Time-varying quadruple collocation for enhanced satellite and reanalysis precipitation data error estimation and integration

Autor: Angelika L. Alcantara, Kuk-Hyun Ahn
Jazyk: angličtina
Rok vydání: 2024
Předmět:
Zdroj: International Journal of Applied Earth Observations and Geoinformation, Vol 127, Iss , Pp 103692- (2024)
Druh dokumentu: article
ISSN: 1569-8432
93502958
DOI: 10.1016/j.jag.2024.103692
Popis: For the past few years, there have been improvements in collocation approaches that do not rely on in-situ observations to estimate data uncertainties. These approaches have been applied in various applications, including data fusion. A notable development is the Quadruple Collocation (QC) approach, which effectively accounts for error cross-correlation between datasets, setting it apart from previous collocation methods. Nevertheless, collocation approaches still presume static errors over time, neglecting the dynamic nature of precipitation data errors. To address this limitation, this study proposes the time-varying quadruple collocation (TV-QC) approach to enhance the estimation of satellite and reanalysis precipitation data uncertainties and to facilitate more effective data fusion. To achieve this, we first compare the performance of time-variant errors (TVE) with time-invariant errors (TIE), and we then assess them against gauge-based errors. Subsequently, we utilize the errors obtained from TV-QC for data merging. Our findings highlight that TVE not only differ significantly from their time-invariant counterparts but also display notably stronger correlations with ground-based errors. Furthermore, our analysis reveals differences between the performances of the individual products. GPM consistently exhibits superior accuracy, while the PERSIANN and ERA5 datasets consistently exhibit weaker and less consistent performance across diverse seasons and climate zones. Additionally, utilizing TV-QC for data fusion leads to the development of a new precipitation product for the Philippines, “PHIL-ADVANCE”, that surpasses the parent datasets in terms of performance. Overall, the integration of time-varying collocation and the QC technique enhances our understanding of precipitation uncertainties and facilitates improved data fusion.
Databáze: Directory of Open Access Journals
Externí odkaz: