Using 1-min Resolution GOES Observations to Diagnose Earlier Signs of Convective Initiation, Determine Cumulus Cloud Updraft Characteristics, and Infer In-Cloud Processes

Since 2012, periodic collections of 1-min resolution super rapid scan operations for GOES-R (SRSOR) GOES-14 observations were recorded for significant weather events over the U.S. as well as specified time periods to mimic the data collection of GOES-R. Monitoring the SRSO imagery allows for the observation of evolving cumulus clouds over a short time period. Cloud top height analyses can show where cloud tops are relative to the level of free convection as determined through the Rapid Refresh (RAP) model. Early observations show that many convective clouds reach and exceed the level of free convection (LFC) however do not end up producing precipitation. Convective clouds typically have to reach a height that is at least 5-10 degrees colder than the level of the LFC to be considered a candidate for convective rainfall. This study will show the development of an algorithm that will detect early signs of convective initiation using the SRSO observations.

In addition to the development of early CI detection, another goal of studying the 1-minute resolution GOES observations is to evaluate the sensitivity of the observations of growing cumulus clouds, specifically, to determine what components of the cumulus processes occurring do these data describe. Since these SRSOR data are collected by GOES-14 at 4 km resolution in the infrared, questions arise as to the character of the signatures of cumulus cloud growth that would be expected to be seen, specifically when one examines the vertical momentum equation appropriate for moist convection. Such signatures include updraft acceleration as a function of the instability profile, hydrometeor loading, and latent heat release.

To date, a total of 71 separate cumulus cloud updrafts were collected over 33 to 152 minute periods on five dates and regions in 2012, 2013 and 2014. Cloud top temperatures (TBs) in the 10.7-µm “window” channel were cataloged as cumulus clouds evolved from the “fair weather” stage to mature cumulonimbus, or a cloud that eventually possessed a new anvil. GOES 3.9 µm channel derived reflectance (ref₃₉) were also available every 1-minute in SRSOR. The ref₃₉ data are used to infer cloud top glaciation, whereas reflectance values falling to below ~9% when 10.7 µm cloud top TBs are <273 K are highly correlated with the transition of cloud water particles to ice crystals (Lindsey et al. 2006). Lastly, proximity soundings of temperature, dew point, mixing ratio and of estimated convective available potential energy (CAPE) were collected from 13 km resolution NOAA Rapid Refresh (RAP) model 0000 hour analysis grids. An analyzed LFC from the RAP model was used in the computation of CAPE. The profile of CAPE was first used to compute the incremental amount of buoyancy an updraft was penetrating through (or using) for each 1-min of cloud growth. For each updraft, a 1-min vertical motion (w, ms^-1) was computed by simply noting the change in geometric height (using the RAP profile) of the updraft every 1 minute, assuming that cloud top TB is approximately the parcel temperature at cloud top (as typical for optically thick clouds), and that the parcel follows a moist adiabat.

Results have shown that the correlations between the accumulated CAPE to a given altitude and w often exceed 0.60, leading to the conclusion that the SRSOR data are observing aspects of cumulus clouds which have rarely been seen by geostationary satellites (with the possible exceptions being from early 3-min observations from GOES, pre-1980, and in 2013 data collections of 2.5 min data from Meteosat Second Generation). Evidence is shown on how updrafts grow, in general, with respect to the capping inversion, the freezing level, and the equilibrium level as anvils form. Further analysis will be presented of SRSOR-observed convection with 1-km resolution Weather Research and Forecasting (WRF) fields for the same time the cumulus clouds were sampled by GOES-14. The WRF model analysis is done as a means of drawing further conclusions on what components of active convection SRSOR data are actually observing. Given the results, discussion is provided as to how such information can be used within products that diagnose and nowcast near-term convective storm initiation as early as possible, future storm intensity, and lightning characteristics of convective clouds.