The Mauna Loa volcano on the island of Hawaii erupted during the overnight hours early on 28 Nov 2022, for the first time since 1984. GOES-West imagery, from both GOES-17 (current operational GOES-West) and GOES-18 (to become operational GOES-West in January) captured the eruption, lava flow, and SO2/Ash output in great detail on the 28th.
Focusing first on all 16 single-bands from both GOES-17 (Fig 1) and GOES-18 (Fig 2), we see that the thermal signature associated with the lava flow is (amazingly) captured in all but only three of the ABI bands. The signature is apparent in all near-infrared and visible bands except for band 1, a testament to just how “hot” the signal was. In the IR, a signal is diagnosed in all but the two more sensitive water vapor bands. Comparing GOES-17 with GOES-18, the signals are very similar, aside from the degradation apparent in a few of the GOES-17 bands due to the cooling system issue.
Figure 1: 28 Nov 2022 GOES-17 ABI channels 1-16 over Hawaii.
Figure 2: 28 Nov 2022 GOES-18 ABI channels 1-16 over Hawaii. Preliminary non-operational data
Now combining multiple channels into a single image in order to track the evolution of the lava thermal signature, we see the rapid growth and expansion of the signal northeastward through the night, in addition to the aerosol output (Fig 2). The animation combines the Hi-Res Top image as the underlay, semi-transparent ch13 IR as the next layer to depict clouds, and a combination of semi-transparent ch-07 (yellow), ch06 (red), ch05 (white) and ch03 (blue). By combining these channel in this order, we are able to depict progressively hotter regions of the volcanic activity, ranging from yellow (coolest) to red to white to blue (hottest).
Figure 3: 28 Nov 2022 GOES-17 IR+NIR imagery. See text for details.
Now observing the volcanic aerosol emission aloft, the GOES-18 SO2 RGB clearly captures the SO2 output from the volcanic eruption among widespread cloud cover, as shades of bright yellow (Fig 4).
Figure 4: 28 Nov 2022 GOES-18 SO2 RGB over Hawaii. Preliminary non-operational data
The SO2 is similarly (slightly less obvious) detected in the Ash RGB. Ash however, is a little more difficult to diagnose given the presence of similarly appearing high clouds in the scene.
Figure 5: 28 Nov 2022 GOES-18 Ash RGB over Hawaii. Preliminary non-operational data
To help monitor the eruption and volcanic plumes, NWS Pacific Region HQ requested a GOES-West (-17) mesoscale sector for the next couple of days. Creating a similar animation as in Fig 3, but instead during the day and leveraging the 1-min imagery and Geocolor as the base layer, we can track evolution in even better detail. The hottest areas of the lava field can be observed, as well as the constant emission of ash and vapor.
Figure 6: 28 Nov 2022 GOES-17 1-min Geocolor+IR+NIR imagery.
SNPP and NOAA-20 also provided stunning imagery of the eruption overnight. The Day Night Band Near Constant Contrast Product revealed the very bright light emitted from the lava flow overnight (Fig 7).
Figure 7: 28 Nov 2022 NOAA-20 DNB-NCC product.
At the same time, the 750 m resolution M13 (~4 um) band captured the thermal signature associated with the eruption in better, quantitative detail (Fig 8). The imagery revealed a large area of very hot BTs, with pixels maxing out over 500K (226C, 440F).
Figure 8: 28 Nov 2022 NOAA-20 band M13 (~4 um). White is cooler BT, black to red is hotter BT.
Volcanic activity continued on the next day, with GOES-17 1-min Geocolor+IR+NIR imagery continuing to show convection above the volcano, and later in the day, reemergence of the large and hot hot spot. Also added to this animation is the ch13 – ch 11 band difference, with bright green values representing likely SO2 emission.
Figure 9: 29 Nov 2022 GOES-17 1-min Geocolor+IR+NIR+SO2 Diff imagery.
GOES-18, still undergoing post launch testing, began collecting 30-second imagery of the volcanic eruption on 30 Nov. The following GOES-18 30-sec Day Fire RGB imagery captures the real-time evolution of the heat signature associated with the lava area alongside clouds and surface vegetation characteristics (Fig 10).
Figure 10: 30 Nov 2022 GOES-18 30-second Fire RGB. GOES-18 data is preliminary and non-operational.
The lava field become relatively stationary with a consistent temperature according to GOES-17 imagery from the pre-dawn hours on the 1st through the day. Additionally, flow shifted such that the SO2 plume was advected south from the volcano.
Figure 11: 01 Dec 2022 GOES-West Volcano AWIPS procedure overnight into the daytime. Lava field hot spot is yellow to red to white (hotter), and SO2 plume is bright green.
A favorable large scale pattern set up across the Great Lake region to result in an impressive Lake Effect Snow Event for Upstate New York. NWS Forecasts of a considerable snow event were out days in advance, and would end of easily verifying. NOAA Satellite Imagery provided insight into this event, from the relevant large scale features down to the individual snow bands.
GOES Water Vapor Imagery, with an overlay of RAP 500 mb height contours, from late on the 16th through the first half of the 18th captured a series of shortwaves advancing through the region, helping to provide large scale forcing and additional moisture to the event (Fig 1). An initial shortwave exited the eastern Great Lakes early on the 17th. The second and more intense shortwave pushed across the eastern Great Lakes late on the 17th, and was easily diagnosed in satellite imagery with increasing moisture and cloudiness (light blue to white to green) apparent on the leading edge. In the wake of this wave, favorable westerly to southwesterly deep flow developed across the eastern Lakes, ahead of another shortwave entering the region early on the 18th.
Figure 1: 17-18 Nov 2022 GOES-East mid-level water vapor imagery.
GOES RGBs can be valuable tools to supplement radar data during Lake Effect Snow Events. Specifically, at night, Nighttime Microphysics RGB imagery helps to differentiate cloud types at night, including the mid-level and glaciated clouds associated with the LES bands (orange) from higher ice clouds (red) and lower liquid clouds (bright green). After sunrise and switching over to the Day Cloud Phase Distinction RGB, the mid-level glaciated clouds appear as yellow, while the lower-level stratus and liquid cloud (tops) are bright cyan and higher ice clouds are red. The animation in Fig 2 transitions from the nightMicro RGB to DCPD RGB, and helps to connect features and how they appear between the two. Also present in the animation is GLM Flash Extent Density, which revealed considerable lightning activity associated with the LES bands, helping to highlight the most intense convection and areas of snowfall. The organization of the LES bands is evident in the NightMicro imagery through the night on both eastern Lakes, and remain organized into the day on the 18th.
Figure 2: 18 Nov 2022 GOES-East Nighttime Microphysics RGB transitioning to Day Cloud Phase Distinction RGB, GLM Flash Extent Density.
VIIRS Day Night Band Near Constant Contrast Imagery from the night of the 17th provides a visible like image of the LES bands, particularly the one that had matured over Lake Erie by this time and the early organization of the Lake Ontario band (Fig 3). The NCC imagery provides unique texture details of the cloud tops not available in IR imagery, helping the user to diagnose convective elements. With a lack of light at this point of the moon cycle, city lights begin to dominate. Being creative with color tables, one can try to isolate the city lights from the clouds, such as what is done in the figure. The city lights are in color, while the clouds are in grayscale.
Figure 3: 0658 UTC (top) and 0747 UTC (bottom) 18 Nov 2022 VIIRS DNB Near Constant Contrast product.
In addition to the large scale support previously mentioned, the smaller scale setup favored the development of Lake Effect Snow bands during the night of the 17th into the 18th. DCPD RGB imagery with sfc, 850 mb, and 700 mb RAP winds, and 700 mb temperatures, show a favorable environment for LES had set up over Lake Erie by the morning of the 18th (Fig 4). The lower level winds showed little variation with height, and the 700 mb temperatures(~-20C) were considerable cooler than the Lake (~10C). Connecting the proper NWP variables with satellite observations can be a useful strategy for maintaining situational awareness, and connecting what the models are showing to what is being observed.
Fig 4: 18 Nov 2022 GOES-East DCPD RGB imagery with RAP model analysis sfc, 850 mb, 700 mb winds, 700 mb temperature. See animation for key.
Leveraging the RGB + Derived Product Readout Menus in AWIPS (satellite > Local Menu Items), one can get even more information out of the satellite imagery. The DCPD RGB procedure, when sampled, reveals cloud top information such as cloud top height and phase (Fig 4a). Sampling the suspected LES band near Buffalo in this example confirms glaciated cloud tops, plus cloud top heights over 10,000 ft.
Figure 4a: 1646 UTC 18 Nov 2022 GOES-East DCPD RGB image with Derived Product Readout information.
A 2-panel animation comparison the DCPD RGB with MRMS Composite Reflectively helps to show the relation between radar echoes and how those features appear in satellite imagery (Fig 5). This can be useful knowledge for when needing to diagnose snow bands below radar beams and in radar poor zones, but also to supplement available radar. Identifying trends such as increased cloud-top glaciation, cooling clouds, and increase in cloud texture could be useful in identifying intensifying snow bands. Re-positioning and re-orientation of show bands, as well as tightening of cloud edges, are also useful pieces of information that may be gleaned from the satellite imagery.
Figure 5: 18 Nov 2022 GOES-East DCPD RGB (top) and MRMS Composite Reflectivity (bottom).
All of the aforementioned features are more easily diagnosed in 1-min imagery, such as the 1-min DCPD RGB imagery shown in Fig 6.
Figure 6: 18 Nov 2022 GOES-East 1-min DCPD RGB.
NUCAPS temperature and moisture profiles provide information about the thermodynamic environment between synoptic balloon launches. NUCAPS profiles sampled from both Lake Erie and Lake Ontario for this event in the middle of the night show decent lapse rates (even morose if you consider the warm lake temperatures and resulting lake-induced instability) and moist profiles (Fig 7).
Figure 7: 0752 UTC over Lake Ontario (top) and 0753 UTC over Lake Erie (bottom) 18 Nov 2022 NUCAPS temperature and moisture profiles. NUCAPS point selection is no right.
Finally, the NESDIS Snowfall Rate product, derived from Microwave data on polar-orbiting satellites, captured intense instantaneous snowfall rates within the column, including rates of at least 0.14 in/hr liquid equivalent (Fig 8)!
Figure 8: 1443 UTC, 1538 UTC, 1827 UTC 18 Nov 2022 NESDIS Snowfall Rate product.
Lake effect snow continued off of the two lakes, off and on, through Sunday evening, finally clearing by Monday morning. The animation in Fig 9, similar to in Fig 2, shows the evolution of the bands from Thursday night through Monday morning.
A broad upper low positioned over the western US resulted in critical fire weather and winter weather conditions across High Plains on 09 Nov 2022. By the start of the day, an east-west oriented quasi-stationary cold front was draped across northern Colorado. IR satellite imagery captured the exact position of the cold front during the day, filling temporal and spatial gaps between surface obs.
Figure 1: 09 Nov 2022 GOES-East 5-min CWIR Imagery, sfc obs. Darker shades of gray = warmer brightness temperatures.
In order to save screen space, and monitor the position of the cold front in addition to clouds and aerosols, one can create a semi-transparent IR overlay on Geocolor imagery. In this case, south of the front, gusty southwesterly winds brought temperatures into the 60s – 70s, critical fire weather conditions, along with blowing dust across southern Colorado. North of the front, temperatures stayed in the 30s-40s during the day, with weaker easterly winds and low stratus well northeast of the boundary.
Figure 2: 09 Nov 2022 GOES-East 5-min Geocolor + semi-trans CWIR Imagery, sfc obs. Darker area relates to warmer brightness temperatures.
NWS/BOU leveraged GOES Imagery in a similar manner: “There are a lot of interesting features on satellite today. The low clouds and fog have stuck around in the far northeast corner keeping temperatures near freezing in Sterling and Julesburg. Meanwhile, a sharp boundary has been draped across the east central plains which is obvious on the shortwave IR from the southeast Denver metro to central Washington County. There are critical to extreme fire weather conditions to the south of this boundary with no fire weather concerns to the north of this boundary. Limon is currently 74 degrees with a relative humidity of 5 percent with gusts up to 53 mph. The Red Flag Warning is verifying there and the warning will be kept in its current state until it expires at 7pm. Latest visible satellite data also shows a couple areas of blowing dust mainly in eastern Arapahoe County and nearby areas.”
Focusing further south, we observe considerable blowing dust in GOES-East Geocolor Imagery out of the San Luis Valley. From NWS/PUB: “As the pressure gradient tightens overhead, southerly and southwesterly winds in excess of 50kt have been reported across the forecast area, and a thick area of blowing dust can be seen across the northern San Luis Valley on enhanced satellite imagery. Alamosa is reporting 2 miles visibility in blowing dust.”
Figure 3: 09 Nov 2022 GOES-East 5-min Geocolor Imagery.
Can the Day Cloud Type RGB (1.37, 0.64, 1.61) be a useful RGB when monitoring for convective initiation? This blog post will attempt to share some observations regarding the Day Cloud Type RGB when monitoring for convective development, particularly in the transition from the developing to mature stages of the convective lifecycle.
Firstly, let’s talk about the components that make up the Day Cloud Type RGB. In your best Michael Buffer voice: “In the red corner, we have the 1.37 µm channel; the green corner, we have the 0.64 µm channel; and in the blue corner is the 1.61 µm channel.” Let’s focus on the 1.37 µm channel, and what it is actually measuring. Onboard NOAA’s GOES-R series of geostationary satellites, the 1.37 µm channel corresponds to Band 4 within the Advanced Baseline Imager (ABI), with a given name of the ‘Cirrus Band.’ Per the CIMSS Day Cloud Type RGB quick guide, compared to the Day Cloud Phase Distinction RGB, “The use of the Band 4 ‘Cirrus Channel’ at 1.38 µm (rather than the 10.3 µm Clean Window Channel) allows for better detection and discrimination of thin and thick cirrus clouds.” Additionally, one of the RGB’s composers, Andrew Heidinger of NESDIS (other composer: Jochen Kerkmann from EUMETSAT), provided insights on the original intent of the RGB to aid in viewing thin cirrus, and can be viewed here.
However (spoiler alert), the ‘Cirrus Band’ can observe clouds besides cirrus. Is your mind blown? Don’t worry, mine was, too, after discovering this fact.
Rather, reflectance back to ABI within this channel is dependent upon water vapor content, and to an extent how it is vertically arranged within the atmosphere. The 1.37 µm channel sits within an atmospheric water vapor absorption region of the electromagnetic spectrum. This means given a sufficient amount of water vapor, incoming solar radiation in the 1.37 µm wavelength travelling into the atmosphere will get absorbed by the atmosphere rather than reflected back to space (and satellite). Since most of the atmosphere’s water vapor is within the lowest portions of the troposphere, there is still opportunity for clouds/aerosols above the absorbing amount of water vapor to reflect solar radiation back toward the satellite. Typically these reflective clouds are high enough in altitude to fall within the “cirrus” category, and hence the given “Cirrus Band” title. But should the atmosphere *not* hold a sufficient amount of water vapor to absorb radiation, reflectance from non-cirrus clouds, as well as other meteorological and non-meteorological features, can be detected.
The other two components of the Day Cloud Type RGB, the 0.64 µm channel and the 1.61 µm, are great for monitoring and tracking cloud motion and texture as well as distinguishing cloud particle phase at the top of clouds.
So how does all of this pertain to monitoring convection? Imagine a very unstable environment ripe for deep convection. It may be comprised of rich moisture within the boundary layer underneath very dry air aloft. So while any clouds within this lowest level moisture may not be reflective in the 1.37 µm channel, once clouds vertically grow into the dry layer aloft, it becomes reflective. In this example, the satellite sensor would only see reflectance from clouds growing above the moist boundary layer, such as the case when Cu “breaks the cap” into a relatively dry portion of the atmosphere, such as an elevated mixed layer.
Use the conceptual model with the image slider above. Each letter corresponds to the conceptual understanding of how much insolation is being absorbed and reflected based on the associated cloud top height. Image slider: left / right
With its inclusion of the 1.37 µm channel, the Day Cloud Type RGB can be an excellent tool to monitor when clouds grow vertically high enough, dependent upon moisture content and its orientation within the troposphere. Couple this with the power of cloud top phase information from the 1.61 µm and general cloud motion and texture information from the 0.64 µm, and you have a tool for monitoring the convective lifecycle.
“But wait, why would I use this RGB when I have the Day Cloud Phase Distinction RGB which that has a proven track record of utility in monitoring the convective lifecycle?” one might ask themselves.
It is true, the Day Cloud Phase Distinction RGB does have this utility, especially when monitoring for convective initiation, or during the transition from the developing to mature stage of the convective lifecycle. However, the Day Cloud Type RGB is starting to show some striking contrast right at the moment of this transition, at least in some cases. Let’s look at an example to compare the two RGBs at time of convective initiation.
In this example near the Tennessee and Kentucky border on September 25, 2022, a cold front forced sustained convection to initiate (here is the associated SPC Severe Weather Event Review from this day). The Day Cloud Phase Distinction RGB is on the top, and Day Cloud Type RGB on the bottom. Both RGBs show convective initiation, but the Day Cloud Type RGB really highlights these vertically growing clouds with its red/orange color starkly contrasted to neighboring shallower cyan clouds. This contrast makes it easy to see to moment of convective initiation. And in this case, the Day Cloud Type RGB can more clearly denote this transition compared to the Day Cloud Phase Distinction RGB where its contrast may be harder to distinguish the vertically growing and glaciating clouds turning green compared to a blue background and cyan clouds.
Image slider showing the Day Cloud Phase Distinction RGB and Day Cloud Type RGB, and their individual compositional contributions in black to white (white representing a higher contribution; black, lower contribution). Clockwise from the upper left: RGB, Red component, Blue component, and Green component. Image slider: left / right
The enhanced contrast within Day Cloud Type RGB compared to the Day Cloud Phase Distinction RGB in this instance is simply due the relatively shallow nature of convection within this environment not acquiring cold enough cloud top temps for the 10.3 µm channel to add a significant contribution within the red component of the Day Cloud Phase Distinction RGB. Conversely, the moisture profile of this environment allowed for a very large contribution of the Day Cloud Type RGB’s red component, the 1.37 µm channel. The contributions of each RGB component is represented in the image above, with each quadrant representing the RGB itself and the channels that make up its components in black and white, with white representing higher contribution. While the blue and green components are nearly identical between these two RGBs, the red component in the upper right quadrant of the image above differs substantially. The 10.3 µm channel in this configuration of the Day Cloud Phase Distinction RGB is not very bright, representing a lack of contribution to the overall RGB (note that one could adjust the range to make it stand out more, however). Again, the converse is illustrated in the 1.37 µm of the Day Cloud Type RGB.
It is important to emphasize the dependency of moisture content and orientation within the troposphere when considering the Day Cloud Type RGB for monitoring the lifecycle of deep convection; and in this case, it just happened to work out nicely. If you look at Nashville’s 18 UTC RAOB within the warm sector just ahead of this convective line (image above), you can see the vast majority of moisture resides below ~600 mb, with a very dry layer aloft. If we associate this ~600 mb layer as the demarcation of sufficiently absorbing water vapor content in the 1.37 µm, the observed environmental temperature at this layer infers cloud tops would only begin to reach the point of glaciation around -10 C. This makes sense as to why we didn’t see a transition of green to red in the Day Cloud Type RGB as the 1.37 µm reflectance likely increased substantially just as cloud top phase was changing, aka glaciating.
Having an RGB to easily diagnose the point of convective initiation can be a powerful tool within a mesoanalyst’s toolbelt. As a mesoanalyst, knowing this point of initiation can help provide confidence when telling the warning met/team, “These clouds are likely initiating, and these are the likely hazards given the near storm environment (blurb on hazards). I suggest you begin radar and storm-scale interrogation in this area.” Additionally as a mesoanalyst, knowing this point can help those involved in messaging, graphic generation, as well as IDSS to locate near-term hazards both in time and space.
So let’s briefly mention some potential strengths and weakness of using the Day Cloud Type RGB for monitoring the transition of developing to mature stages of the convective lifecycle:
Strengths:
Can easily illustrate convective initiation where moisture content is mostly confined within the boundary layer beneath very dry layer aloft.
Won’t succumb to seasonal thermal differences (like the 10.3 µm channel), outside of seasonal and latitudinal differences in moisture content and solar reflectance.
Weaknesses:
Environments composed of deep moisture throughout the troposphere (like the tropics), may see a delayed signal (the appearance of orange/red coloring) beyond ‘true’ convective initiation.
Environments composed of very dry air throughout the troposphere (typical in arid, very cold, and/or high elevation climates), early-lifecycle clouds will likely have strong orange/red coloring, even before ‘true’ point of initiation. This may mislead a forecaster to incorrectly signal a transition from infant to maturing clouds.
Due to dependence of solar radiation, can only be used in daytime.
Limited use when higher level clouds/aerosols are upstream or above area of interest.
It cannot be stressed enough that having knowledge of the moisture profile in the area of interest is crucial to applying this technique. Knowing how moisture is parsed throughout the troposphere will aid your understanding of when the 1.37 µm component will begin contributing to the RGB.
It is also worth mentioning the Day Cloud Type RGB’s lack of cloud top temperature information. This information can give convective lifecycle details including rate of growth of convection. If such information is desired, it is recommended to utilize the Day Cloud Phase Distinction RGB, or utilize the single band 10.3 µm.
This type of application of the Day Cloud Type RGB is still in its infancy, with much of the content within this blog post still under review by peers within the satellite applications community. Therefore, proceed with caution. That being said, if you happen to notice any cases where this application either excels or fails, please reach out to the author and/or comment within this post.
Many thanks to Patrick Ayd (NWS DLH) for sparking interest in utilizing the 1.37 µm for convective monitoring, Andrew Heidinger (NESDIS) and Jochen Kerkmann from (EUMETSAT) for composing the Day Cloud Type RGB, and the TOWR-S team for including the Day Cloud Type RGB within AWIPS. Additionally, many thanks to all who have contributed to conversations regarding this potential application.
Satellite Liaison Blog 
