Operational AVHRR based Monitoring of Central American Volcanoes

1
Operational AVHRR based Monitoring of Central
American Volcanoes

• Peter Webley KCL

M. J. Wooster (KCL), J.A. Saballos, W. Strauch,
K. Dill (INETER) P. and J. Stephenson (BURS)
2
Operational Remote Sensing system

• Operational test of previously derived technique
• Design a AVHRR system to monitor volcanoes in
  real time
• Examine both the usefulness of the system for
  volcano monitoring and the dissemination of such
  data into the local communities for volcanic
  hazard and early warning
• Funding for project
• DfID grant - 2½ year project

3
AVHRR and Thermal Monitoring?

• AVHRR Advanced Very High Resolution Radiometer
• NOAA operates the POES series, upon which the
  AVHRR sensor is mounted.
• Orbit at 850 km.
• Scans through ? 55 either side of the orbital
  track
• 5 spectral channels .
• For thermal monitoring, Channels 3 and 4 are most
  useful.
• As surface temperatures rise, the thermal
  emmittance increases much more rapidly in the MIR
  than in the TIR
• Important as at the resolution of the AVHRR, the
  areas of the volcano at highly elevated
  temperatures are likely to be highly sub-pixel
• Possible for to detect these increases over and
  above the ambient background emission, even
  though the hot area may comprise less than 0.1
  (i.e. 1000th) of the pixel area.

4
AVHRR Real time Volcano Monitoring Project

• Objectives
• To investigate the potential improvements to
  local volcano monitoring and hazard mitigation
  operations that can be gained via use of locally
  captured satellite remote sensing data in
  developing countries.
• The project includes both physical and social
  science components in order to make this
  assessment, and the main data used is from AVHRR,
  captured locally in Nicaragua.
• Ideal Accomplishments
• Installation of AVHRR receiving station in
  Nicaragua (INETER).
• Design and operation of fully automated data
  capture and pre-processing system.
• Extensive case studies into the application of
  AVHRR to monitor and detect thermal volcanic
  activity.
• Design of automated analysis system to monitor
  volcanoes in Nicaragua, Guatemala, El Salvador
  and Costa Rica.
• Development of web based interface for data
  download and thermal monitoring by outside
  geo-science organisations.
• Investigation of auto email alerts based on AVHRR
  signals
• First test of an operational system in developing
  country
• Involves local volcanologists and Increased
  application of data and results.
• Other systems in US and Japan

5
AVHRR Receiving Station and Data

• Image area 2600 km 2600km
• 6 - 8 satellite passes per 24 hours (day and
  night imagery)
• Much more than the single daytime image online at
  the SAA, operated by NOAA.
• Data
• 5 Spectral Channels
• Measuring Reflectance and Temperature
• At 1.1 km resolution at nadir
• Analysis
• Use T3 T4 to quantify at volcanic activity
• T3 increases rapidly in presence of sub pixel
  hotspot
• Determine radiance anomalies for the hotspot
  pixels
• Use T4 T5 for ash cloud test
• Provide time series of important data and
  automated alerts

6
Computing/Data Storage

• Computing
• Data Capturing
• Data Analysis
• Software
• BURS
• ENVI/IDL
• Data files
• RAW 10 40 MB
• ENVI BIL 90 180 MB
• Data Storage
• Per day 400 500 MB
• CD backup
• Copied each week
• Storage within INETER
• Development to DVD storage
• Analysis outputs
• Stored on PC
• Displayed on website

7
Number of AVHRR passes?

• 16th January 2005
• 9 passes successfully captured
• 17th January 2005
• 7 passes successfully captured

8
Volcanoes analysed by the AVHRR monitoring system

• Nicaragua
• Cerro Negro, Concepcion, Cosiguina, Masaya,
  Mombacho, Momotombo, San Cristobal, Telica
• Guatemala
• Acatenango, Atitlan, Cerro Quemado, Fuego,
  Pacaya, Santa Maria/Santiaguito, Tacana,
  Tecuamburro
• El Salvador
• Izalco, San Miguel, San Salvador, Santa Ana
• Costa Rica
• Arenal, Irazu, Poas, Rincon

9
Planning orbits
Downloads orbit element data daily from NOAA and
uses this to accurately predict the next
satellite overpass time and position
10
AVHRR Capture System

• Data Capture
• Data calibration
• Data converted to universal format
• Pre-processing fully automated

• NOAA Scheduler
• Carries out orbit planning, data capture and data
  calibration
• Conversion to universal format automated

11
Analysis System Summary

• Uses IDL/ENVI
• Automatically compiles and runs code if satellite
  pass within past 45 minutes
• If no pass, then will close and re-load in 45
  minutes
• Loads AVHRR scene into ENVI and extracts the
  following data
• Channels 1 to 5, Latitude and Longitude
• Satellite Azimuth, Satellite Elevation, Sun
  Azimuth and Sun Elevation
• Image covers 8 volcanoes in Nicaragua, 8 in
  Guatemala, 4 in Costa Rica and 4 in El Salvador
  (not necessarily all at once)
• Uses image geocoding to find the pixel
  corresponding to the volcano summit
• Possible geo-location uncertainty of a few
  pixels.
• For each volcano, program carries out the
  following analyses
• Find the Max T3 T4 close to the identified
  summit pixel
• Determines thresholds to detect hot/anomalous
  pixels on volcano
• Determines radiance anomalies at these anomalous
  pixels
• Carries out cloud analysis of region surrounding
  the volcano

12
Volcano Activity Identification Stage 1

• Finds the Maximum T3 T4 value in a 7 by 7 grid
  around the summit pixel. This is the updated
  volcano summit location.
• Determines this to be the location of the volcano
  and then selects a 7 by 7 grid around this.
• Determines the mean and standard deviation of
  this new array
• Analyses the grid to determine which pixels are
  anomalous. If none, then no further radiance
  calculations
• If some then uses these anomalous pixels in
  radiance calculations

13
Radiance Calculations Stage 2 (a)

• Calculates the Radiance for Channels 3, 4 and 5
  from Planck equation
• ? wavelength (m) T is temperature (K)
• L is Spectral Radiance (W/m2/sr/µm) h is
  Plancks Constant (6.610-34 Js)
• k is Boltzmanns constant (1.3810-23 J/K) c
  is the speed of light (3108 m/s)
• Background Radiance Anomaly
• Difference between AVHRR T3 and the surrounding
  non-thermally anomalous background pixels in the
  same spectral channel, T3.
• The advantage
• Effect of the atmosphere on the radiance anomaly
  measure should be minimised
• The disadvantage
• Value of the retrieved thermal anomaly effected
  if hotspot and background pixels at different
  elevations

14
Radiance Calculations Stage 2 (b)

• Equivalent Radiance Anomaly
• Difference between the AVHRR T3 signal and the
  simulated T3 signal modelled using the T4
  observations at those same pixels using the
  Planck function.
• The advantage
• Effect of differing elevations on the radiance
  anomaly is largely avoided
• The disadvantage
• Differences in the atmospheric effect between T3
  and T4 can affect the radiance anomaly
  calculations,
• May also be underestimated if the volcanic
  hotspot significantly effects the T4
  observations.
• Simulated Radiance Anomaly
• Calculate the empirical linear best-fit
  regression relationship between the AVHRR T3 and
  T4 data around the volcanic hotspot
• This relationship is then used with the AVHRR T4
  data of the volcanic hotspot pixels to simulate
  the AVHRR T3 signal at these locations would be
  in the absence of the volcanic hotspot.
• These simulated T3 observations are then
  subtracted from the actual MIR observations to
  calculate the observed volcanic thermal anomaly.
• Advantage
• Allows for a difference in atmospheric effect
  between T3 and T4.
• Disadvantage
• Only works if a strong regression relationship is
  found between these two channels in the
  background data selected.

15
Outputs from Analysis (1)

• Text file based outputs for each volcano
• Channels 1 to 5 at summit and hotspot
• Channels 3 4 and 4 5 at summit and hotspot
• No of Ash and saturated pixels
• Three radiance anomalies
• Distance between summit and hotspot
• CI and cloud value for summit pixel

Cloud index Summit value
Radiance anomalies
Date and time of image
Channels 3, 4 and 5 at hotspot
16
Outputs from Analysis (2)

• Fully georeferenced images for each country and
  volcano
• Processed data in real-time for own personal
  analysis
• Gridded text files of AVHRR data for each volcano
• Time series plots of text data.
• Data sent automatically by FTP to web server
• ARCVIEW data for each country and each volcano
• ENVI image data for each country and each volcano
• ENVI GIS shape files for each country and each
  volcano
• Georeferenced imagery for each country and each
  volcano
• Time series text files for each country and each
  volcano
• Time series figures for each country and each
  volcano

17
Georeferenced Imagery for Nicaragua
Cerro Negro
18
Georeferenced Imagery for Guatemala
Fuego
19
Georeferenced Imagery for El Salvador
Izalco
20
Georeferenced Imagery for Costa Rica
Poas
21
Threshold Analysis

• Uses predefined thresholds to determine volcano
  alert level
• Four Levels. Maximum 3, Minimum 0
• Threshold levels for Equivalent Radiance anomaly
• Level 0 1 1.6 W/m2/str/µm
• Level 1 2 3.2 W/m2/str/µm
• Level 2 3 6.4 W/m2/str/µm
• Reads in the AVHRR Channel and Radiance anomaly
  time series for the volcano
• Determines if the Cloud Index gt 0.6 or if it is a
  daytime pass.
• Determines if the Radiance anomaly has passed
  over the predefined threshold for the volcanos
  alert level
• Counts the total number of alerts in the past 15
  images
• If 2 or more than volcano alert level rises
• If less than 2 than volcano alert level stays the
  same
• If 3 or more times below threshold then alert
  level drops
• If Radiance anomaly has passed over threshold
  data is written to text file for attachment to
  e-mail.
• Equivalent Radiance anomaly

22
E-mail alert system

• Data written as text to a .txt file for
  attachment to an e-mail
• Uses a e-mailing software and a batch file to
  auto email to a mailing list for the country,
  based on the volcano.
• Mailing lists for each country.
• avhrr_gua_at_gf.ineter.gob.ni
• avhrr_nic_at_gf.ineter.gob.ni
• avhrr_sal_at_gf.ineter.gob.ni
• avhrr_cor_at_gf.ineter.gob.ni
• e.g. Nicaragua, for each of the eight volcanoes
  monitored an e-mail is sent to the
  avhrr_nic_at_gf.ineter.gob.ni list if the volcano
  has passed the defined thresholds.

• 3 alert level changes
• 6 e-mails sent

23
Example of E-mail Alert
24
Real-time data accessible from website
http//sat-server.ineter.gob.ni/

• Fully georeferenced imagery ENVI Image files
• Time series of Radiance anomalies E-mail alert
  system
• Gridded data around each volcano

25
Summary

• AVHRR system analyses for 24 volcanoes across
  Central America
• AVHRR can detect and monitor the thermal
  signature of volcanic activity in real time
• System analyses for thermal activity of the
  volcanoes
• System is fully automated
• Capture
• Analysis
• Automated E-mail alert for all volcanoes
• Website interface interface to thermal analysis
  for 24 volcanoes, 24 hours per day
• Automatically updated

Developments

• Improvement and automation of ash cloud detection
• Reduction in time to send data to FTP server
• Most recent 3 5 images displayed on website