Toward reliable water resource assessment: Correcting bias in satellite-based rainfall and model-derived discharge in Kenya’s Central Highlands

Felicia Yeboah, Afua Owusu, Mansoor Leh, Komlavi Akpoti, Kirubel Mekonnen, Elias Muluken Adamseged, Emma Odera and Naga Velpuri

PLOS Water, 2026, vol. 5, issue 9, 1-38

Abstract: Reliable precipitation and streamflow data are essential for hydrological modelling and water resources management, yet observational networks in many African regions remain sparse. This study evaluated five satellite precipitation products (SPPs) CHIRPS, PERSIANN-CCS, PERSIANN-PDIR, TAMSAT, and MSWEP and modelled stream-discharge data from VegET–mizuRoute model against in situ observations across the Central Highlands of Kenya for 2012–2021. Performance was assessed at eight precipitation and six discharge stations using pairwise statistics (RMSE, NSE, CC, RMB) and categorical metrics (POD, FAR) on a monthly scale. Four bias correction techniques (Least Squares, Quantile Mapping, Delta Method, Random Forest) were assessed to enhance the utility of the best-performing SPP and the modelled discharge data. CHIRPS and MSWEP consistently outperformed the other products: CHIRPS achieved RMSE of 34.85–134.26 mm, NSE of −0.94 to 0.44, and CC of 0.52–0.87, while MSWEP recorded RMSE of 24.17–125.66 mm, NSE of −0.91 to 0.65, and CC of 0.45–0.84. The VegET–mizuRoute model reproduced seasonal flow dynamics but showed station-dependent biases, with NSE from −2.28 to 0.23, RMB from −0.59 to 0.77, and CC from 0.46 to 0.71. Following bias correction, Random Forest yielded substantial improvements for both precipitation and discharge, underscoring the capacity of machine learning to address complex, nonlinear biases. It raised precipitation NSE to a consistently positive 0.85–0.95 across all eight stations while halving RMSE at most sites (9.05–46.23 mm). For discharge, Random Forest improved NSE at most stations to 0.80–0.88, with CC rising to 0.92–0.95, except at one station. Our findings confirm that systematic bias remains a dominant error source in global hydrological products in this region, and that machine learning-based bias correction is essential for the operational reliability and accuracy of these datasets for water management and flood applications.

Date: 2026
References: Add references at CitEc
Citations:

Downloads: (external link)
https://journals.plos.org/water/article?id=10.1371/journal.pwat.0000623 (text/html)
https://journals.plos.org/water/article/file?id=10 ... 00623&type=printable (application/pdf)

Related works:
This item may be available elsewhere in EconPapers: Search for items with the same title.

Export reference: BibTeX RIS (EndNote, ProCite, RefMan) HTML/Text

Persistent link: https://EconPapers.repec.org/RePEc:plo:pwat00:0000623

DOI: 10.1371/journal.pwat.0000623

Access Statistics for this article

More articles in PLOS Water from Public Library of Science
Bibliographic data for series maintained by water ().