International EOSNPP Direct Readout Meeting 2005

1
Biodiversity Monitoring The MODIS and AVHRR DB
Station, Mexico Gerardo Lopez, Isabel Cruz, Bart
Wickel, Joanna Acosta, Sergio Cerdeira. Comisión
nacional para el conocimiento y uso de la
biodiversidad - CONABIO
International EOS/NPP Direct Readout Meeting
2005 Benevento, Italy October 3-6, 2005
2
Comisión nacional para el conocimiento y uso de
la biodiversidad
National commission for knowledge and use of
biodiversity
Inter-ministerial commission headed by the
President of Mexico

• Main tasks / objectives
• Maintenance of the national biodiversity system
  (SNIB)
• Support biodiversity studies
• Advise governmental institutions and other
  public sectors
• Divulge information on biological diversity

3
CONABIO
structure
internal

• Support for GAP analysis
• Priority regions
• Invasive species

• Central American cooperation early forest fire
  detection

• Review and evaluation of Remote Sensing related
  projects sponsored by CONABIO

• Analysis of satellite imagery

• Projects- Agave- Islands

4
CONABIO
Cooperation
international
national

• Fire detection algorithm
• Burned area mapping

• New antenna Chetumal
• MODIS dataSAR data

• National system for rapid forest fire detection
• Burned area mapping
• Training

• Technology transfer web portal

• Central American cooperation early forest fire
  detection
• Burned area mapping

5
Remote Sensing Group
Activities

• Forest fires (MODIS)
• Rapid response system
• Fire propagation risk
• Burned area mapping
• Vegetation monitoring (MODIS)
• Time series analysis NDVI / EVI, LAI, FPAR, NPP
• Rapid environmental assessment (MODIS)
• NDVI anomaly mapping ----- Fire propagation risk
• Case studies (Landsat, SPOT, IKONOS)
• Land use change studies
• Applied ecological studies

6
MODIS direct broadcast station
Remote Sensing Group
MODIS DB station Mexico City

• MODIS DB station (X band antenna)
• 8 passes per day receiving MODIS Aqua and
  Terra
• AVHRR (L band antenna)
• Mainly NOAA 12 and 16

7
MODIS direct broadcast station
Remote Sensing Group

• Moderate resolution imaging spectroradiometer
  (MODIS)
• 36 bands
• 250m - 500m - 1 km resolutions
• Advanced Very High Resolution Radiometer (AVHRR)
• 5 bands
• 1.1 km resolution

Ecosystem applications terrestrial
studies Hot-spots detection Vegetation
monitoring Land cover change detection
studies Oceanic studies Sea Surface Temperature
(climate change) Chlorophyll -a concentration
(ocean colour)
MODIS DB station Mexico City
8
Other data sources
Remote Sensing Group

• OTHER data sources
• SPOT antenna Mexican Navy - INEGI
• Landsat data archive
• SPOT
• 4 bands
• Visible-IR
• 2.5-20 m resolution
• Applications
• Vegetation studies
• Landuse change studies
• Pattern recognition (natural gas wells
  identification)

9
Rapid fire detection
Program history
Publication for Guatemala Costa Rica El Salvador
Implementation of other sensors
Automation of the program
Hot spot detection with AVHRR and MODIS
DMSP
1998
1999
2000
2004
2001
2003
2002
2005
2006
Start hot spot detection with AVHRR and same day
publication
Increase in response time
Fire propagation index, publication for entire
Central America
Publication for Guatemala
10
Rapid fire detection program
Program history

• Problem
• Severe wild fires during El Niño year 1998
• 2000-2004 7533 wildfires annualy
• 196,972 ha affected annually
• Causes
• Agricultural activity, 43
• Intentional, 21
• Accidental, 31
• Other, 5
• Affected vegetation
• Pasture 40
• Shrub and woodland 41
• Forests 19

(Source CONAFOR, 2005)
11
Rapid fire detection program
processing
Reception
AVHRR
MODIS
Total time between reception and publication 1.5
hours gt 40 min
12
Rapid fire detection program
Web-products

• Web fire products Mexico and CA
• Response time next fire season lt 40 min
• Detailed maps of forest fire sites
• Low to high tech information provisione-mail gt
  ArcSDE server gt pdf reports gt ArcIMS map

Report tables
Quicklook images
Hot spot images
EVI images
13
Rapid fire detection program
Mexico Central America
Guatemala
Honduras

• Total time between reception and publication
• Currently 1.5 hours Jan 2006 lt 40 min
• Suite of productsLow gtgt High tech

Nicaragua
14
Fire propagation risk
analysis

• Historical NDVI series, 36 months

Historical NDVI
HANTS
Theoretical NDVI
Normalizeddifference
Actual NDVI
NDVIAnomaly

• Vegetation Cycles
• Perennial
• Deciduous
• Agricultural areas

15
Fire propagation risk
Product validation

• April 2004
• Red areas indicate high negative NDVI anomaly
• Blue points indicate detected hot spots

• NDVI anomaly ? a fire prediction tool
• Indicates areas with a higher risk of
  propagation

16
DLR antenna Chetumal 2006
Future data

• ASAR (25m, 15-45, HH, VV, HV, C-band)
• ERS-2 (10, 30 m, 23, VV, C-band)
• TerraSAR (1, 3, 16 m, HH, VV, HV, X-band)
• Forest fragmentation
• Biomass estimation
• Fire scar mapping
• Forest mapping
• Wetland delineation
• Soil moisture studies
• DEM generation
• 2007 MERIS
• Fire detection
• Vegetation monitoring
• 2007 Rapid eye
• Regional vegetation studies
• 2008 IRS??, IKONOS??

DLR X-band antenna Chetumal
17
Future projects

• National level
• NDVI / EVI time series analysis
• Vegetation anomalies and forest fires
• LAI / FPAR / NPP
• VCF analysis for deforestation and fragmentation
• Rapid response burned area mapping
• Regional level
• Case study Magueyes
• Case study Mangroves
• Validation
• LAI / VCF products

18
Cloud forest project
Cloud occurrence
Cloudiness map derived from MODIS data (Jan - Jun
2005)
gt60
19
Cloud forest project
Cloud occurrence
gt60
20
Cloud forest project
Cloud occurrence
gt60
21
Cloud forest project
Remote Sensing studies

• Remote Sensing Studies
• Cloud properties (Occurrence, height,
  temperature etc.)
• Forest fragmentation
• Effects of land use change on cloud base

22
Cloud forest project
analysis

• Field studies
• Catchment waterbalance
• Hydro-meteorological profiles
• Quantification of water gains
• GIS analysis
• Analysis of biodiversity databases
• Definition of recuperation areas

23
Burn scar detection
fire effects
Every year in Mexico more than 80,000 hectares
are affected by fire. Fires have a huge impact in
global and local environmental processes Identifyi
ng and mapping burned areas is a critical stage
in fire management
Photography by Fulvio Eccardi
24
Burn scar detection using MODIS, 500m, 32-Day
composites
as part of AQL 2004 (Latin America Burn Scars
project)

• Objective
• Identifying and mapping burn scars for entire
  Latin America using MODIS imagery with a common
  methodology but taking advantage of the local
  expertise to generate a product adapted to every
  region.
• Main tasks defined in Nov. 2004 (Santiago, Chile)
• Definition of methodology for mapping burn scars
• Responsibilities for each participating
  institution
• Output products and data dissemination
• Objectives for next workshop in Dec. 2005
  (Mexico City)
• Integration of results
• Expose process of validation
• Comparison of local methodologies

25
Input data for burn scar algorithm MODIS data
Continental subsets of MODIS 500m 32-days
composites derived from 8-day surface reflectance
composites (MOD09A1 product) obtained from GLCF -
UMD.
26
Input data for burn scar algorithm NDVI
NDVI Normalized Difference Vegetation Index
calculated from MOD09A1 product.
27
Input data for burn scar algorithm NBR
NBR Normalized Burn Ratio (Rogan and Franklin,
2001) which is a normalized ratio of NIR and SWIR
bands, calculated from MOD09A1 product.
28
Output data for burn scar algorithm
Burn scars detected in areas of at least 30 of
tree cover based on VCF Vegetation Continuous
Fields (MOD44B) product from 2001 obtained from
GLCF - UMD.
29
Change detection
Input File of Fourier components Method Algebra
of bands

• Band by Band
• Layerstack

30
Areas with change
31
Difference between mean and anual

• Input
• File with smoothed EVI values for 4 years
• Layerstack with files of smoothed EVI values for
  each year

32
Hardware
analysis
Sun sparc300GB 1.0 GHz Solaris 5.7 TeraScan
40GB DLT
MODIS level 0
Dell Power edge 28002TB dual 3.6 GHz CentOS
4.1 TeraScan NASA DAAC processing
MODIS level 1 2
Sun fire V8901 TB quad 1.2 GHz Fire
processing GIS analysis
Sun StorEdge L8 8x 320 GB Tape
5TB SAN
Fire maps Image processing
Dell poweredge2 TB dual 3.0 GHz Image web server
Sun fire V210500 MB dual 1.34 GHz Fire web server
33
Future objectives
analysis

• Improvements Fire monitoring program
• Burned area mapping 2004-2005
• New fire-portal January 2006
• Reduction of response time to lt 40 minutes
• Fire propagation risk modelling
• Data distribution
• MODIS imagery
• MODIS data products Mexico and Central America
• Regional case studies
• Cloud forest
• Mangroves
• Invasive species

34
Future objectives Sea Surface Temperature
analysis
35
NDVI and hot-spots (in red) detected in 2005,
Mexico