Macro-Scale Correction of Precipitation Undercatch in the Midwest/Great Lakes Region

Abstract:

Precipitation gauge undercatch is a serious problem in the context of using observed meteorological data sets for hydrologic modeling studies in regions with cold winters, such as the Midwest. Attention to this matter is urgently needed to support hydroclimatological research efforts in the region. To support hydrologic modeling studies, a new hybrid gridded meteorological dataset at 1/16 degree resolution based on data from CO-OP station records, the U. S. Historical Climatology Network, the Historical Canadian Climate Database, and Precipitation Regression on Independent Slopes Method has been assembled over the Great Lakes and Midwest regions from 1915-2013 at daily time step. Preliminary hydrologic simulations results using the Variable Infiltration Capacity hydrology model with this hybrid gridded meteorological dataset showed that precipitation gauge undercatch was a very significant issue throughout the region, especially for winter snowfall and simulated streamflow, which were both grossly underpredicted. Correction of primary CO-OP station data is generally infeasible due to missing station meta data and lack of local-scale wind speed measurements. Instead, macro-scale post processing techniques were developed to adjust the regridded precipitation product from CO-OP station records from 1950-2013 forwards, accounting for undercatch as a function of regridded wind speed simulations obtained from NCAR Reanalysis. Comparisons of simulated and observed streamflow over seven river basins in the Midwest were used to evaluate the datasets constructed using different combinations of meteorological station inputs, with and without undercatch corrections. The comparisons show promise in producing corrected precipitation data sets from 1950-2013 for hydrologic modeling studies, with substantial improvements in streamflow simulation from the uncalibrated VIC model when gauge undercatch corrections are included.