SPATIOTEMPORAL ANALYSIS OF RADAR-ESTIMATED PRECIPITATION DURING THE BOREAS SUMMER 1994 FIELD CAMPAIGNS

A truck-mounted C-band precipitation radar was operated in the Boreal Ecosystem-Atmosphere Study southern study area (SSA) north of Prince A lbert, Saskatchewan, for the duration of the intensive field campaigns , May to September 1994, Radar reflectivities were transformed to rain fall intensities using the Marshal-Palmer relationship at 10-min inter vals for 2 x 2 km pixels. Hourly accumulations of this product are ana lyzed for an area that coincides roughly with the southern modeling su barea (SMSA), a 50 x 40 km rectangle inside the SSA where the greatest focus of field activities took place. For validation, rainfall measur ements from a network of 12 tipping bucket and Belfort gauges are used . Although there are significant differences between the rainfall meas urements of single gauges and the overlying radar pixels, the area-ave raged (SMSA) radar rainfall coincides fairly well with the average rai nfall at the gauges. The conditional mean rainfall intensity over the SSA is shown to be closely related to its standard deviation, whereas the area-averaged rain rate is related to its frequency of occurrence and rainfall coverage by power laws. The fractional area covered by ra infall above certain thresholds follows a lognormal distribution. It i s demonstrated how such relationships can be used by modelers to deter mine the sub-grid-scale characteristics of precipitation. Using a spat ially distributed hydrology-vegetation model, the effect of the spatia l resolution of the rainfall fields on moisture and energy fluxes betw een the surface and the atmosphere is investigated. There are large di fferences in the modeled fluxes if only the rainfall information from the gauge network is used to produce model precipitation field inputs, as compared to using the 2 x 2 km resolution radar rainfall fields. U sing aggregated radar precipitation at decreasing resolutions results in increasing differences of the area-averaged fluxes from the 2 x 2 k m base run. In general, the mean absolute differences for area-average d latent and sensible heat fluxes and surface runoff are less than 10% . Locally though, the effect of using lower-resolution rainfall input fields on surface energy fluxes can be quite large.