Development and validation of an anti-leakage least-squares spectral analysis-based zenith tropospheric delay model for the East African region.
| dc.contributor.author | Matsukuni, Nicholas | |
| dc.date.accessioned | 2026-09-15T07:08:07Z | |
| dc.date.available | 2026-09-15T07:08:07Z | |
| dc.date.issued | 2026 | |
| dc.description | Dissertation | |
| dc.description.abstract | Accurate estimation of Zenith Tropospheric Delay (ZTD) remains challenging over East Africa because of strong spatial and seasonal atmospheric variability and the limited distribution of continuous GNSS observations. Existing empirical models such as GPT3 and GTrop mainly describe temporal ZTD variability using predefined harmonic terms, which may not adequately represent the more complex regional seasonal behaviour. This study developed a regional ZTD model for East Africa using Anti-Leakage Least-Squares Spectral Analysis (ALLSSA) and ERA5 reanalysis data from 2013 to 2020. Unlike fixed-harmonic approaches, the developed model represents temporal ZTD variability using statistically significant frequencies identified adaptively from the time series, allowing the spectral structure to respond to regional differences in atmospheric behaviour. ERA5 atmospheric profiles were used to derive ZTD, while the vertical dependence was represented using a reference ZTD and a height-dependent decay parameter. Sensitivity analysis showed that increasing the number of ALLSSA iterations beyond N=40 produced little further reduction in RMSE, and this value was adopted for model development. Internal evaluation against ERA5-derived ZTD yielded a mean correlation coefficient of 0.99 and a mean RMSE of 25.19 mm, showing that the model retained the dominant variability represented in ERA5. Independent validation using GNSS-derived ZTD from the Nevada Geodetic Laboratory at nine stations in Uganda, Kenya, and Tanzania gave a mean RMSE of 21.99 mm and a mean bias of -5.30 mm. Compared with GTrop and GPT3, ALLSSA reduced the mean RMSE by approximately 29-32% and the absolute mean bias by approximately 48-62%. The improvement was not uniform across the region. It was greatest in the equatorial sub-region, where ZTD exhibits a bimodal seasonal cycle, and became smaller farther south where the temporal behaviour is predominantly annual. Seasonal bias showed a less consistent pattern, with GPT3 or GTrop producing smaller absolute biases under some conditions despite their larger RMSE. The results show that allowing the temporal representation of ZTD to adapt to the frequencies present in the regional signal provides an advantage over fixed-harmonic empirical models, particularly where seasonal atmospheric variability is more complex. The developed approach therefore provides a basis for improved regional ZTD estimation over East Africa. | |
| dc.description.sponsorship | Dr. Richard Cliffe Ssenyunzi ; Assoc Prof. Saphina Biira ; Busitema University | |
| dc.identifier.citation | Matsukuni, N. (2026). Development and validation of an anti-leakage least-squares spectral analysis-based zenith tropospheric delay model for the East African region. [Unpublished dissertation]. Busitema University. | |
| dc.identifier.uri | https://bdears.busitema.ac.ug/handle/123456789/9518 | |
| dc.language.iso | en | |
| dc.publisher | Busitema University | |
| dc.title | Development and validation of an anti-leakage least-squares spectral analysis-based zenith tropospheric delay model for the East African region. | |
| dc.type | Other |