Evaluation of the North American Regional Reanalysis (NARR) precipitation fields in a topographically complex domain.

The North American Regional Reanalysis (NARR) precipitation product was evaluated using station observations and catchment water yield in British Columbia (BC), Canada, at inter-annual, monthly, and daily time scales. A structural break occurred in 2003, associated with exclusion of Canadian precipitation gauge data from NARR's data assimilation process beginning in that year. The NARR product under-predicted precipitation in mountainous regions, over-predicted in the northern region of BC's Interior Plateau, and catchment-averaged NARR precipitation was less than observed water yield in coastal BC. The product was unable to reproduce even the seasonal pattern at three stations. This study highlights uncertainties associated with the NARR precipitation product, and presents a cautionary tale that is likely relevant not only to the application of NARR in BC, but may be relevant for other re-analysis products and other regions with complex topography and sparse station networks. [ABSTRACT FROM AUTHOR]