An Analysis of Precipitation Isotope Distributions across Namibia Using Historical Data.

• Main Authors: Kudzai Farai Kaseke, Lixin Wang, Heike Wanke, Veronika Turewicz, Paul Koeniger
• Format: Article
• Language: English
• Published: Public Library of Science (PLoS) 2016-01-01
• Series: PLoS ONE
• Online Access: http://europepmc.org/articles/PMC4856410?pdf=render

Global precipitation isoscapes based on the Global Network for Isotopes in Precipitation (GNIP) network are an important toolset that aid our understanding of global hydrologic cycles. Although the GNIP database is instrumental in developing global isoscapes, data coverage in some regions of hydrological interest (e.g., drylands) is low or non-existent thus the accuracy and relevance of global isoscapes to these regions is debatable. Capitalizing on existing literature isotope data, we generated rainfall isoscapes for Namibia (dryland) using the cokriging method and compared it to a globally fitted isoscape (GFI) downscaled to country level. Results showed weak correlation between observed and predicted isotope values in the GFI model (r2 < 0.20) while the cokriging isoscape showed stronger correlation (r2 = 0.67). The general trend of the local cokriging isoscape is consistent with synoptic weather systems (i.e., influences from Atlantic Ocean maritime vapour, Indian Ocean maritime vapour, Zaire Air Boundary, the Intertropical Convergence Zone and Tropical Temperate Troughs) and topography affecting the region. However, because we used the unweighted approach in this method, due to data scarcity, the absolute values could be improved in future studies. A comparison of local meteoric water lines (LMWL) constructed from the cokriging and GFI suggested that the GFI model still reflects the global average even when downscaled. The cokriging LMWL was however more consistent with expectations for an arid environment. The results indicate that although not ideal, for data deficient regions such as many drylands, the unweighted cokriging approach using historical local data can be an alternative approach to modelling rainfall isoscapes that are more relevant to the local conditions compared to using downscaled global isoscapes.