PBCOR — Global bias-
State-of-the-art gauge-based climatologies — such as WorldClim, CHPclim, and CHELSA — seriously underestimate precipitation over most major mountain ranges. The Precipitation Bias CORrection (PBCOR) dataset provides global gap-free bias correction maps derived using streamflow observations from 9372 stations worldwide.
For more information, see the following open-access paper:
- Beck, H. E., T. R. McVicar, M. Zambrano-Bigiarini, C. Alvarez-Garret, O. M. Baez-Villanueva, J. Sheffield, D. Karger, and E. F. Wood: Bias correction of global high-resolution precipitation climatologies using streamflow observations from 9372 catchments, Journal of Climate, doi:10.1175/JCLI-D-19–0332.1 (2019).
The latest version (1.0) of the PBCOR dataset, including a readme with information about the files, can be downloaded here. By using the data in any publication you agree to cite the above-mentioned paper.
Frequently asked questions
Why didn’t you bias correct other popular precipitation datasets?
The three precipitation climatologies that we used as baseline (i) have a high resolution, (ii) incorporate a large number of gauge observations, and (iii) explicitly account for orographic effects. They can therefore be expected to provide more accurate climatological precipitation estimates than other datasets. Other datasets can be bias corrected by rescaling them to match one of the bias-corrected climatologies.
Can I use the correction factors for other versions of WorldClim, CHELSA, or CHPclim?
This is is not recommended as each version exhibits unique bias patterns.
Why don’t your bias-corrected precipitation climatologies agree with my gauge observations?
First, your gauge observations may underestimate precipitation due to gauge under-catch. Secondly, your (point scale) gauge observations may not be representative of the 0.05° grid-cells. Thirdly, the baseline climatologies may be wrong. Finally, our bias corrections may be wrong, due to (i) errors in the streamflow, potential evaporation, or catchment boundary data; (ii) uncertainty in Fu’s (1981) w parameter used to infer precipitation from streamflow; and (iii) uncertainty in the random forest regression of the bias correction factors.
Which of the three corrected climatologies is best?
That depends on the region under consideration. In general, one can assume that the climatology that incorporates the largest number of gauge observations for a particular region is best. Maps of the gauge observations incorporated in each climatology can be found in the corresponding publications (Fick and Hijmans, 2017, Karger et al., 2017, and Funk et al., 2015).
What about the bias corrections applied by the GPCP and GPCC datasets?
The GPCP and GPCC climatologies (i) have a fairly coarse 0.5° spatial resolution and (ii) were corrected for gauge under-catch by interpolation of correction factors derived from sparse and unevenly distributed station networks. They therefore do not adequately represent mountainous regions as illustrated in the PBCOR publication (Beck et al., 2019, Fig. 7).
The PBCOR dataset was developed by Hylke Beck (Princeton University, Princeton, NJ, USA) in collaboration with Eric Wood, Tim McVicar, Mauricio Zambrano-Bigiarini, Camilla Alvarez-Garret, Oscar Baez-Villanueva, Justin Sheffield, and Dirk Karger. The precipitation and potential evaporation dataset developers are thanked for producing and making available their datasets. The following organizations are thanked for providing streamflow data: the United States Geological Survey (USGS), the Global Runoff Data Centre (GRDC), the Brazilian Agencia Nacional de Aguas, EURO-FRIEND-Water, the European Commission Joint Research Centre (JRC), the Water Survey of Canada (WSC), the Australian Bureau of Meteorology (BoM), and the Chilean Center for Climate and Resilience Research (CR2, CONICYT/FONDAP/15110009).