Contributions of Dense Pressure Observations to Mesoscale Analyses and Forecasts

In an effort to improve the analysis and short-term forecast of mesoscale phenomena, the assimilation of dense surface pressure observations was examined using an ensemble Kalman filter. Over the Pacific Northwest, an order of magnitude more pressure observations beyond the METAR network were obtained. A bias correction procedure was developed to improve the usability of these observations, and this procedure is shown to be effective at reducing errors in the analysis and forecasts after assimilating bias-corrected observations. Comparisons of assimilating different densities of pressure observations indicate that additional pressure observations are able to reduce the domain-averaged surface pressure analysis errors by a statistically significant amount. The adjustments made by the additional pressure observations are localized to known mesoscale phenomena, and persist for several hours into subsequent forecasts from the new analyses. Forecasts after assimilating dense pressure observations led to better timing of frontal passages and improved the forecast of a localized convective band. Three-hour ensemble cycling experiments over a month-long period showed that assimilating more dense pressure observations reduced domain-averaged three-hour forecast errors in surface pressure and surface and upper-level wind and temperature fields by statistically significant amounts. Furthermore, the assimilation of three-hour pressure tendency observations is also seen to yield three-hour forecast errors of surface fields that are competitive with errors when assimilating dense pressure observations, suggesting that pressure tendency can be a viable alternative to assimilating raw pressure without the need for bias correction.