguish pulsars from other compact radio sources, such as using the variability of pulsars caused by scintillation (e.g., Dai et al. 2016). These imaging domain searches could complement previous all-sky surveys at high galactic latitudes (e.g., Keane et al. 2018) since they have relatively short integration time, which could have missed faint pulsars and long period pulsars (e.g., Tan et al. 2018; Caleb et al. 2022).

In this paper, we describe broadband observations of 18 high Galactic latitude pulsars with DM from \(\sim 7\) to 45 pc cm−3 using the UWL receiver system. Details of the observation, data reduction, and data analysis are given in Sect. 2. In Sect. 3, we present the flux density measurements and spectral indices, rotation measures and polarization pulse profiles, and prospects of discovering pulsars in radio continuum images. Discussion and conclusions of our results are given in Sect. 4.

2 Observations and data analysis

2.1 Observations

A sample of 22 pulsars were observed with Parkes using the UWL receiver as part of the project P1059 .Footnote 1 Each pulsar was observed twice (except for PSR J0125−2327 having three epochs) with an integration time of \(\sim 60\) minutes. The observational parameters of all pulsars are listed in Table 1, where the last column presents the detection/non-detection status of each pulsar. These pulsars are located at high Galactic latitudes (\(|gb|\) > 20∘) and do not have flux density and polarization measurements at around 1.4 GHz. 14 of these pulsars were originally discovered by Parkes multi-beam surveys (Manchester et al. 1996; Lyne et al. 1998; Burgay et al. 2006; Burke-Spolaor et al. 2011; Keane et al. 2018; Morello et al. 2019) and our 1 hour observations are deep enough to produce high signal-to-noise ratio (S/N) pulse profiles. Seven pulsars were discovered with Arecibo telescope (Lorimer et al. 2007; Camilo et al. 1996) and one was discovered in Molonglo pulsar survey (Manchester et al. 1978).

Table 1 Observational parameters for 22 high galactic latitude pulsars which include position (RAJ \(\&\) DECJ), spin period (P0), dispersion measure (DM), distance to pulsar (DIST), observation date (MJD), and integration time (Int. Time). The position coordinates (RAJ \(\&\) DECJ), period (P0), DM, and distance (DIST) are taken from ATNF pulsar catalogue (Manchester et al. 2005). The last column indicates the detection status of pulse profile of pulsars in their respective folded data. The letters “N” and “Y” represent the detection status as NO and YES for detection/non-detection, respectively
Full size table

The Medusa backend was used in combination with the UWL which provides a radio-frequency coverage from 704 to 4032 MHz (Hobbs et al. 2020). For each pulsar observation, UWL is used in coherently de-dispersed search mode and data was recorded with 1 MHz channel bandwidth, 2-bit sampling, 4-polarizations, and 128 μs time resolution. Before each pulsar observation, a calibration observation of ∼ 120 s duration was obtained by pointing the telescope slightly offset (\(<1\) arcminute) from the pulsar and injecting a pulsed calibration square wave signal of known periodicity directly into the feed.

2.2 Data reduction

The search-mode data was folded using the DSPSR software package (van Straten and Bailes 2011) with sub-integrations of duration 30 s modulo the barycentric pulsar spin period to form a pulse profile with 1024 phase bins using the pulsar ephemerides from the ATNF Pulsar Catalogue Footnote 2 (Manchester et al. 2005). The calibration data was also folded using DSPSR package with sub-integrations of duration 10 s, and 128 phase bins. Folded data were then processed with the PSRCHIVE software package (Hotan et al. 2004) and the data reduction progressed as follows.

Firstly, the calibration observation was examined by visual inspection using the pazi program to remove sub-integrations and channels that are corrupted due to radio frequency interference (RFI). We used the pcm program together with the calibration observation and the polarization calibration solutions based on observations of J0437−4715 to calibrate our data. The absolute flux density was obtained using the fluxccal program together with flux calibration solutions based on observations of a known source PKS 0407−658. Both the polarization and absolution flux density solutions are provided and updated by the ATNF.Footnote 3 Finally, the calibrated pulsar observation was examined visually, and corrupted sub-integrations and channels (if any) were removed from the data. The Stokes parameters follow the astronomical conventions outlined by (van Straten et al. 2010) and the Institute of Electrical and Electronics Engineers (IEEE) defines the left and right-hand circular polarization to be positive and negative, respectively.

We successfully detected 18 pulsars out of 22 observed. In these 18 detections, six pulsars (J0125−2327, J1403−0314, J2154−2812, J2215+1538, J2253+1516, J2307+2225) were detected across the whole band. Three pulsars (J1453+1902, J1517−32, J2354−22) can only be detected at below 3 GHz. The remaining nine pulsars can only be detected at below 2 GHz. For four undetected pulsars (J0006+1834, J0410−31, J1940−2403, J1947−4215), we extracted the frequency band (1216−1728 MHz) with 512 frequency channels for both epochs and carried out periodic search with pulsar searching software package PRESTO (Ransom 2001). We searched for periodic signals at dispersion measures (DM) within ±10 pc cm−3 around their nominal DMs. No pulsars were detected in both epochs.

2.3 Data analysis

2.3.1 Flux density measurements

To measure pulsar flux density and spectral index, we averaged the calibrated profile of each pulsar in time and polarization and averaged in frequency to \(8\times 416\) MHz sub-bands. We used the PSRCHIVE program PAAS to create standard templates from our observations, and then the PSRFLUX program to measure the flux density in each sub-band. PSRFLUX first matches the template with the observation and then computes the area under the standard template. This is accomplished by performing a basic least squares fit on the measured profile, which is supposed to be a scaled replica of the template with an offset. The uncertainty in flux density measurements is derived from the least squares uncertainty estimation. In Table 2, we present flux densities from two epochs for eight sub-bands (each with 416 MHz bandwidth). For some pulsars, flux densities were not measured for all eight sub-bands due to the non-detection of these pulsars in some sub-bands.

Table 2 18 high galactic latitude pulsars included in this table with multiple epoch wide-band flux density measurements, fractional linear \(\mathrm{L}/\mathrm{I}\), circular \(\mathrm{V}/\mathrm{I}\), and absolute circular polarization \(|\mathrm{V|}/\mathrm{I}\). Fractional polarization is measured for epoch1 except for PSR J1012–2337. 1\(\sigma \) uncertainties on the last quoted digit are given in parenthesis. Measurements for some sub-bands of several pulsars are missing due to the non-detection of the pulse profile in the corresponding sub-band
Full size table

2.3.2 Faraday rotation measures

To measure pulsar RMs, we extract the lowest of eight sub-bands (704-1728 MHz) with a total of 1024 frequency channels. We measure the RM of each pulsar using the rmfitFootnote 4 program of PSRCHIVE. An initial estimation of the RM is found by brute force examining a RM range of ± 1500 rad m−2. At each trial RM, rmfit adjusts for the corresponding Faraday rotation and calculates the linear polarization \(L = \sqrt{(Q^{2}+U^{2})}\) across the on-pulse phase bins, where U and Q are linear stokes parameters. Subsequently, rmfit fits a Gaussian profile to the resulting RM spectrum to determine the peak RM. Later, this RM value is refined by selecting a narrower RM range (e.g., −300 rad m−2 to 300 rad m−2) with smaller RM steps. rmfit refinement algorithm divides the data into two equal frequency bands, and each data slice is integrated over frequency, and then calculates a weighted differential position angle (\(\Delta { \mathrm{PA}}\)). Only the phase bins with significant linearly polarised flux (\(>3\sigma \)) are used to calculate \(\Delta {\mathrm{PA}}\). If \(\Delta {\mathrm{PA}}\) exceeds its uncertainty, the data is corrected with initial guess of RM by brute force, and then \(\Delta {\mathrm{PA}}\) is computed again. The final best-fit RM is reported once the \(\Delta {\mathrm{PA}}\) is less than its uncertainty. RM measurements for 17 pulsars are listed in Table 3. We were not able to measure the RM of PSR J1517−32 and did not detect any significant linear polarization. We did not correct the ionospheric Faraday rotation in Table 3.

Table 3 18 pulsars are included in this table with pulsar names, RMs obtained in this work (RMobs), previously measured RMs (RMcat), parameters such as spectral indices (\(\alpha \)) and corresponding cut-off/tur-over frequencies from spectral model fits, dispersion measure (DM), scintillation bandwidth (\(\nu _{\mathrm{d}}\)) and timescales (\(\tau _{\mathrm{d}}\)) at 1.4 and 3 GHz, and modulation indices (\(m\)) at 1.3 GHz. The table includes 8 new RM results that have not been previously published. All RMs are measured using epoch1 observation except for PSR J1012−2337. We have included the references to papers with previous RM measurements in table footnotes
Full size table

3 Results

3.1 Flux density and spectral indices

In Table 2, we present flux densities of each pulsar from two epochs in eight sub-bands each with 416 MHz bandwidth. The wide-band and multi-epoch flux density measurements allow us to calculate the spectral index. For this purpose, we split the whole UWL band into 16 sub-bands each with 208 MHz bandwidth for at least two epochs of observations and measure flux densities in each of them. Due to the low signal-to-noise ratio, the flux density can not be measured in all sub-bands except for five pulsars (J0125−2327, J2154−2812, J2215+1538, J2253+1516, J2307+2225). To compensate for the effects of scintillation on pulsar spectra, we averaged the flux densities obtained from corresponding sub-bands at multiple epochs for each pulsar.

In general, the majority of the pulsar spectra can be modelled by simple power-law. However, deviations from simple power-law are observed in many pulsars, manifesting as low-frequency turnover, spectral steepening at high frequencies, or break in the pulsar spectra. To account for this, we have implemented the software pulsar\(\_\)spectra,Footnote 5 which is an open source pulsar flux density catalogue and automated spectral fitting software that identifies the optimal spectral model among different morphological classes, including simple power-law, broken power-law, low frequency turn-over, and high frequency cut off. Further details regarding the spectral fitting methodology and each model can be found in Swainston et al. (2022).

Out of 18 pulsars, the spectral behavior of 11 pulsars in our sample can be well described by simple power-law. Two pulsar showed spectrum steepening at high frequencies. Five pulsars showed low frequency turn-over. The best fitting power-law spectral models are shown as dashed black lines (see Fig. 1). Orange shaded region is the 1\(\sigma \) uncertainty of the best-fitting model. The associated spectral indices, turn-over frequencies, and cut-off frequencies are listed in the Table 3.

Fig. 1
figure 1figure 1figure 1

Flux density spectra for the 18 pulsars with the black dashed line as bes-fitting model mentioned at the lower left corner of each plot. Orange shaded envelope is the 1\(\sigma \) uncertainty of the best-fitting model

Full size image

3.2 Rotation measures and polarization profiles

We measured RMs for 17 pulsars, including seven new measurements (for pulsars J1000+08, J1126–38, J1335–3642, J1403–0314, J1708+02, J1947–18, J2354–22). The RM and its uncertainty for each pulsar are listed in Table 3. We also listed previous RM measurements of these pulsars from the ATNF pulsar catalogue RM measurements (Manchester et al. 2005). As seen in Table 3, the majority of RM measurements (RMobs) are in agreement with the ATNF catalogue values (RMcat). Seven of our 18 pulsars (J1012–2337, J2154–2812, J2205–1444, J2215–1538, J2234–2114, J2253–1516 and J2307+2225) were studied recently by Posselt et al. (2023), we measured slightly different RMs with significantly lower uncertainties as compared to RMs reported by Posselt et al. (2023). This is due to the wide frequency range covered by UWL, along with the higher S/N profiles produced by the long integration time. Two of the RMs we measured are significantly different from previous measurements. These include J0843+0719 with RM\(_{\mathrm{cat}} =\) 40(4) rad m−2 (Ng et al. 2020) compared to RM\(_{\mathrm{obs}}= \) 49(1) rad m−2 and J1453+1902 with RM\(_{\mathrm{cat}} =\) 3.2(6) rad m−2 (Spiewak et al. 2022) compared to RM\(_{\mathrm{obs}} =-5(6)\) rad m−2.

Multi-frequency polarization pulse profiles for each pulsar were produced after averaging in time and correcting for the Faraday rotation using RMobs presented in Table 3. We produce averaged polarization pulse profiles for eight different frequency sub-bands (centered at 0.9 GHz, 1.3 GHz, 1.7 GHz, 2.1 GHz, 2.6 GHz, 3 GHz, 3.3 GHz, and 3.8 GHz) each with 416 MHz bandwidth. The baselines for the stokes I, Q, U and V profiles have been set to zero mean. The linear polarisation L was calculated as \(L = \sqrt{(Q^{2}+U^{2})}\), and the noise bias in L was corrected according to the equation 11 in Everett and Weisberg (2001). The similar bias in \(|\mathrm{V|}\) was corrected as described in Dai et al. (2015). Additionally, we measured the fractional linear (L/I), circular polarization (V/I), and absolute circular polarization (\(|\mathrm{V|}\)/I) for each sub-band. The on-pulse phase range with a significant signal exceeding 3\(\sigma \) was used to calculate the fractional linear, circular, and absolute circular polarization. In order to compare polarization fractions at different frequencies, the same on-pulse phase bin ranges were selected for all sub-bands.

In Fig. 2, we present the on-pulse fractional polarizations of 18 pulsars in different frequency sub-bands. The polarization profiles are shown in Fig. 3. Black, red, and blue lines represent the total intensity (I), linear polarization (L), and circular polarization (V), respectively. Polarization angle (P.A) of linear polarization is plotted in a smaller panel on top of each profile. Because of the different channels excised due to RFI, the central frequencies for the 8 sub-bands for each pulsar may differ. Polarization profiles of several pulsars in some sub-bands are not shown due to the low signal-to-noise ratio.

Fig. 2
figure 2figure 2figure 2

The distribution of on-pulse fractional polarization vs frequency for all pulsars. The top, middle and bottom panel show linear, circular and absolute circular fractional polarization, respectively. All pulsar names are labelled in the upper right corner with the same colours as their plots. Right-hand panels: Histograms showing the distribution of fractional polarization of 18 pulsars at 1.3 GHz

Full size image
Fig. 3
figure 3figure 3figure 3figure 3figure 3figure 3figure 3figure 3figure 3

Average polarization profiles for 18 pulsars in different frequency bands between 704 and 4032 MHz, with polarization position angles in the upper panels. The red line is the linear polarization profile, the blue line is the circular polarization profile, and the black line is the total intensity profile

Full size image

The top panel of Fig. 2 illustrates the on-pulse fractional linear polarization (L/I) of 18 pulsars in different frequency sub-bands. Over the UWL frequency range, the highest fractional linear polarization is observed to be \(\sim 84\)% for J1453+1902 at \(\sim 900\) MHz, while pulsars like J1517−32 show low levels of linear polarization at low frequencies. PSR J1453+1902 shows a high fractional linear polarization (\(>40\)%) in all sub-bands that it was detected. PSR J1708+02 shows a large fractional linear polarization (\(\sim 50\%\)) at the lowest frequency sub-band and it decreases drastically at higher frequencies. At around 1.3 GHz, as shown by the histogram in the top panel, most of the pulsars show moderate fractional linear polarization (∼2−20\(\%\)) and the average degree of linear polarization for the 18 pulsars is \(\sim 16\%\). We show that the fractional linear polarization increases as a function of frequency for some pulsars (e.g., J1403+0314, J1947−18, J2215+1538, J2253+1516) and decreases in several cases (e.g., J0125−2327, J1708+02, J2354−22). It is also evident that the fractional linear polarization of pulse profile components can show different frequency evolution. For example, for pulsar J0125−2327, the fractional linear polarization of the leading component decreases with increasing frequency whereas that of the trailing component increases with increasing frequency.

The middle and bottom panels of Fig. 3 show the on-pulse fractional circular (\(\mathrm{V}/\mathrm{I}\)) and absolute circular polarization (\(\mathrm{|V|}/\mathrm{I}\)) as a function of frequency, respectively. The highest fractional circular polarization is \(\sim 25\)% right-handed and \(\sim 25\)% left-handed for pulsars J2215+1538 and J2154−2812, respectively. For most of the pulsars, the fractional circular polarization is in the range of \(\sim 2-20\%\) over the bandwidth. The average degree of circular polarization for the 18 pulsars is \(\sim 7\%\) at 1.3 GHz which is lower in comparison to previous results (e.g., Sobey et al. 2021) with an average fractional circular polarization of \(\sim 16\%\). However, several pulsars (e.g., J2215+1538, J2253+1516, and J2307+2225) show low levels of absolute circular polarization, but it increases drastically at higher frequency sub-bands (>2.5 GHz). J2215+1538 shows left-handed circular polarization at the lowest two sub-bands and a sense reversal of circular polarization at higher frequencies. On the contrary, a few pulsars like J1453+1902, J2154−2812, and J2205+1444 show a decrease in absolute fractional circular polarization with increasing frequency, although for PSR J2154−2812 increases surprisingly (\(\sim 20\%\) jump) at the highest frequency sub-band that it was detected. Different components of pulsars often have opposite/same signs of circular polarization and opposite/same frequency dependence. For example, for pulsar J1335−3642 and J2234+2114, leading and trailing components have opposite and same signs of circular polarization, respectively. Overall, our results are consistent with previous polarization measurements (e.g., Sobey et al. 2021; Johnston and Kerr 2018).

3.3 Detecting pulsars in radio continuum images

The Rapid ASKAP Continuum Survey (RACS) is the first large-area survey conducted with the full 36-antenna ASKAP telescope (McConnell et al. 2020). The first data release of RACS (also known as RACS–low) consists of 903 images covering the sky south of declination \(+41^{\circ}\) made over a 288 MHz band centred at 887.5 MHz. The median root-mean-square (rms) noise is \(\sigma _{\mathrm{med}}=0.25\) mJy/beam. For a significant detection with \(5\times \sigma _{\mathrm{med}}\), the expected source continuum flux density is \(\sim 1.25\) mJy.

As shown in Table 2, six of our pulsars (J0125–2327, J1947–18, J2215–1538, J2234+2114, J2253+1516, and J2307+2225) have flux densities higher than 1.25 mJy at 900 MHz. We searched for their radio continuum counterparts in RACS images and detected three (J2215+1538, J2234+2114, J2253+1516) out of the six pulsars in RACS images. PSR J2253+1516 is detected with a flux density of \(\sim 2.5\) mJy/beam which is roughly consistent with our Parkes measurements with flux density varying between 0.5−2.9 mJy. Our Parkes measurements showed that it is highly variable at 0.9 GHz and 1.3 GHz. It is also close to a bright radio continuum source and its continuum flux density could be affected by imaging artifacts. PSR J2215+1538 is clearly detected in RACS-low with a flux density of ∼7.8 mJy, which is consistent with our Parkes measurements (see Table 2). PSR J2234+2114 is marginally detected with a flux density of \(\sim 0.98\) mJy/beam. This is much lower than our Parkes flux densities at 900 MHz varying between ∼1.6−2.5 mJy which suggests its strong flux variation at around 1 GHz. Pulsars, such as J0125−23, J1947−18, and J2307+2225, have measured flux densities well above the RACS \(5\times \sigma _{\mathrm{med}}\) detection threshold but are not visible in RACS images. Again, this is likely due to the strong scintillation of these pulsars. We will discuss this in Sect. 4.

4 Discussion and conclusions

In this paper, we presented wide-band observations of 18 pulsars at high Galactic latitudes (\(|gb|>20^{\circ}\)). We measured the wide-band flux density and spectrum of each pulsar at multiple epochs and presented their wide-band polarization pulse profiles. We also measured RMs for 17 pulsars, including seven new measurements (for pulsars J1000+08, J1126–38, J1335–3642, J1403–0314, J1708–02, J1947–18, J2354–22). Three pulsars (J2215+1538, J2234+2114, J2253+1516) can be identified in ASKAP RACS continuum images, and we compared their flux densities with our Parkes measurements.

Observations at different epochs showed that the flux density of many of these pulsars varies significantly across a wide band. This is most likely due to interstellar scintillation as the radio emission from pulsars propagates through the ionized ISM. The DM of our pulsars ranges from \(\sim 7\) to 45 pc cm−3, and at frequencies near 1 GHz we expect them to be in the strong scintillation regime. To estimate the values of diffractive scintillation bandwidth (\(\nu _{\mathrm{d}}\)) and the scintillation time-scale (\(\tau _{\mathrm{d}}\)), we utilise models and empirical fits to data, as well as ISM turbulence models. We estimate the scattering time (\(\tau _{ \mathrm{s}}\)) using empirical relation obtained by (Krishnakumar et al. 2015) (assuming \(\tau _{\mathrm{s}}\) ≈ \(\nu ^{-\mathrm{4}}\) scaled to 1.4 GHz and 3 GHz, respectively):

$$ \tau _{\mathrm{s},\mathrm{1.4\,GHz}} (\mu \,\mathrm{s) = 1.20 }\times \mathrm{10^{-5} DM^{2.2} (1 }+\mathrm{ 0.00194 DM^{2})}, $$
(1)
$$ \tau _{\mathrm{s},\mathrm{3\,GHz}} (\mu \,\mathrm{s) = 5.06 }\times \mathrm{10^{-7} DM^{2.2} (1 }+\mathrm{ 0.00194 DM^{2}). } $$
(2)

Assuming a homogeneous medium with a Kolmogorov spectrum (Cordes and Rickett 1998), \(\nu _{\mathrm{d}}\) can be obtained by,

$$ \nu _{\mathrm{d}} (\rm MHz) = \frac{1.16}{2\pi \tau _{\mathrm{s}}}. $$
(3)

The scintillation time-scale, \(\tau _{\mathrm{d}}\), can be estimated as (Johnston et al. 1998),

$$ \tau _{\mathrm{d}} = \frac{3.85 \times 10^{4} \sqrt{D \nu _{\mathrm{d}}}}{\nu V} , $$
(4)

where V is the transverse velocity of the pulsar in km s−1 and \(\nu \) is the observing frequency in GHz. Since the transverse velocity of pulsars in our sample is not measured, we have used 300 km s−1 which is the average transverse velocity of known pulsar population (Sobey et al. 2021). The distance to the pulsars, \(D\) (in kpc), are taken from ATNF pulsar catalogue (Manchester et al. 2005).

At low DMs, such as in our case, the scattering timescales \(\tau \) tend to be larger than predicted by Krishnakumar et al. (2015). Therefore, we advise caution as there may be a tendency to overestimate the scintillation bandwidth when approaching the lower limits of both DM and \(\tau \) achievable at a given frequency (Oswald et al. 2021). The spectral index measurements for the pulsars with a wide scintillation bandwidth can be affected by scintillation. However, to compensate for the effects of scintillation on spectral index measurements, we used averaged flux densities using at least two epochs of observations for each pulsar.

In Table 3, we listed \(\nu _{\mathrm{d}}\) and the \(\tau _{\mathrm{d}}\) at 1.4 and 3 GHz. We also calculated the modulation index (e.g., Kumamoto et al. 2021) at 1.3 GHz to quantify the flux variation (see the last column in Table 3), although we only have two measurements of most pulsars. Several pulsars, such as PSRs J1126−38 and J1517−32, show relatively small flux variation across the whole band. These pulsars have the highest DMs in our sample and therefore small \(\nu _{\mathrm{d}}\) and \(\tau _{\mathrm{d}}\), and their scintillation effects are smoothed out with our long integration and wide bandwidth. On the other hand, pulsars like J0125−2327, J2307+2225, and J2354−22 have low DMs and show large flux variation across the whole UWL band. However, we found that several pulsars with relatively high DMs and small \(\nu _{\mathrm{d}}\) and \(\tau _{\mathrm{d}}\), such as J1403−0314 and J2253+1516, also show large flux variations across the whole band. While at high frequencies (\(>2.5\) GHz), \(\nu _{\mathrm{d}}\) and \(\tau _{\mathrm{d}}\) become comparable to our bandwidth and integration time and could lead to large variation, their flux variations at low frequencies are difficult to understand. We also suggest that the non-detection of four pulsars (J0006−1834, J0410−31, J1940−2403, and J1947−4215) is caused by strong scintillation of these pulsars.

Pulsars can be detected as variables or slow transients in radio continuum surveys. For example, the Phase I Pilot Survey of the ASKAP VAriables and Slow Transients (VAST) survey, which covered a total of ∼ 1646 square degrees comprising two different sky regions, identified 28 sources at 888 MHz, including seven known pulsars with high variability (Murphy et al. 2021). The highest DM of pulsars identified by VAST is ∼ 30 pc cm−3, which is comparable with the DMs of pulsars discussed in this paper (DM ∼ 7−45 pc cm−3). Here we showed that, at even higher frequencies (\(\sim 3\) GHz), pulsars at high Galactic latitudes with relatively low DMs can also be strong variables. Therefore, transient surveys at higher frequencies, such as the Very Large Array Sky Survey (VLASS; Lacy et al. 2020), should be able to detect a population of pulsars at high Galactic latitudes.

The observed strong flux variations of these pulsars also suggest that previous all-sky pulsar surveys were likely to miss a significant population of pulsars at high Galactic latitudes. Limited by telescope observing time, previous all-sky pulsar surveys were shallow and only covered the sky once. For example, previous Parkes surveys such as the Survey for Pulsars and Extragalactic Radio Bursts (SUPERB; Keane et al. 2018) and the High Time Resolution Universe Pulsar Survey (HTRU; Keith et al. 2010) used an integration time of 270 s and 560 s per pointing, respectively. New all-sky pulsar surveys at low frequencies (e.g., the MWA SMART survey; Swainston et al. 2021) and all-sky continuum surveys (e.g., ASKAP EMU; Norris et al. 2011) have the potential to discover more pulsars and help us better understand the Galactic distribution of pulsars.

11 pulsars (see Table 3) in our sample have spectra that follow a simple power-law with spectral indices between –0.75 (PSR J0125–2327) and –4.30 (PSR J1012–2337) with the mean of –2.15. The previous studies show that the mean spectral index of pulsars with a simple power-law is around –1.60 (e.g., Jankowski et al. 2018). While the mean spectral index is steeper for simple power-law, our results are limited by a very small sample size. Five pulsars exhibit a clear low-frequency turnover. Particularly noteworthy examples are PSR J0125–2327 and PSR J1403–0314, which demonstrate the most steep spectral indices of –6.56(4) and –7.93(2), respectively. These pulsars exhibit gradual turnovers with smoothness parameter \(\beta \) below unity, typically associated with the synchrotron self-absorption process (Izvekova et al. 1981; Jankowski et al. 2018). The other three pulsars (PSR J2215+1538, PSR J2234+2114, and PSR J2354–22) have sharper turnovers with \(\beta =\) 2.1, which is a special case of thermal free-free absorption model (e.g., Kijak et al. 2017; Jankowski et al. 2018). PSR J2234+2114 and PSR J1947–18 are the only pulsars in our sample to exhibit high frequency cut-off, which is a model to describe the coherent emission of the electrons accelerated by the pulsar electric field (Kontorovich and Flanchik 2013). Following the Jankowski et al. (2018), we estimated the emission height of PSR J2234+2114 as \(z_{e} = 20.0 \pm 3.4\) km. This suggest that the emission height is remarkably low and emphasize the need for additional constraints at higher frequencies to assess the model validity and its underlying assumptions.

The high fraction of polarization, circular polarization particularly, has been suggested to be a key parameter to distinguish radio pulsars from other compact radio sources in continuum surveys. Large-scale radio continuum surveys showed that a significant fraction of the polarised sources is pulsars (Lenc et al. 2018), although other compact radio sources such as cataclysmic variables (Mutel and Morris 1988) and flare stars (Lynch et al. 2017) can also be circularly polarised. Recent follow-up observations of highly circularly polarised sources in ASKAP surveys led to the discoveries of PSRs J1431−6328 (Kaplan et al. 2019) and J0523−7125 (Wang et al. 2022). Our measurements of fractional circular polarization of these high Galactic latitude pulsars agree with previous results of normal pulsars and MSPs (e.g., Dai et al. 2021). The majority of known pulsars show a fractional circular polarization less than \(\sim 20\%\). While there is a great potential for detecting new pulsars using circular polarization information, relatively weak pulsars with low fractional polarization require additional information and/or methods (e.g., Dai et al. 2016) to be effectively identified in radio continuum surveys.

With our long integration time, we observed systematic drifting of the spin period of PSR J1947−18, which strongly suggests that PSR J1947−18 is in a potential binary system. PSR J1947−18 has a spin period of 2.6 ms and a DM of 25.2 pc cm−3. It was originally discovered in high Galactic latitude parts of the HTRU survey (Morello et al. 2019), which has an integration time of 560 s per pointing. To constrain the orbital period, we split the one-hour observation into 10 chunks and then measured the pulse period within each chunk using Tempo2. Although the time span is short, we tried to fit for a circular orbit. Our current best-fitting suggests that the orbital period is 8.4 hour and the projected semi-major axis is 0.035 light seconds. These parameters indicate an extremely low-mass companion. However, more observations covering the entire orbital period are required to tightly constrain the orbital parameters.