Gregory L. West and W. James Steenburgh Spurious Grid-Scale Convection in the North American Regional Reanalysis Gregory L. West and W. James Steenburgh Department of Meteorology and Cooperative Institute for Regional Prediction, University of Utah 1. Introduction While developing a cyclone climatology, we discovered spurious grid-scale convection (SGSC) in the North American Regional Reanalysis (NARR). Since ~25% of papers published in AMS journals in the last 5 years use reanalysis data, this study examines the frequency, distribution, and characteristics of NARR SGSC. The NARR Assimilates data and combines with Eta 3-h forecasts using EDAS 32km/45 layer, North American Domain 1979-2004, long term consistent dataset Betts-Miller-Janjić convective parameterization (Janjić 1990, Janjić 1994) Zhao microphysics (Zhao and Carr 1997, Zhao et al. 1997) 4. SGSC Example: Antecedent Conditions Precipitation produced by the convective parameterization at point A agrees with radar. Spurious area of grid-scale precipitation produced at point B, which develops into SGSC over the subsequent 3 h. 0300 UTC Radar 0300 UTC Convective precipitation 0300 UTC Grid-scale precipitation 6. Distribution of SGSC in the NARR 1981: 62 SGSC events distributed over North America and southern Oceans. 1997: 95 SGSC events distributed over North America and southern Oceans. 2003: 2235 SGSC events, 92% of them over the oceans, only 8% over North America. 2. What is Spurious Grid-Scale Convection? Occasionally the convective parameterization in models is unable to remove instability faster than the rate at which it’s being created through ascent and saturation. The instability is aliased to the grid-scale, resulting in grid-scale ascent, saturation, and precipitation. 5. SGSC Example: Maturity The grid-scale instability continues, resulting in an intense grid-scale updraft. CMORPH Processing Error In January 2003, the NARR switched from CMAP to CMORPH precipitation data over the oceans. A processing error with the CMORPH data led to incorrect distribution/amounts of precipitation from 8°N to ~40°N, where CMORPH data was assimilated, causing widespread SGSC events. The error has been corrected, and the NARR is being rerun from 2003-present. a) 0300 UTC29 Jun 2003 Grid-scale precipitation b) 0600 UTC29 Jun 2003 Convective precipitation c) 0600 UTC 29 Jun 2003 d) 1200 UTC 1 Mar 2003 3. Development of Moist Instability Although synoptic environments in which SGSC forms vary, it is typically preceded by a near saturated conditionally unstable layer, or a moist absolutely unstable layer. 7. Summary/Conclusions SGSC is present in the NARR ~100 events/yr prior to 2003 Most events persist for ~6 h, ~2-5% persist >6 h Characterized by maxima in grid-scale precipitation (with erroneous scale, location, and intensity), vertical velocity, relative humidity, mid-level equivalent potential temperature (qe), precipitable water, mid to low-level absolute vorticity, and the presence of CAPE Occasionally accompanied by low-level cold pools and pressure/geopotential height anomalies Probably little or no effect on long-term means SGSC could affect studies examining extreme events, cases, or that use automated searches to identify phenomena, due to spurious extrema present in certain fields 0300 UTC 1 Mar 2003 (a,b) Virtually all precipitation is produced by the grid-scale scheme. Over Kansas, the convective parameterization removes sufficient instability, and precipitation is produced by the convective parameterization only. (c) SGSC features large vertical velocity maxima, sometimes >100 cm s-1, locally high qe air, and relative humidity maxima. (d) SGSC is occasionally accompanied by low-level cold pools and high pressure maxima resulting from sub-cloud evaporative cooling. Precipitable water and low to mid-level vorticity maxima also usually exist (not shown). An example from the NARR of a nearly moist absolutely unstable layer from 900-650 hPa. An example of a moist absolutely unstable layer in a sounding (Bryan and Fritsch 2000).