Quantitative Volcanic Ash Remote Sensing at NOAANESDISSTAR and CIMSS - PowerPoint PPT Presentation

Title: Quantitative Volcanic Ash Remote Sensing at NOAANESDISSTAR and CIMSS


1
Quantitative Volcanic Ash Remote Sensing at
NOAA/NESDIS/STAR and CIMSS

• Mike Pavolonis (NOAA/NESDIS/STAR)
• and
• Justin Sieglaff (CIMSS)

2
Outline

• Addressing operational challenges
• AVHRR product development
• GOES product development
• GOES-R and NPOESS product development

3
Some Addressable Operational Challenges

• Heavy reliance on time consuming manual analysis
  of satellite imagery
• Volcanic clouds sometimes closely resemble
  meteorological clouds in conventional imagery
• Quantitative information on ash cloud height,
  concentration, and microphysics is not available

4
Some Addressable Operational Challenges

• Heavy reliance on time consuming manual analysis
  of satellite imagery - automated scanning of
  satellite imagery for volcanic ash clouds can now
  be performed reliably, allowing for automated
  warnings to be passed onto analysts/forecasters,
  as manual analysis of every satellite image in
  real-time is often not practical (Dynamic
  Tasking and Adaptive Sensing)
• Volcanic clouds sometimes closely resemble
  meteorological clouds in conventional imagery -
  cloud microphysical retrievals can be used to
  identify clouds which are easily mistaken for
  meteorological clouds
• Quantitative information on ash cloud height,
  concentration, and microphysics is not available
  - these retrievals can be performed and are
  important for forecasting the dispersion of ash
  clouds and ash fallout

5
Some Addressable Operational Challenges

• Heavy reliance on time consuming manual analysis
  of satellite imagery - automated scanning of
  satellite imagery for volcanic ash clouds can now
  be performed reliably, allowing for automated
  warnings to be passed onto analysts/forecasters
• Volcanic clouds sometimes closely resemble
  meteorological clouds in conventional imagery -
  cloud microphysical retrievals can be used to
  identify clouds which are easily mistaken for
  meteorological clouds
• Quantitative information on ash cloud height,
  concentration, and microphysics is not available
  - these retrievals can be performed and are
  important for forecasting the dispersion of ash
  clouds and ash fallout

NOAA/NESDIS/STAR and UW-CIMSS are developing
products for current and future GOES and POES
that address these operational challenges.
6
AVHRR Project Description

• Using advanced spectral and spatial techniques, a
  fully automated ash detection algorithm along
  with an ash height, particle size, and
  concentration retrieval were developed for the
  AVHRR.
• NOAA has funded us to transition these products
  from research to NOAA/NESDIS operations.
• The volcanic ash products are scheduled to be
  produced operationally in the Spring of 2010, but
  they will be available prior to then for
  evaluation.

7
Example Ash Detection Results
Kasatochi (2008)
Redoubt (2009)
8
Example Ash Detection Results
Kasatochi (2008)
Redoubt (2009)
Very thin ash cloud is detected without false
alarms
Ice topped portion of volcanic cloud is detected
using a different methodology
9
Volcanic Eruption or Thunderstorm?
GOES-12 IR Window
GOES-12 Ash/Ice
From the 1544 UTC VAAC message REMARKS WE HAVE
RECEIVED A REPORT FROM NAVY AVIATION IN THE AREA
INDICATING AN ERUPTION OF SANTA ANA TO
FL460. From the 1601 UTC VAAC message REMARKS
THE ERUPTION IS NOW BELIEVED TO HAVE STARTED AT
ABOUT 1400Z AND AN AREA PREVIOUSLY THOUGHT TO BE
THUNDERSTORMS IS NOW IDENTIFIED AS THICK ASH. A
FORECAST WILL BE ISSUED AS SOON AS POSSIBLE
10
Ice Topped Volcanic Clouds
Mount Spurr (August 1992)
Ice topped ash clouds, which are common in the
earliest phase of ash cloud evolution, can only
be reliably identified using advanced cloud
microphysical retrieval techniques.
Pavolonis et al., 2006
11
Redoubt 03-23-09 - 1146 UTC
Most BTDs were gt 0K due to larger reff and
multilayered clouds
12
Redoubt 03-23-09 - 1430 UTC
Thin layer of ash is captured well.
13
Spurr 09-17-92 - 1245 UTC
A large area of the cloud is dominated by smaller
particles.
14

• Particle size likely plays a large role in
  determining ash residence time.
• The September 1992 Mount Spurr cloud remained
  visible in satellite imagery for 80 hours as it
  impacted the lower 48, while the 2009 Redoubt ash
  clouds could not be detected with conventional
  satellite imagery lt 12 - 18 hours after each
  eruption.

15
Kasatochi - 08 - 08 - 2008 - Loop
16
Sarychev - 06 - 14 - 2009 - 0915 UTC
17
AWIPS Visualization
Ash Mass Loading
Ash Probability
Ash Effective Radius
Ash Cloud Height
18
IC4D Visualization
Ash Mass Loading
19
Algorithm Development for the Current GOES Imager

• Project Goal Apply our volcanic cloud algorithms
  to the current series of GOES imagers. Once
  algorithms are deemed reliable, pursue a
  transition to operations.
• Products produced ash probability, ash height,
  ash effective radius, and ash concentration.

20
Redoubt 03/26/09 - 2030 UTC

• We are pursuing similar algorithms with GOES, but
  we are also exploring ways to utilize the
  temporal information.
• We hope to add these algorithms to GOES
  operations within the next few years.

21
Future GOES and POES

• GOES-R ash requirements ash detection and height
  (includes ash mass loading and particle size)
• NPOESS ash requirements There are no specific
  ash requirements for NPOESS, but a slightly
  modified form of the GOES-R algorithm can be
  applies to the VIIRS.

More accurate and frequent ash products will be
generated in the GOES-R era (with good coverage
of southern Alaska). NPOESS (especially VIIRS
and CrIS combined) will allow for better ash
remote sensing compared to POES.
22
Ash Detection
The ash detection results are expressed as a
confidence value.
23
Ash Detection
Eruption of Karthala (November 11/25/2005)
24
Ash Detection
Full disk results indicate that while the
probability of detection is high for most ash
clouds, while the probability of false alarm is
low.
Ash Detection
Ash cloud
RGB
25
Ash Retrieval
RGB Image
Ash Height
Ash Loading
Total Mass 117 ktons
The height and mass loading products are free of
visual artifacts and have reasonable spatial
patterns for this moderate sized eruption.
26
Ash Retrieval
RGB Image
Ash Height
Ash Loading
Ash Cloud
Total Mass 8.8 ktons
The height and mass loading products are free of
visual artifacts and have reasonable spatial
patterns for this light sized eruption.
27
Validation
The retrieved height agrees well with the CALIOP
cloud boundaries.
SEVIRI RGB
Dust Cloud
CALIOP Backscatter
28
Summary

• Reliable automated quantitative ash detection is
  now possible.
• Retrievals of ash height, effective particle
  radius, and concentration provide important
  additional information that can potentially be
  used in a prognostic manner.
• The ABI on GOES-R will have an unprecedented
  combination of spatial, temporal, and spectral
  resolution. In order to take full advantage of
  these new capabilities, quantitative volcanic
  cloud products are needed to help alert
  forecasters to the presence of volcanic clouds,
  as manual analysis over all volcanoes at each
  image time (in real-time) is not feasible.
• Similar approaches should be taken with other
  aviation hazards such as fog, turbulence, and
  deep convection.
• Close collaboration with the user community
  through the PG is vital for improving operational
  satellite capabilities.

29
Backup Slides
30
Why Automated Ash Detection?

• Manual analysis of every satellite scene may not
  always be possible, especially with the next
  generation of GEO satellites. Un-instrumented
  volcanoes especially need to be monitored
  carefully by satellite.
• Ash locations are needed to constrain ash
  property retrievals
• Automated detection will aid data mining and
  climate applications

31
Improving Upon Established Ash Detection
Techniques
The 11 - 12 ?m split-window brightness
temperature difference has traditionally be used
to detect ash.
The split-window technique is not suitable for
automated ash detection, though, because it is
hampered by numerous false alarms (right) and
missed detection due to water vapor absorption
(top).
From Pavolonis et al. (2006)
32
Improving Upon Established Ash Detection
Techniques
By adding additional spectral information, we
were able to increase the probability of
detection, especially when the split-window BTD gt
0 K.
From Pavolonis et al. (2006)
33
Improving Upon Established Ash Detection
Techniques
The use of additional spectral information also
greatly improved the global false alarm rate.
From Pavolonis et al. (2006)
34
Improving Upon Established Ash Detection
Techniques

• Despite these improvements, the skill in
  differentiating volcanic ash from other
  atmospheric and surface features is still not
  good enough for an operational setting.
• Thus, additional spectral and spatial based
  algorithm improvements were needed.

From Pavolonis et al. (2006)
35
Improving Upon Established Ash Detection
Techniques

• Improve the spectral sensitivity of split-window
  measurements to volcanic ash by explicitly
  accounting for background conditions (e.g.
  surface temperature, surface emissivity, water
  vapor, etc) on a pixel by pixel basis.
• Make advanced use of spatial information.

36
Ice Topped Volcanic Clouds
11 - 12 ?m BTD
3.75 ?m Reflectance
?obs(12, 11 ?m)
37
Ice Topped Volcanic Clouds
11 - 12 ?m BTD
Large BTDs, ?s, and NIR reflectance values are
indicative of very small particles,which are
fairly rare for optically thick ice clouds.
3.75 ?m Reflectance
?obs(12, 11 ?m)
38
Ash Retrieval

• Retrievals of ash concentration (optical depth
  and particle size) have been limited to case
  studies. Automated real-time capable retrieval
  algorithms are lacking in operational settings.
• An optimal estimation procedure (Heidinger and
  Pavolonis, 2009) is used to retrieve the ash
  cloud top temperature, emissivity, and
  microphysical parameter for pixels determined to
  contain ash by the detection algorithm.
• The results are used to compute an ash
  concentration.
• Only infrared channels are used, so the results
  and day/night independent and the procedure is
  fully automated.
• It is hoped that these retrievals can be used to
  improve dispersion models.

39
Ash Retrievals

• Cloud temperature is a free parameter
• Atmospheric water vapor, surface temperature, and
  surface emissivity are accounted for in the
  retrieval
• A multilayered correction can be applied
• The 1D VAR retrieval produces error estimates for
  each of the retrieved parameters

40
Ash Detection
Ash detection under difficult multilayered
conditions is improved when the low cloud layer
is approximately accounted for.
Ash detection with multilayered correction
41
Summary

• Automated quantitative ash detection requires an
  approach that can isolate the cloud microphysical
  signal from the background signal and make
  maximum use of spatial information in order to be
  of operational quality (high detection, low false
  alarm rate).
• Retrievals of ash height, effective particle
  radius, and concentration provide important
  additional information that can potentially be
  used in a prognostic manner.
• We are applying similar detection and retrieval
  approaches to the GOES imager and we are
  developing more advanced algorithms for GOES-R.
• Our goal is an automated combined LEO/GEO global
  volcanic ash monitoring system that will be a
  reliable tool for volcanic ash forecasters.
• We welcome collaborations on all of these issues.