Operational forest fire monitoring in brazil
Download
1 / 47

Operational Forest Fire Monitoring in Brazil - PowerPoint PPT Presentation


  • 269 Views
  • Uploaded on

Operational Forest Fire Monitoring in Brazil Wilfrid Schroeder, M.Sc. PROARCO - Fire Monitoring System Brazilian Institute for the Environment and Natural Renewable Resources – IBAMA [email protected] About us... IBAMA is the Major Environmental Agency in Brazil

loader
I am the owner, or an agent authorized to act on behalf of the owner, of the copyrighted work described.
capcha
Download Presentation

PowerPoint Slideshow about 'Operational Forest Fire Monitoring in Brazil' - paul2


An Image/Link below is provided (as is) to download presentation

Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author.While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server.


- - - - - - - - - - - - - - - - - - - - - - - - - - E N D - - - - - - - - - - - - - - - - - - - - - - - - - -
Presentation Transcript
Operational forest fire monitoring in brazil l.jpg

Operational Forest Fire Monitoring in Brazil

Wilfrid Schroeder, M.Sc.

PROARCO - Fire Monitoring System

Brazilian Institute for the Environment and Natural Renewable Resources – IBAMA

[email protected]


About us l.jpg
About us...

  • IBAMA is the Major Environmental Agency in Brazil

  • Responsible for Forestry, Animal life, fishery, etc..

  • One of the Primary Goal is to Manage and Protect the Brazilian Legal Amazon


The challenge l.jpg
The Challenge:

  • Large land area requiring wide scale monitoring system

  • Little or no access from surface: observations need to be made from above

  • Illegal logging activities going on over remote areas

  • New land areas being created using fire as a tool for clearing fields

  • Large number of vegetation fires


Slide4 l.jpg

The Brazilian Amazon

Total Area: 5.2 Million km2

Number of States Covered: 9


Slide5 l.jpg

The Arc of Deforestation

Total Area: 1.6 Million km2

Number of States Covered: 7


The beginning l.jpg
The Beginning

  • Late 80’s – Start using AVHRR’s afternoon pass

    • Technical cooperation with INPE

  • Problems with detection algorithm

    • High number of spurious fires detected

  • Limited field inspection

    • Few satellite hot spot coordinates visited


The need l.jpg
The Need

  • January-March 1998 – The Great Roraima Forest Fire

    • Little operational capacity at that moment prevented early detection and combat

  • Operational Fire Monitoring Facility made necessary

    • Pressure from the international community

  • July 1998 – The PROARCO system was established


The concept l.jpg
The Concept

  • Intensive use of Remote Sensing and GIS technologies for fire monitoring

  • Use of meteorological data for fire risk assessment

  • Quick access to reports/bulletins

    - providing near real time data through internet, fax, and vehicles with satellite communication capability

  • Increase law enforcement activities

  • Have the local communities involved


Remote sensing fires l.jpg
Remote Sensing Fires

  • System based on previous AVHRR use experience

  • Detection algorithm experiencing constant improvement

  • Use of evening overpass (NOAA-12)

    • to avoid saturation from bright surfaces

  • Satellite hot spot data being used for field inspections based on different alert levels (green and yellow)

    • Hot spot location and persistence criteria


Algorithm basics l.jpg
Algorithm Basics

  • Fixed threshold method

    • Adjusting limits through histogram analyses

    • Trial and error

  • AVHRR Channel 3

    • Separating all potential fires through saturation

  • AVHRR Channel 1, 2 and 4:

    • Eliminating bright targets (clouds, water bodies, bare soils,…)


Overall performance l.jpg
Overall Performance

  • Number of spurious fires greatly reduced


Slide13 l.jpg

Overall Performance

  • Poor image navigation are still noticed occasionally


Slide14 l.jpg

Overall Performance

  • Day to day variation as a limiting factor


Slide15 l.jpg

Overall Performance

  • Image Acquisition Problems

Courtesy of INPE


Slide16 l.jpg

Overall Performance

  • Resulting Spurious Fires


Slide17 l.jpg

Overall Performance

  • Similar Effects Affecting the NDVI

Zoomed area showing a large number of false green pixels

CPTEC/INPE


Slide18 l.jpg

Overall Performance

Worth Mentioning – NOAA/AVHRR

  • Detection algorithm performing well

  • Image navigation still requires operator’s hands-on

  • Pixel distortion towards the edge of the image reduces detection capacity and affects hot spot statistics

  • Image acquisition characteristics affect the quality of derived products


Need for improvement l.jpg
Need for improvement

GOES Fire Detection Algorithm

July 2000 – Implementation of CIRA’s RAMSDIS system based on GOES-8 data

Cloud Masking

Tb4 > 2ºC

Day: Tb2 > 17ºC

Potential Fires

Night: Tb2 > 41ºC

Statistics

Sunglint Model

Persistence


Slide20 l.jpg

Need for Improvement

  • Improved monitoring capability (every 30min)

  • Reduced Response Time


Slide21 l.jpg

Need for Improvement

  • Satellite data quality assessment facilitated


Slide22 l.jpg

Need for Improvement

  • Great results from visual image interpretation

Northern Sectors

Southern Sector


Slide23 l.jpg

Need for Improvement

Poor results from automated fire detection algorithm


Slide24 l.jpg

Need for Improvement

  • Persistence Check


Slide25 l.jpg

Need for Improvement

Worth Mentioning – GOES

  • Hot spot location errors are found to be in the 2km range

  • Visual image interpretation has been able to detect 100% of the major fires in National Parks all over Brazil

  • Response time is averaging 2 hours at most cases

  • Coincident meteorological analyses helps planning fire combat management in near real time

  • Constant image acquisition geometry

  • Coarse spatial resolution introduces high number of spurious fires

  • Automated detection is still of limited use


A new era l.jpg
A New Era

  • July 2001 – MODIS hot spot data via ftp access


Slide27 l.jpg

A New Era

  • September 2001 – visual in-flight inspection of MODIS hot spot coordinates showing great results


Slide28 l.jpg

A New Era

  • Rapid Response System images used as a confirmation

Where you see smoke there will be a fire!!

Courtesy of NASA


Slide29 l.jpg

A New Era

  • Day to day variation also observed


Slide30 l.jpg

A New Era

Worth Mentioning – MODIS

  • Hot spot location errors are found to be in the 250m range

  • Coincident high resolution visible images favors fire confirmation during day time

  • Pixel distortion creates similar problems observed with NOAA/AVHRR – what is made worst by non-overlapping images near the equator


Keep on moving l.jpg
Keep on Moving

  • January 2002 – DMSP OLS data made available through NGDC / C. Elvidge et al.


Slide32 l.jpg

Cities

State Boundaries

Keep on Moving

  • Good image navigation


Slide33 l.jpg

Keep on Moving

  • Noise from the South Atlantic Magnetic Anomaly

Multi-angle Imaging SpectroRadiometer (MISR) Instrument aboard NASA's Terra Spacecraft

Courtesy of NASA


Slide34 l.jpg

Keep on Moving

Spurious Fires Detected


Slide35 l.jpg

Keep on Moving

  • Fire detection requiring operator’s hands on

  • Stable lights file outdate as a limiting factor


Slide36 l.jpg

Keep on Moving

Worth Mentioning – DMSP

  • Good correlation with NOAA/AVHRR

  • Sources of contamination limits detection capacity to larger fires (increasing omission error by the use of more restrictive thresholds)

  • Image acquisition time does not match fire peak activity hours

  • Stable lights file must be updated on a regular basis


Slide37 l.jpg

Data QA

  • Helicopters and small aircrafts are used to field inspect the hot spot coordinates, feeding back the monitoring system with valuable information for fine tuning the satellite fire detection algorithms and methods

  • Airborne sensors are used during specific satellite data validation campaigns

  • Satellite data inter-comparison helps identifying commission/omission errors and assessing image navigation problems


Slide38 l.jpg

Data QA

  • Airborne satellite data validation campaigns


Slide39 l.jpg

Data QA

Prescribed Burn at IBGE Reserve

in Brasília – September 2000

Visible Band

Forest Mapper Instrument

IR (8.55 mm)

Fire Mapper Instrument


Slide40 l.jpg

Data QA

Agricultural Burning in a Cerrado Area (savana) in the State of Tocantins - September 2000

Visible Band

Forest Mapper Instrument

IR 8.55mm

Fire Mapper Instrument


Slide41 l.jpg

Data QA

  • IR sensors to be used onboard orbital platforms

  • Airborne System:

  • Pair of Kodak MegaPlus digital cameras (Forest Mapper)

  • IR Sensor (Fire Mapper)

Courtesy of NASA

Infrared Spectral Imaging Radiometer (ISIR) Image over Namib Desert Acquired from Space Shuttle Discovery on 7 August 1997


Slide42 l.jpg

Data QA

SIVAM Aircrafts


Slide43 l.jpg

Data QA

  • GIS system for satellite data ingestion


Slide44 l.jpg

Final Considerations

Latest Improvements

  • Increasing spatial resolution (visible channels) allowed for visual confirmation of fires in the images (smoke plumes)

  • Improved navigation parameters reducing processing time (no GCP collection needed) and making field inspection easier

  • Increasing spectral resolution / mid-IR channel saturation facilitating fire/non fire discrimination


Slide45 l.jpg

Final Considerations

Remaining Points

  • Varying pixel size through image cross section imposes some significant limitations to hot spot data applicability (specially with polar orbiting spacecrafts)

  • Full global cover every 12 hours is imperative. Tropical areas are affected by little image overlapping between consecutive orbits

  • Geostationary automated hot spot detection suffers from low confidence problems caused by spatial resolution limitations


Slide46 l.jpg

Acknowledgements

  • INPE – CPTEC

  • United States Forest Service – USFS

  • CIRA – Colorado State University

  • NASA Goddard Space Flight Center

  • University of Maryland

  • National Geophysical Data Center

  • World Bank



ad