1.1.

Characteristics of GOES Imagers

The two Earth viewing sensors on the current GOES series are the imager and the sounder.⁷ The nominal field of view is approximately 1 km at nadir for the imager visible band, although the detectors are oversampled in the east-west direction.⁷ The technical aspects can be found in the GOES N/O/P data book.⁸ Many operational products from these sensors are derived, such as the cloud products (height, properties, etc.), atmospheric motion, biomass burning, smoke, and surface properties (e.g., land surface temperature). Current plans call for GOES-14 to become an operational satellite in 2015. The GOES-14 imager has five spectral bands (Fig. 1); see Fig. 6.1 of Ref. 3 for a plot of the visible spectral response function. The current GOES imager routine operational schedule scans a full disk (FD) image every 3 h as well as a CONUS or “extended Northern Hemisphere” sector as often as every 15 min. In contrast, for one scan mode, the ABI on the GOES-R series will provide FD images every 15 min, plus CONUS images every 5 min and mesoscale images every half minute (or two locations of approximately $1000\mathrm{km}\times 1000\mathrm{km}$ every minute).⁹ Hence, the special 1-min imagery of GOES-14 foreshadows what will be routinely available from the ABI on the next generation geostationary satellite series (GOES-R), scheduled to be launch ready in 2016, and operational in early 2017. GOES-S, the second in the series, is currently slated to launch in 2017 and become operational in 2020. A second possible scan mode is that the ABI only scan FD images, but do so every 5 min. The ABI will be better than the current imager in a number of significant ways. For example, faster coverage rate, improved spatial resolution (by a factor of four), on-orbit calibration of all bands, improved bit depth, more spectral bands (by over a factor of three), and improved Image Navigation and Registration.⁹ This means the imagery from the ABI will not have the gaps that the current SRSOR has when a FD is scanned. Nominally, the period for a given low Earth orbiting satellite is approximately 90 min, although there are multiple polar-orbiters. There is little overlap between orbits over the mid-latitudes and tropics so these observations primarily provide snapshots of the atmospheric state and cannot capture the rapid evolution of phenomena as is routinely captured by satellites in a geostationary orbit.

Fig. 1

GOES-14 imager spectral response functions (SRF) for the infrared (IR) bands. For the reference, a high-spectral resolution Earth-emitted spectrum (for the US standard atmosphere) is also plotted. The visible band (1) centered at 0.63 μm is not shown.

1.2.

Super Rapid Scan Operations for GOES-R

The GOES-R imager will be able to routinely take mesoscale images every 30 s (or swapping between observations of two widely separated phenomena every minute—e.g., an Atlantic hurricane and a Great Plains supercell tornadic storm). These images from the ABI can be acquired even while concurrently scanning larger areas, such as the continental United States and the FD. In general, the SRSOR schedule from GOES-14 consisted of 26 images every 30 min, with the exception of the 30-min gap every 3 h during FD images and during stray light (SL) testing (04:45 to 09:14 UTC). There was no SL outage during the late October collection of Hurricane Sandy (known as “Super Storm Sandy” in some media outlets) data, since the special scans needed for the SL correction algorithm had already been collected. The SRSOR allows for more rapid imaging than the standard Rapid Scan Operations (RSO) or even the SRSO. While the SRSOR scans an area nominally each minute, the RSO is only as fine as 5 min (and at times only 10 or 30 min). The SRSO has 1-min imagery periods that are broken up by other longer scans of approximately 5, 9, and 30 min. All these schedules require a FD imaged every 3 h, which leaves the 30-min gaps. The 1-min sectors from the current GOES imagers are approximately $1100\mathrm{km}\times 1900\mathrm{km}$ over the mid-CONUS (Fig. 2). More information on these operational scans can be found at NOAA’s Office for Satellite and Product Operations (OSPO) web page: http://www.ospo.noaa.gov/Operations/GOES/schedules.html, while http://www.ospo.noaa.gov/Operations/GOES/14/srsor.html lists the scans associated with the SRSOR 2012 experiment. For most of the experiment, GOES-14 was positioned at a longitude of 105 deg west, although the images of Hurricane Sandy were scanned from approximately 90 deg west. The location of the scanned area was determined the preceding afternoon, based on a number of factors such as the SPC outlooks. These coordinates were then communicated with the satellite operators. In general, the same schedule and center point were operated for 24 h.

Fig. 2

GOES-14 imager sample displaying the spatial coverage during 1-min imaging. Image start time of 20:15 UTC on August 16, 2012.

The schedule that was run during the SRSOR test varied due to a number of factors, including the potential weather, other satellite schedules, and out-of-storage planned testing (Table 1). The “default” schedules allowed for testing of changes to the routine schedules. The Continuous East Full Disk is a schedule that takes successive FDs, nominally every 30 min. More information, such as the daily schedule, coverage area, and satellite animations, on the SRSOR in 2012 can be found at the University of Wisconsin-Madison’s Cooperative Institute for Meteorological Satellite Studies (CIMSS) web page: http://cimss.ssec.wisc.edu/goes/srsor/GOES-14_SRSOR.html. Scientists at the Cooperative Institute for Research in the Atmosphere collected the SRSOR data in the real time, converted it to the proper format, and sent it to the SPC in Norman, Oklahoma, where it was displayed in their National Advanced Weather Information Processing System. This supplemental real-time 1-min data feed assisted the SPC forecasters by providing more frequent scans of preconvective clouds.

Table 1

Starting day (with ordinal day number), schedules, and location with SRSOR from GOES-14 during 2012.

As can be seen in Table 1, GOES-14 data collection continued until September 24, 2012, at 10:14 UTC when GOES-14 was needed for the operational service during a major anomaly (outage) of GOES-13. From that moment until GOES-13 was restored to service October 18, GOES-14 operated as the operational east satellite. “The root cause of the GOES-13 anomaly unusual large Filter Wheel vibration as the root cause for the sounder east/west scan motor overload. The large vibration disturbance was further transmitted onto the imager/sounder optical bench which was also affecting imager East/West scan control system.” (NOAA National Environmental Satellite, Data, and Information Service (NESDIS) memo: “GOES-13 Anomaly Report—September 23, 2012 Imager and Sounder Instrument Scan Motor Overload,” January 2013).

After GOES-13 returned to operations, the 1-min imaging from GOES-14 was not resumed. As Hurricane Sandy neared Jamaica, track forecasts increasingly suggested an impact with the United States East Coast was imminent. At the suggestion of researchers at CIMSS and NOAA NESDIS in Madison, Wisconsin, a request was sent to satellite operations warning of an “epic storm along the east coast” the following week. At this time, GOES-14 was slated to be placed back in storage mode within a few hours, when the request for special 1-min data from the GOES-14 imager of Hurricane Sandy through landfall was approved. NOAA NESDIS OSPO restarted the 1-min special SRSOR scans within 24 h. During the SRSOR time frame in 2012, GOES-14 collected approximately 30,000 images (or over 500 GB).

2.1.

Convection

The effect of “varying temporal sampling” was investigated in relation to several severe storms, showing the importance of rapid scanning, given that “storms evolve significantly on time scales shorter than the current GOES operational scan strategies.”¹⁰ Many applications regarding severe weather will benefit from more rapid imaging.¹¹

The GOES-14 imager captured convection developing extremely rapidly over Missouri and Arkansas on August 16, 2012, the first day of the SRSOR experiment (Fig. 3). A time sequence of 90 min of 10.7-μm IR window channel imagery demonstrates the rapid convective development of the storm in Arkansas circled in Fig. 3 and magnified in Fig. 4. Note the cold cloud top pixels associated with over shooting cloud tops. The corresponding high-temporal (1 min) resolution animation shows rapidly developing convection over the region (Video 1). According to the NOAA Storm Reports, the storm in Fig. 3 had a severe wind report with it at 18:27 UTC (“A tree was blown down onto Highway 10”). The NOAA Storm Reports for this day from the SPC can be found at http://www.spc.noaa.gov/climo/reports/120816_rpts.html.

Fig. 3

GOES-14 imager of one storm (near the center of the state of Arkansas, USA) for two of the spectral bands: visible band (a) and IR window (b). The image start time is 18:00 UTC on August 16, 2012.

Fig. 4

GOES-14 imager of one storm over a 90-min time window. The image series start time is 17:15 UTC on August 16, 2012, while the end time is 18:45 UTC. The IR window (band 4) is shown for each frame, with the coldest brightness temperatures (BTs) color-enhanced.

Video 1

GOES-14 imager visible band image animation of rapidly developing convection over the state of Arkansas, USA, on August 16, 2012 (QuickTime, 6 MB) [URL: http://dx.doi.org/10.1117/1.JRS.7.073462.1] (Also available at http://remotesensing.spiedigitallibrary.org/data/Journals/APPRES/926148/JARS_7_1_073462_ds001.mov).

The cloud top rapid cooling for the storm circled in Fig. 3 as seen in the 1-min data is not well represented when the data are subsampled in time to represent operational RSO (blue line) or normal operations (red line) (Fig. 5). The cloud-top emissivity (ordinate) represents the cloud compared to an optically thick cloud at the tropopause, where the emissivity “${\epsilon }_{\mathrm{tot}}$ is a measure of how close the cloud radiative center is to the tropopause.”¹⁰ This higher time resolution is especially important when monitoring rapidly developing storms. The time change in ${\epsilon }_{\mathrm{tot}}$ can be interpreted as a cloud-top cooling rate.

Fig. 5

Time series of maximum 10.7 μm cloud-top emissivity (ordinate) as seen from the GOES-14 imager of the one storm circled in Fig. 3 from August 16, 2012, as a function of time after $t_0$ (17:15 UTC). Note that the rapid cooling associated with the pulsating updrafts depicted with the data each minute.

On September 08, 2012, there was an EF0 tornado reported in the Queens borough of New York City at approximately 15 UTC: http://www.spc.noaa.gov/climo/reports/120908_rpts.html. GOES-13 (operational GOES-East at the time) also viewed this cell. Unfortunately, the touch-down occurred when GOES-13 was performing a routinely scheduled full-disk image and only collected imagery at 14:45 UTC, 15:15 UTC, and 15:45 UTC (housekeeping on GOES-13 was performed for 10 min starting at 15:34 UTC). Nevertheless, the animation of the visible and IR imagery shows the development of the tornadic cell along the convergence line. The 30-min gap between images, however, makes computation of convective predictors complementary to radar such as cloud-top cooling ineffective. GOES-14 was in SRSOR on September 8, and its (almost) every minute imagery captured the evolution of the tornadic cell very well. The 1-min imagery allows careful scrutiny of the cloud-top temperature as the cells evolve. The loop, from 14:45 UTC to 15:09 UTC, spans the time of tornadic activity, and the initial cloud-top cooling and subsequent anvil spreading is readily apparent (Video 2). Note that the GOES-13 is animated with the normal time between images (15 or 30 min), while the GOES-14 is in SRSOR (Video 3).

Video 2

GOES-14 imager visible (a) and IR window (b) bands animation of convection over the state of New York, USA, on September 08, 2012 (QuickTime, 4 MB) [URL: http://dx.doi.org/10.1117/1.JRS.7.073462.2] (Also available at http://remotesensing.spiedigitallibrary.org/data/Journals/APPRES/926148/JARS_7_1_073462_ds002.mov).

Video 3

GOES-13 (a) and GOES-14 (b) imager visible band image animation of convection over the state of New York, USA, on September 08, 2012 (QuickTime, 6 MB) [URL: http://dx.doi.org/10.1117/1.JRS.7.073462.3] (Also available at http://remotesensing.spiedigitallibrary.org/data/Journals/APPRES/926148/JARS_7_1_073462_ds003.mov).

Total lightning flash detections from northern Alabama Lightning Mapping Array (NA-LMA) and Earth Networks Total Lightning Network (ENTLN) were monitored at 1-min intervals within 15 km of the center of an overshooting cloud-top producing storm cell south of Huntsville, Alabama, on September 2, 2012 (Fig. 6). OTs were detected by the GOES-R algorithm for 30 consecutive GOES scans, though an OT was evident in visible imagery prior to 23:33 UTC. Radar vertically integrated liquid (VIL) and enhanced echo top height were analyzed from the Huntsville and Birmingham, Alabama, Weather Surveillance Radar 1988, Doppler. The ENTLN network did not transmit valid data for 3 min from 00:00 to 00:03 UTC (i.e., 24:00 to 24:03 on the ordinate) on the 3rd. Lightning activity began to increase in association with (1) IR cloud-top cooling and ascending cloud top height and (2) an increase in radar echo top height and VIL. The NA-LMA and ENLTN lightning time trends follow each other quite well. Lightning maximized near the time of a severe wind report at 23:56 UTC. The minimum IR brightness temperature (BT) preceded the maximum flash rate by approximately 7 min, though a secondary maximum in echo top and IR BT occurred at the time of severe wind. Although detection of continuous OTs for an extended period did not clearly indicate a particular time of significant weather, the combined presence of continuous OTs, high/increasing VIL, and high/increasing flash rate may be indicative of impending high impact weather and may improve the forecaster’s situational awareness.

Fig. 6

An analysis of the co-evolution of GOES-14 SRSOR, WSR-88D, and total lightning observations for an individual storm cell over northern Alabama on September 2–3, 2012. (a) GOES-14 SRSOR minimum IR BT within the storm cell, GOES-R overshooting cloud top (OT) detection algorithm output, and cloud-top height derived from the length of shadow produced by OT penetration above the surrounding anvil combined with the 16.1 km equilibrium level from the Birmingham, Alabama, sounding at 00 UTC on 3 September. (b) Total lightning flash detections from the NA-LMA and ENTLN within 15 km of the storm cell center. (c) The maximum WSR-88D derived vertically integrated liquid and precipitation echo top height within the storm cell.

2.2.

Fires/Biomass Burning

The geostationary perspective allows one to monitor the diurnal signatures of a number of phenomena, including wildfires. The SRSOR is a great improvement over the operational RSO (Ref. 12). The “start” of a fire is observed in both the visible loop (Video 4) and shortwave IR window loop (Video 5). Both spectral bands also show the smoke plumes from the fires. It should be noted that the BT of some fires is fairly constant over time, while other fires have widely varying BTs. Of course rapid scan images of fires can help monitor these changes and understand how they are associated with changes of wind speed/direction, moisture, fuel loads, etc. This may have implications in air quality models, to make sure the fire properties that are being assimilated are not aliased. How the same fire can rapidly change is demonstrated with GOES-12 RSO from 15:15 UTC May 9, 2007, to 04:15 UTC May 11, 2007. This is from the Bugaboo-Scrub fire in Georgia, USA. As can be seen in Fig. 7, the BT of the 3.9-μm band varies tens of Kelvin over just minutes. The 10.7-μm band does not show these rapid changes. The 3.9-μm band saturates at approximately 342 K for this imager, yet the similar band on the ABI at will be increased to 400 K. So, the ABI will improve not just the dynamic range, but also the spatio-temporal attributes.

Video 4

GOES-14 imager visible band image animation of smoke over the states of Oregon, Washington, and Idaho, USA, on September 9, 2012 (QuickTime, 8 MB) [URL: http://dx.doi.org/10.1117/1.JRS.7.073462.4] (Also available at http://remotesensing.spiedigitallibrary.org/data/Journals/APPRES/926148/JARS_7_1_073462_ds004.mov).

Video 5

GOES-14 imager near-infrared band 2 (3.9 μm) image animation of fires and smoke over the states of Oregon, Washington, and Idaho, USA, on September 9, 2012. The hottest pixels are color-enhanced (QuickTime, 3 MB) [URL: http://dx.doi.org/10.1117/1.JRS.7.073462.5] (Also available at http://remotesensing.spiedigitallibrary.org/data/Journals/APPRES/926148/JARS_7_1_073462_ds005.mov).

Fig. 7

Time series of the Bugaboo-Scrub fire in Georgia, USA, from 15:15 UTC May 9, 2007, to 04:15 UTC May 11, 2007. The IR window BT values are plotted, where $T$ is the value of the hot spot, TB is the background value and 4 and 11 refer to the 3.9- and 10.7-μm bands, respectively. TOA is the top of the atmosphere solar flux.

2.3.

Hurricanes

GOES-14 SRSOR imagery of four tropical cyclones were acquired in 2012: Helene, Isaac, Leslie, and Sandy. Due to the many days of SRSOR data and the catastrophic impact of Hurricane Sandy on the most populous areas of the east coast of the United States, this case is of the most interest. The multiday 1-min interval scans were collected during Hurricane Sandy, from October 25, 2012, at 17:45 UTC to October 31, 2012, at 22:41 UTC, roughly centered on the storm. Each afternoon the expected coordinates of the storm for the following day were communicated by NOAA NESDIS scientists to the satellite operators for subsequent data collection (Fig. 8).

Fig. 8

GOES-14 imager coverage each day of Sandy during the experimental SRSOR. The images have been remapped onto a Mercator projection.

The SRSOR scans from the GOES-14 imager were the most rapid scan images of a hurricane ever recorded (Video 6). Over 7000 images of Hurricane Sandy were acquired from GOES-14. The loops were shown on the Weather Channel, CNN, NBC, Reuters, and other media outlets, in addition to being used experimentally by NOAA [including OPC, Weather Prediction Center (WPC), NHC, NWS Eastern Region, and NWS Southern Region]. Some of the loops were also posted on YouTube and collectively received over 1 million views. A high-definition YouTube animation of the visible band during daylight hours was posted: http://youtu.be/f2tPHiMAB5U?hd=1 and http://youtu.be/p72XhZT6KdA?hd=1 (the second link is set to play the loop back in 1 min).

Video 6

GOES-14 imager visible band image animation of Hurricane Sandy from 25 October to 31 October 2012 (QuickTime, 43 MB) [URL: http://dx.doi.org/10.1117/1.JRS.7.073462.6] (Also available at http://remotesensing.spiedigitallibrary.org/data/Journals/APPRES/926148/JARS_7_1_073462_ds006.mov).

The coverage location varies each day following the storm. The animation of the visible band (Video 6) only shows the sunlit hours. On several of the days, the hurricane animations depict wind shear of approximately 180 deg. A more detailed analysis of Hurricane Sandy can be found in Folmer et al.¹³

These nonoperational GOES-14 data of Hurricane Sandy were not acquired by the Comprehensive Large Array-data Stewardship System (CLASS) in real time. So that these unique datasets might be more widely available, the Space Science and Engineering Center (SSEC) Data Center transferred the GOES Variable (GVAR) data to the CLASS. The total amount of Hurricane Sandy data transferred was on the order of 98 GB. These data, along with other SRSOR data from both 2012 and 2013, can be accessed at CLASS or the SSEC Data Center or other sites.

2.4.

Smoke/Cloud Interactions

While super rapid scan information is most needed to monitor convection that can provide earlier warning for severe weather, it is also useful for monitoring the complex interaction between many other phenomena, such as smoke, clouds, and precipitation. One example shows smoke that appears to suppress the growth of a cumulus cloud field because it reduces the solar insolation necessary to force the convection (Video 7). Smoke is also hypothesized to affect the microphysical processes within clouds, leading to a decrease in precipitation efficiency.¹⁴

Video 7

GOES-14 imager visible band image animation of smoke/cloud interactions from August 17, 2012 (QuickTime, 3 MB) [URL: http://dx.doi.org/10.1117/1.JRS.7.073462.7] (Also available at http://remotesensing.spiedigitallibrary.org/data/Journals/APPRES/926148/JARS_7_1_073462_ds007.mov).

The example in Video 7 highlights how unique 1-min imagery provides visual evidence of what is happening physically in the atmosphere and allows scientists to visualize fine-scale motions. With 15-min imagery, it is difficult to visualize how the smoke interacts with the cumulus cloud field, but in Video 7 the interaction becomes more obvious and provides scientists with a unique perspective.