5.1 Limitations of the study

Beam blockage and path-integrated attenuation are only two of the possible sources of uncertainties that could affect data quality. Furthermore, attenuation was only considered for the ground radar, while the spaceborne radars may also experience attenuation. However, this has already been acknowledged and addressed by the corresponding SR data teams, by providing attenuation-corrected datasets. The SR data product used in the analysis are already attenuation-corrected and the values were taken as-is. To take this approach further, the framework of quality-weighted averaging could and should be extended to the SR data as well, including not only attenuation, but also the effects of non-uniform beam filling.

It should be noted that the approaches in calculating beam blockage and path-integrated attenuation are not absolute. In this thesis, they were calculated as part of the processing chain, and not as readily available quantities. Exploring these variables further are research topics in themselves, but a sufficient method had to be selected. For the beam blockage fraction calculation, the method presented by Bech et al. (2003) was used. The radar’s field of view is simulated in their model using information on the scan geometry and the surrounding topography based on a digital elevation model. The implementation of this model assumes that there is no vertical gradient of refractivity, so in cases where super-refraction of the atmosphere may be possible, this method may not apply. As for the retrieval of the PIA, the specific differential phase (\(K_{DP}\)), which was used to calculate PIA following the algorithm proposed by Vulpiani et al. (2012). This method was already proven to work well with the dataset (Crisologo et al. 2014). While other \(K_{DP}\) retrieval methods exist (Bringi and Chandrasekar 2001; Y. Wang and Chandrasekar 2009), it is beyond the scope of this thesis to explore their differences, but certainly, the philosopher’s stone has not yet been found when it comes to the reliable reconstruction of \(\Phi_{DP}\) and \(K_{DP}\), and hence the retrieval of PIA.

Aside from the methodological, data limitations also exist. TRMM and its successor GPM was launched in 1997 and 2014, respectively, therefore the SR-GR method of bias estimation can only be done as far back as the start of the corresponding instrument. Additionally, with the study area close to the equator, the swath density is not as high as it is nearer the poles. There are at most two SR overpasses that go over the Philippines everyday, and they don’t always intersect with the radars used in this study. Applying the methods to study areas with high swath overpass frequency could provide more samples to increase the reliability of the results.

Unfortunately, we were not able to gain access to the calibration and maintenance history of the Subic and Tagaytay radars as recorded by PAGASA, and manual cross-checking of the changes in radar calibration against the maintenance records was not possible. Determining the exact dates of maintenance events that could have changed the radar calibration and comparing them with the observed changes in the analysis would allow us to differentiate between the effects of deliberate calibration and those of instrument instability and system drift. The findings with respect to the annual calibration changes of the Subic radar were relayed to radar engineers in PAGASA, and they confirmed that some hardware changes (i.e. magnetron replacement) were performed in 2014, but no further details were given.

The observed variability of the calibration biases over time could result from cumulative effects of several sources of uncertainty along the entire data collection and generation process. Examples of such sources could be uncertainty of beam propagation due to fluctuations in atmospheric refractivity; non-uniform beam filling; residual errors in the geometric intersections of the volume samples; and rapid changes in precipitation during the time interval between the radars being compared, to name a few. Additionally, hardware instabilities owing to the effects of temperature, thermal expansion, and gradual degradation of the system, can also contribute to calibration drifts.

References

Bech, Joan, Bernat Codina, Jeroni Lorente, and David Bebbington. 2003. “The Sensitivity of Single Polarization Weather Radar Beam Blockage Correction to Variability in the Vertical Refractivity Gradient.” Journal of Atmospheric and Oceanic Technology 20 (6): 845–55. http://journals.ametsoc.org/doi/abs/10.1175/1520-0426(2003)020%3C0845:TSOSPW%3E2.0.CO;2.

Bringi, V. N., and V. Chandrasekar. 2001. Polarimetric Doppler Weather Radar: Principles and Applications. Cambridge University Press.

Crisologo, I., G. Vulpiani, C. C. Abon, C. P. C. David, A. Bronstert, and Maik Heistermann. 2014. “Polarimetric Rainfall Retrieval from a C-Band Weather Radar in a Tropical Environment (the Philippines).” Asia-Pacific Journal of Atmospheric Sciences 50 (S1): 595–607. https://doi.org/10.1007/s13143-014-0049-y.

Vulpiani, Gianfranco, Mario Montopoli, Luca Delli Passeri, Antonio G. Gioia, Pietro Giordano, and Frank S. Marzano. 2012. “On the Use of Dual-Polarized C-Band Radar for Operational Rainfall Retrieval in Mountainous Areas.” Journal of Applied Meteorology and Climatology 51 (2): 405–25. https://doi.org/10.1175/JAMC-D-10-05024.1.

Wang, Yanting, and V. Chandrasekar. 2009. “Algorithm for Estimation of the Specific Differential Phase.” Journal of Atmospheric and Oceanic Technology 26 (12): 2565–78. https://doi.org/10.1175/2009JTECHA1358.1.