Landsat Ecosystem Disturbance Adaptive Processing System

LEDAPS Update

Jeff Masek, NASA GSFC 10/27/2009

Landsat Ecosystem Disturbance Adaptive Processing System

descended from MODIS Adaptive Processing System (MODAPS) LEDAPS as a Project . North American forest disturbance, 1990-2005 LEDAPS as a Processing System large-area, reflectance-based Landsat analyses

Background
Global estimates of carbon fluxes often exclude effects of land cover change and disturbance Patch size often small requires Landsat-type data analysis North American Carbon Program Science Plan calls for analysis of disturbance from satellite data
2001 modeled NEP fluxes
source gC/m2/yr sink

source: Potter, 2003, EOS

7km

1985

1988

1999

LEDAPS Processing Overview

Landsat Geocover TM, ETM+ Calibration Atmospheric Correction Cloud/Snow masking Landsat MSS

Radiometric Normalization

Radiometrically Consistent Surface Reflectance Dataset (1975-2000)

Disturbance (biomass gain) / Regrowth (biomass loss) via change detection

Aggregation

Disturbance/Recovery Products for Carbon Assessments

QA/ Validation

Analysis

Preprocessing

Disturbance Index Example

Olympic Peninsula

1988
5km

2000

Disturbance Index Change

Map

LEDAPS 1990s Disturbance rate ~0.9%/yr US average; Stand-replacing only

0 50 100 % cell area disturbed 1990-2000

Masek et al., RSE, 2008

Sampling Approach
S.N. Goward, North American Forest Disturbance and Regrowth since 1972

Disturbance history

~25 Sample Sites

- random sample stratified by forest type - constrained by forest cover per scene and geographic dispersion - Known sampling probability per scene

Biennial Image Time Series (1972-2004)

Regrowth dynamics

10-year revisit intervals are too long for mapping disturbance with Landsat-type data
1 0.9 0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1 0

Virginia (p15r34)

Fraction of Disturbance Detected by LEDAPS

1988

1990

1992

1994

1996

1998

2000

Year of disturbance

0 0.4x105 LEDAPS Area Disturbed (ha)

0 1.0x105 FIA Forests < 20 years (ha)

FIA Forest Age vs. LEDAPS disturbance (150km cells)

FIA Area < 20 years / 20.0 = Stand-scale turnover area per year (~1985-2005) LEDAPS = stand-clearing disturbance (~1990-2000)

LEDAPS Acres/Yr Disturbance

R2=0.749

FIA Acres/Yr Turnover (per 150km cell)

Average Forest Age 2005-2105

2005

<20

>135

2055

2105

Age

Age

Age

Sim. Year

Sim. Year

Sim. Year

Most US Forests are not in equilibrium with current disturbance rates SE, PNW nearly in equilibrium - some future aging Northern forests get older (lower disturbance) Rocky Mountain forests get younger (higher disturbance)

Net Ecosystem Productivity

2005

2055

2105

2.0e5 tC/yr
4.00E+08

3.50E+08

3.00E+08

US Forest Sink (tC/yr)

2.50E+08

Turner (1995) = 2.36e8

2.00E+08

NAFD * 1.50
1.50E+08

Williams/Collatz= 1.68e8

1.00E+08

NAFD * 0.50

5.00E+07

0.00E+00 1990

2010

2030

2050

2070

2090

2110

Year

Landsat Disturbance Albedo

Disturbance (fire) has been linked to albedo increase (cooling) What is the effect of harvest, forest management, etc? Initial study using MODIS Albedo product and space/time substitution Feng Gao/ Yanmin Shui leading effort to create Landsat albedo products

LEDAPS Architecture
INGEST lndpm Metadata parsing AROP Automated registration and orthorecification Calibration, TOA reflectance Cloud/snow/ shadow mask 6S Atmospheric Correction to SR Revised reflectancebased cloud mask

Scene RDBMS

PROCESSING

lndcal

Job Scheduling & Control

lndcsm

Data

lndsr

cloud.for

SR products

Individual science modules are available for download

Atmospheric Correction
Based on MODIS/6S radiative transfer approach water vapor from NCEP re-analysis data ozone from TOMS, EP-TOMS topographic-dependent Rayleigh correction Aerosol optical thickness estimated from imagery using the Kaufmann et al (1997) Dense, dark vegetation approach - estimate blue reflectance based on TOA SWIR 2 - difference between TOAblue and SRblue gives AOT - interpolate valid targets across image

ETM+ Comparison with MODIS

900

Reflectance (*0.0001)

Red spectral band

600

300

0 130
4000

160

190

220

250

280

Day of Yea r 2 001 MODIS red (Band 1) ETM+ red (Band 3) Near-infrared spectral band

Reflectance (*0.0001)

3000

Saskatchewan, Canada
Landsat (LEDAPS) reflectance MODIS daily reflectance (highest QA observations)

2000

1000 130 2500

160

190

220

250

280

Day MODIS NIR (Band 2) of Year 2001 ETM+ NIR (Band 4) Shortwave infrared spectral band (1.55-1.75 mm)

Reflectance (*0.0001)

2000

1500

1000 130

160

190

220

250

280

MODIS Band 6

Day of of Year Day Year 2001

ETM+ Band 5

Name Curtis Woodcock Randy Wynne Susmita Sen Thomas Hilker Mike Wulder Jennifer Dungan Rama Nemani Gail Schmidt Tom Maiersperger Saurabh Channan Steve Prince Arnon Karnieli Matthew Smith Daniel Slayback Jin Chen Zhe Zhu Gabriele Bitelli Qingling Zhang Paul Morin Erica Meta Smith Devendra Singh Anderson, Martha Egidio Arai Nicholas Coops Junchang Ju Junli Li Marion Stellmes Zhengwei Yang Glynn Hulley Crystal Kolden

Institution Boston U. Virginia Tech Virginia Tech. U. U. British Columbia CFS ARC ARC SAIC SAIC UMD UMD Ben Guirion U. GSFC GSFC Beijing Normal U. Boston U. Universit di Bologna Columbia University University of Minnesota UC Berkeley Department of Science and Technology, INDIA USDA Brazilian Institute for Space Research University of British Columbia South Dakota State U. UCLA University of Trier, Germany USDA NASS NASA JPL Clark U.

email curtis@bu.edu wynne@vt.edu susmita@vt.edu thilker@interchange.ubc.ca Mike.Wulder@NRCan-RNCan.gc.ca Jennifer.L.Dungan@nasa.gov rama.nemani@nasa.gov Gail.L.Schmidt@saic.com Thomas.K.Maiersperger@saic.com schannan@umiacs.umd.edu sprince@umd.edu karnieli@bgu.ac.il matthew_smith@ssaihq.com daniel.slayback-1@nasa.gov chenjin@ires.cn zhuzhe@bu.edu gabriele.bitelli@unibo.it zqling.bu@gmail.com lpaul@umn.edu forester@nature.berkeley.edu ds.chahar@nic.in Martha.Anderson@ARS.USDA.GOV egidio@ltid.inpe.br nicholas.coops@ubc.ca Junchang.Ju@sdstate.edu jli@geog.ucla.edu stellmes@uni-trier.de Zhengwei_Yang@nass.usda.gov glynn.hulley@jpl.nasa.gov ckolden@clarku.edu

LEDAPS x x x x x x x X x x x x

STAR-FM

AROP

x x x x

x x x x x x x x x x x x x x x x x x x x

Current Activities
Extend LEDAPS disturbance mapping to 5-year epochs, 1990-2005 and integrate available ASTER imagery (thru 2011) Support Goward et al North America Forest Dynamics (NAFD) project (thru 2011) Support UMD MEASURES Forest Cover Change ESDR (thru 2013) Support Gitelson et al NACP project (corn/soybean NPP; thru 2011) Additional collaborations with Nemani/ARC; Roy/SDSU; Wulder/CFS, Wynne/VTU; Chen/UT

ESDR of Global Forest Cover Change

Major deliverables:
Global, Fine resolution (< 100 m) forest cover change (FCC) ESDR
1990-2000 2000-2005 1975-1990 for southern South America

Global fine resolution (< 100 m) surface reflectance ESDR

1990, 2000, and 2005;

Global 250-m vegetation continuous field (VCF) based FCC ESDR from 2000 to 2005;

Other Deliverables:
Fragmentation products derived from the fine resolution FCC products; FCC ESDR products aggregated from the fine resolution and the 250 m FCC products to 250 m, 500 m, 1 km, and 0.05 grids for use by carbon, biogeochemical and hydrological modelers; Subsets of the above products for protected areas of the world and their buffer zones.

Suggestions for Products

Landsat-based surface reflectance should be pursued
- facilitates biophysical applications (e.g. NDVI, LAI), integration with canopy reflectance models, data fusion, and robust time series analysis - not inherently more difficult than MODIS, ASTER - need validation over special terrain (ice, bright desert)

Accurate cloud mask probably more difficult

Great opportunities for integrating MODIS/VIIRS and Landsat

- direct radiometric fusion a la STAR-FM - use MODIS to establish expected seasonal signal (phenology pattern), compare to Landsat observation - ARC, SDSU Measures projects as prototypes

How to move toward Global Land Cover / LCC / Disturbance?

- Landsat Project => USGS (NASA?) => Science Community

ETM+ Surface Reflectance Mosaic

Atmospheric Correction

1990s Landsat-5 mosaic

TOA reflectance Surface reflectance

100 km

BOREAS Study Region

100 km 100 km

Effect of Atmospheric Correction

(MOD9A surface reflectance) (ETM+ reflectance), 8/3/00
10
Before AC (TOA reflectance)

8 6 4

After AC (surface reflectance)

8/3/2000 acquisitions

Dr (%)

2 0 -2 -4 -6 Band 1 Band 2 Band 3 Band 4 Band 5 Band 7

TM TOA

TM SR