ASAP Convection Monitoring and Nowcasting Products: August 18, 2005 Kristopher Bedka*, John Mecikalski+, Wayne Feltz*, and Todd Berendes+ * Cooperative Institute for Meteorological Satellite Studies, UW-Madison + University of Alabama in Huntsville

Project and Product Description This case study demonstrates convective storm nowcasting and monitoring products developed within the NASA-sponsored Advanced Satellite Aviation-weather Products (ASAP) project and other SSEC/CIMSS research efforts for the August 18, 2005 Wisconsin tornadic thunderstorm ouutbreak.  The objective of the ASAP product is to introduce new satellite-based techniques and data sets to the FAA Aviation Weather Research Program (AWRP)-Product Development Teams (PDTs) to aid in their recognition and forecasting of weather-related aviation hazards.  The convective storm nowcasting work shown here represents one of the many facets of the ASAP project, see http://cimss.ssec.wisc.edu/asap/ for a full project description. The convective storm nowcasting products shown here utilize GOES-12 1 km VIS and 4-8 km IR data at 15 minute time resolution.  The IR data are remapped to the 1 km VIS resolution within the McIDAS software package in order to retain the highly detailed cloud structure that the VIS channel provides.  The VIS and IR data are both ingested into a convective cloud classification algorithm.  This algorithm is a based upon an unsupervised learning system that incorporates 1) VIS reflectance, 2) VIS texture (is a cloud top "lumpy" or uniform in appearance?, does the cloud feature have a distinct edge?), 3) IR brightness temperatures, and 4) IR channel differences.  Six categories of convectively induced clouds are identified by this algorithm: small immature Cu, mid-level "towering" Cu, mature deep Cu, overshooting convective cloud tops, thick cirrus anvil clouds, and thin cirrus.  We only process pixels classified as one of these 6 classes (there are other numerous other classes such as fog, stratus, and clear sky), thus saving a substantial amount of computational time and allowing for cloud monitoring and nowcasting over large geographic domains in near-real time.  Click here to view this and other ASAP convective storm monitoring and nowcasting products. The UW-CIMSS satellite-derived wind algorithm (Velden et al. 1997, 1998) is implemented here in a way that captures high-density mesoscale flow patterns.  These wind vectors are used to track cumulus cloud features back in time to identify cumulus cloud growth rates.  Cloud features that are identified as rapidly growing towering cumulus with tops that have dropped below freezing within the last 15-30 mins are flagged as a strong candidate for convective initiation 30-60 minutes in the future.  Convective initiation (CI) is defined here as the transition from below to above 35 dBZ radar reflectivity.  The intricacies of this convective storm initiation nowcasting process is fully described in a Mecikalski and Bedka (2005) Monthly Weather Review paper: (ftp://ftp.ssec.wisc.edu/asap/documents/papers/MWR_MecikalskiBedka_GOESConvInit.pdf) and the mesoscale wind processing is described in a Bedka and Mecikalski (2005) Journal of Applied Meteorology paper: (ftp://ftp.ssec.wisc.edu/asap/documents/papers/JAM_BedkaMecikalski_MesoscaleAMV.pdf).  A publication on this new convective cloud classification algorithm is forthcoming by Todd Berendes et al. (University of Alabama-Huntsville).  Contact Kris Bedka for more information on this product.

Analysis of the Pre-Storm Environment Mesoscale satellite winds and atmospheric stability parameters are shown in Figure 1 to illustrate the wind shear and instability present before the development of the tornadic thunderstorms that impacted south-central WI.  The mesoscale wind field shows veering winds with height across southern WI.  The flow was SSW below 800 mb but shifted to the west above this level, which aided in inducing rotation within the thunderstorms that later developed. Interesting to note is the contrast in 700-800 mb flow direction and speed between south-central and southeast WI.  A veering wind profile was present in South-Central WI, whereas the flow was closer to unidirectional in the eastern portion of the state.  Satellite-observed flow at jet stream level (100-300 mb) topped out at speeds of about 70 kts.  Ongoing ASAP work involves understanding the "error" characteristics of mesoscale winds relative to NOAA wind profiler and radiosonde measurements over the SGP ARM-CART site.  Contact Kris Bedka for more information on this research. The GOES single field of view (SFOV) convective available potential energy (CAPE) field shows that an axis of higher instability was present across southern WI.  This instability was produced by an axis of higher temperature and moisture induced by strong solar heating and low level advection (below 800 mb) by southwesterly wind flow ahead of the existing convection in western WI.  Maximum CAPE values were estimated near 2000 J/kg, which was evidently adequate to develop and maintain tornadic storms across southern WI, when balanced with the low level wind shear from the veering wind profile.

Click for larger images

Figure 1: (left) Mesoscale satellite-derived winds at 2132 UTC illustrating the complex flow present across the Upper Midwest. Only 25% of the total wind barbs are shown here so most vectors can be viewed clearly. (right) GOES-12 Sounder single field of view (SFOV) convective available potential energy (CAPE) showing the axis of instability in southern WI.

Convective Event Evolution   There are 2 types of java applets below, one combining GOES-12 1km VIS imagery with convective cloud classification and satellite winds with fading capability, and the other showing a composite radar reflectivity mosaic with time stamps closest to the times of GOES-12 imagery.  The event is broken up into 3 time periods, one from 2002-2145, 2155-2245, and 2255-2345 UTC because the java applet would cause your machine to run out of memory.  As convection develops, the convective cloud mask shows a general upscale progression from small cumulus through towering cumulus up to deep convection and overshooting top.  A comparison of the radar mosaic with the convective cloud mask shows that precipitation often begins when a cloud reaches the mid-level towering cumulus classification.   The classifier captures the active updraft portion of the cloud quite well with the overshooting top class.  This portion of a cloud should be avoided by aviation interests, as strong updrafts are responsible for producing convectively induced turbulence.  New NASA-sponsored research at SSEC/CIMSS will focus on exploring satellite observations of turbulent convective clouds in greater detail using MODIS imagery. New convective storm growth is evident within 2 time periods in the animations below, 2115-2145 UTC in Grant and Iowa counties and 2230-2300 UTC in Lafayette/Iowa counties and over Lake Michigan.  The full disk scan at 2100 UTC prevents us from producing CI nowcasts for the 2115-2145 CI event, as the convective storm initiation nowcasting technique described above currently requires a 3 image sequence with 15 min (or less) resolution.  Thus, we will focus on the 2230-2300 CI event with imagery zoomed in over Southern Wisconsin. See this link if you get an "Out of Memory" error or messages such as "Applet AniS notinited" and "Loading Java Applet Failed..." on your java applet status bar. 2002-2145 UTC ASAP Imagery Fader/Animator 2002-2145 UTC Composite Reflectivity Mosaic Animation 2155-2245 UTC ASAP Imagery Fader/Animator 2155-2245 UTC Composite Reflectivity Mosaic Animation 2255-2345 UTC ASAP Imagery Fader/Animator 2255-2345 UTC Composite Reflectivity Mosaic Animation

Southern Wisconsin Convective Storm Initiation Table 2 shows the CI nowcasting criteria utilized to produce Figure 3a below.  These criteria were determined through a detailed comparison of satellite-observed cumulus signatures/time trends and future radar behavior for varying times of the year/day, regions of the the continental U. S., and dynamic/thermodynamic regimes.  A pixel is given 1 "point" for each CI interest field criteria that is met.  Pixels that meet  7 of 8 (or greater) criteria (shown in red in 3a) are considered to have high potential for future thunderstorm development; pixels that meet 5 or 6 criteria (green) have a lower potential. Convection nowcasting products from the 2225 UTC GOES-12 image are presented below in Figure 3a-b and 3e-h.  For this case, new convective storm development occurred quite rapidly over small areas, such that our satellite nowcasting products will only highlight the limited number of growing, newly glaciated convective cloud tops classified as mid-level "towering" cumulus (Fig 3f).  We focus on towering cumulus because their enhanced vertical development is indicative of the stronger forcing and vertical motion, which are vital for future precipitation production.  When compared to the circled regions of the 2225 UTC (Fig 3c) and 2245 UTC (Fig 3d), many of the red nowcast pixels in SW WI and over Lake Michigan do develop into cells with future reflectivity > 35 dBZ.  Nevertheless, much of the development over Lake Michigan appears to be missed by the Fig. 3a nowcasts.  Figure 3g shows a 15 minute cloud-top cooling rate estimate using the techniques describe above.  Essentially, a differencing is perfomed between the 10.7 micron image at 2225 UTC and 2210 UTC, incorporating the cloud motion vectors shown in Fig. 3b.  This product below shows that the cloud growth is captured across the entire line of convection that eventually developed over Lake Michigan.  But, since we only nowcast for towering cumulus (cyan) pixels, only a small portion of the convective line was captured within the nowcast.  Thus, the cloud-top cooling rate estimate can be used as a stand-alone product for monitoring and predicting future thunderstorm development. An examination of the wind vectors along the IL/WI border in Fig. 3b demonstrates that the atmospheric dynamics conducive for tornadic thunderstorm development were evident across southern WI.  Winds below 900 mb were from the SSE and veered to the SSW by 800-900 mb.  Above 800 mb, the flow veered to the WSW.  Not much speed shear was present, but the balance between the directional shear and instability were evidently sufficient for producing and maintaining long-lived tornadic storms.  It is quite remarkable that we can visualize these changing dynamics with height using exclusively current generation satellite technology.  This demonstration with the GOES-12 instrument shows that mesoscale analyses using future generation higher spatial (.5 to 2 km) and temporal resolution (5 minute) GOES-R ABI imagery will greatly improve our understanding and observation of severe thunderstorm environments.

Figure 2: Convective initiation nowcasting criteria used to produce Figure 3a.  See the description and the Mecikalski and Bedka (2005) MWR paper referenced above for a detailed explanation of the origins of these criteria.

 

Click for larger images

a)		b)
c)		d)
e)		f)
g)		h)

Figure 3: a) ASAP convective storm initiation nowcast product at 2225 UTC.  Red highlight high CI potential, green shows lower potential.  Ovals correspond to developing convective cells in Fig. 3c-d.  b) Mesoscale satellite-derived wind vectors. All vectors are shown. c) A composite radar reflectivity mosaic at 2225 UTC and d) 2245 UTC e) GOES-12 1 km VIS satellite imagery f) ASAP convective cloud classification product. g) 15 minute cloud top cooling rate estimate using the satellite winds shown in Fig 3b.  h) The 6.5-10.7 micron channel difference showing cloud tops near the tropopause (orange) versus low cumulus (blue).  This product is one of the interest fields shown above in Fig. 2. 2225 UTC Figure 3 ASAP Imagery Fader

References Bedka, K. M., and J. R. Mecikalski, 2005: Application of Satellite-Derived Atmospheric Motion Vectors for Estimating Mesoscale Flows, J. Appl. Meteor. In press. Mecikalski, J. R., and K. M. Bedka, 2005: Forecasting convective initiation by monitoring the evolution of moving cumulus in daytime GOES imagery. Mon. Wea. Rev. In press. Velden, C. S., C. M. Hayden, S. J. Nieman, W. P. Menzel, S. Wanzong, and J. S. Goerss, 1997: Upper-tropospheric winds derived from geostationary satellite water vapor observations. Bull. Amer. Meteor. Soc., 78, 173-195. Velden, C. S., T. L. Olander, and S. Wanzong, 1998: The impact of multispectral GOES-8 wind information on Atlantic tropical cyclone track forecasts in 1995. Part I: Dataset methodology, description, and case analysis. Mon. Wea. Rev., 126, 1202-1218.

Back to August 18 Case Study Page