modis land data processing – current situation and progress

MODIS Land Data Processing – Current Situation and Progress

MODIS Land Data Processing – Current Situation and Progress

APEIS Capacity Building Workshop on Integrated Environmental Monitoring of the

Asia-Pacific Region

Beijing, ChinaSeptember 20, 2002

Mr. Robert WolfeNASA GSFC Code 922, Raytheon ITSS

Sept. 20, 2002 – APEIS Workshop – R. Wolfe 2

Moderate Resolution Imaging Spectroradiometer (MODIS)

Moderate Resolution Imaging Spectroradiometer (MODIS)

• Launched December 1999 on NASA EOS Terra Polar Orbiting Satellite

• Whisk-broom scanner, 100° Field of View

• Orbit: Attitude 705 km, Inclination 98.2°, mean-period 98.9 min., 16-day repeat cycle, 10:30 am equatorial crossing time

• Senses entire equator every two days, daily full coverage above 30° latitude

• 36 spectral bands: 29 - 1 km, 5 - 500 m and 2 - 250 m nadir spatial resolution

• Second MODIS on Aqua Satellite

– 1:30 PM equatorial crossing time


Terra Launch: Dec. 18, 1999First Image: Feb. 24, 2000

Aqua Launch: May 04, 2002First Image: June 26, 2002

Sept. 20, 2002 – APEIS Workshop – R. Wolfe 4July 25, 2002 – Lower Baja Peninsula

MODIS Terra(1750)

MODIS Aqua(2100)


MODIS Land BandsMODIS Land Bands

Bandnumber

Spatialresolution

Wavelength,nm

Waveband region

1 250 m 620-670 Red

2 250 m 841-876 Near-infrared

3 500 m 459-479 Blue

4 500 m 545-565 Green

5 500 m 1230-1250Near-infrared

6 500 m 1628-1652Shortwave infrared

7 500 m 2105-2135Shortwave infrared


Science TeamScience Team

• Formed 11 years ago

• 3 Disciplines: Land, Oceans, Atmosphere

• Responsibilities:– Algorithm Theoretical Basis Document

– Algorithms in Software: PGEs (Product Generation Executables)

– QA and Validation

• Support Groups– Science Data Support Team (SDST)

• Includes: Discipline Support Groups (Coordination)

– MODIS Calibration Support Team (MCST)• Includes: MODIS Flight Operations


MODIS Land Science Team ProductsMODIS Land Science Team Products

• Goals :

– operationally produce terrestrial remotely sensed products that may be used by expert and non-expert community

– establish a 10 year record that has continuity with precursor systems, e.g., AVHRR, and the future NPP and NPOESS VIIRS missions

• The MODIS products were developed primarily to serve the global change research community (MODIS has global, near daily coverage) and have many other potential applications

• The MODIS Science Team was completively selected to develop peer-reviewed product generation algorithms (24 Principal Investigators, lead: Vince Salomonson)


MODISTrue color

East Coast ZoomMarch 3 2000


MODIS Production and DistributionMODIS Production and Distribution

GSFC DAAC

Level 0 Instrument Data

NSIDC DAAC EDC DAAC

MODAPS

Level 1 Products

Ocean and Atmosphere

Products

Land Snow and Ice Cover

Products

Land Non-cryospheric

Products

User Community

Level 1, Ocean and Atmos.

Products

User Community


MODIS Land Products* / ESE Research Themes

• Energy Balance Product Suite– Surface Reflectance– Land Surface Temperature– BRDF/Albedo– Snow Cover

• Vegetation Parameters Suite– Vegetation Indices– LAI/FPAR– NPP/PSN

• Land Cover/Land Use Suite– Land Cover/Vegetation Dynamics– Vegetation Continuous Fields– Vegetation Cover Change– Fire and Burned Area

Global Water Cycle and Energy

Balance

*Dependencies between products

Biology and Biogeochemistry of Ecosystems and the Global Carbon Cycle

Land Cover and Land Use Change

Atmospheric Chemistry and Aerosols


Product CharacteristicsProduct Characteristics

• HDF-EOS Format

• Average file size is about 70 MB

Type Interim Archive Total L2 Swath 1 5 6 ISIN/SIN 8 30 38 LAEA (Polar) 3 2 5 CMG (Geog.) 0 19 19 Total 12 56 68


Level 1 Process FlowLevel 1 Process Flow

EDOS

Level 0 Science, Ancillary

C1 Level 1A,

Geolocation

C3 Level 2 Cloud

Mask, Atmospheric

Profiles

D0 Ancillary Data (DAO, NMC)

C2 Level 1B Calibrated Radiances


Calibration Summary Calibration Summary

Terra MODIS• Terra MODIS has been stable for over 22 months• The multiple OBCs have proven to be valuable for on-orbit

characterization• Changes in system level response are well understood and have been

incorporated into the L1B software• One subsystem (power supply) has failed and a second (formatter) is

marginal (was just replaced) • There are no indications of additional problems• Current Geolocation is very accurate (~50 m)

Aqua MODIS• Aqua MODIS’ ground tests indicate performance better than Terra

MODIS• Initial on-orbit data indicates only minor changes from the pre-launch

test resultsBarnes/MCST

MODIS identifies

cloudclasses

Hi cloud

Mid cloud

Low cloud

Snow

Clear

Menzel/MODIS

Atmosp.


C1, C2, C3 Level 1 Geo.,

L1B, Cld. Mask, Atmos. Prof.

L1 Level 2 Snow, Sea Ice, Fire

L3 Level 2 Land

Surf. Refl.

L4/4d Level 2/3 Daily

Land Surf. Temp./Emiss.

D0 Ancillary Data (DAO, NMC)

L5P Level 2G/3 Daily

Snow, Sea Ice

L5 Level 2G/3 Daily Geoang, Pointers,

Agg./Text., Land Surf. Refl.,

Fire, Snow

L10 Level 3 8-day

Land Surf. Refl., Fire (+daily),

Land Surf. Temp./Emiss. Snow, Sea Ice,

LAI/FPAR

Future

L12 Level 3 16-day

VI, BRDF, Intermed. VCC

L16a/b Level 3 daily,

8-day PSN

L14 Level 3 32-day Land Cov. DB,

VCC, VCF

L18 Level 3 96-day

Land Cover/Change

L16c Level 3 yearly

NPP

MODIS Land Process Flow


Surface Reflectance/Atmospheric Correction Products

Surface Reflectance/Atmospheric Correction Products

• Reflectance corrected for gaseous and aerosol scattering and absorption, surface adjacency effects caused by variations in land cover and atmospheric/surface coupling effect

• Building block for generation of a number of land products: VI, BRDF/Albedo and LAI/FPAR


Southern Africasurface reflectance - not correctedfor aerosols

MODLAND/Vermote


Southern Africa correctedfor aerosol

MODLAND/Vermote


8-Day Surface Reflectance Composite (detail)8-Day Surface Reflectance Composite (detail)

At-launch algorithm(till Nov. 2000)

Revised algorithm (Min Blue & shadow filter & min view angle)

MODLAND/Vermote


MODIS BRDF/Albedo ProductMODIS BRDF/Albedo Product

• Provides global measures of albedo, BRDF adjusted surface reflectance and surface anisotropy

• Inputs – Gridded Surface Reflectance

• Outputs – 16 days at a 1km gridded spatial resolution. – Albedos

• Bihemispherical (white-sky) • Directional hemispherical (black-sky) albedos local solar noon• 7 Spectral and 3 Broadband Albedos

– Nadir BRDF-adjusted Reflectances (NBAR)• 7 Spectral bands at mean overpass solar zenith angle

– BRDF kernel weights in 7 spectral bands


0.0 0.1 0.2 0.3 0.4 0.5

Global Broadband White-Sky Albedo (0.3-5.0µm)April 7–22 2001

10 km resolution, Hammer-Aitoff projection,produced by MODIS BRDF/Albedo Team

no data

MODLAND/Strahler et al.

No data


Global Composite Map of Nadir BRDF-Adjusted Reflectance (NBAR)April 7–22 2001

10 km resolution, Hammer-Aitoff projection,produced by MODIS BRDF/Albedo Team

no data

MODLAND/Strahler et al.

True color, MODIS Bands 2, 4, 3 No data


Key parameter of the global energy balance, reflecting much of the incident solar radiation back to space

Snow and Ice CoverSnow and Ice Cover

Eight-Day Composite Global Climate Modeling Grid Snow Map (5.6-km resolution, 1/20°) December 18 - 25, 2000


Nov 1-7, 2000 Nov 8-15, 2000

Change in maximum snow extent between two Change in maximum snow extent between two composite periods seen above (1.8 million sq. km)composite periods seen above (1.8 million sq. km)

9.0 million sq. km of snow cover 10.8 million sq. km of snow cover

Snow on both

Snow on Nov 1-7 only

Snow on Nov 8-15 only

Clouds

MODLAND/Hall et al.


Land Surface Temperature (LST) and Emissivity

Land Surface Temperature (LST) and Emissivity

• Split-window and statistical regression techniques are used

• Good indicator of both the energy balance at the Earth’s surface and the greenhouse effect

• Useful for a wide variety of climate, hydrological, ecological, and biogeochemical studies

Eight-Day Composite day-time 1km LST (k) Sept. 13 - 20, 2000


Daytime LST as the red component, and Nighttime LST as the green and blue components (after histogram-equalization enhancement). Pixels with daytime LST values but without nighttime LST values are in red..

8-day LST Day-Night Comparison

LST NightLST Day

Color Composite


Vegetation IndicesVegetation Indices

• Used for global monitoring of vegetation conditions– Normalized Difference

Vegetation Index (NDVI) and an improved Enhanced Vegetation Index (EVI)

– EVI uses self-correcting atmospheric and soil calibration factors

– Play an important role in the derivation of the land cover and land cover change, FPAR, LAI, and thermal products


NDVI vs. EVINDVI vs. EVI

NDVI EVI


LAI and FPARLAI and FPAR

• Leaf Area Index (LAI) and Fractional Photosynthetically Active Radiation (FPAR)– LAI defines an important

structural property of a plant canopy which is the one-sided leaf area per unit ground area

– FPAR measures the proportion of available radiation in the photosynthetically active wavelengths (400 to 700 nm) that a canopy absorbs

LAI March-25-2000


MODIS LAI

MODIS FPAR

MODLAND/Myneni et al.


MODIS Global 1-km Land CoverMODIS Global 1-km Land Cover

Inputs Primary:

View-angle corrected (nadir) surfacereflectance (NBAR), 7 land bands

Ancillary:

Spatial Texture from 250-m; enhanced vegetation index; snowcover; land surface temperature; directional information

Outputs

IGBP 17 classes, UMd 14-classes; BGC 6 BiomesClassification Confidences

Approach Supervised approach, Artificial neural networks, Decision trees, GAM Statistical Boosting Global Test Sites for Training and Testing System for Terrestrial Ecosystem Parameterization ( 920 sites ) – site labeling

test sites for training

MODLAND/Strahler et al

Beta Product Release date – April 27, 2001; Provisional Product June 1, 2001


Vegetative Cover Conversion and Continuous Fields

Vegetative Cover Conversion and Continuous Fields

Vegetative Cover Conversion

• Detects and labels changes in land cover

• Uses the 250m Surface Reflectance products

Vegetation Continuous Fields

• Percent cover for basic land cover types

• Post-launch product

Vegetation Continuous Fields


IGBP Land Cover Units (17)IGBP Land Cover Units (17)

Based on Life-Form, Height, and Cover Fraction

Natural Vegetation (11)

Evergreen Needleleaf

Forests

Evergreen Broadleaf Forests

Deciduous Needleleaf Forests

Deciduous Broadleaf Forests

Mixed Forests

Closed Shrublands

Open Shrublands

Woody Savannas

Savannas

Grasslands

Permanent Wetlands

Developed and Mosaic Lands (3)

Urban and Built-Up Lands

Croplands

Cropland/Natural Vegetation

Mosaics

Nonvegetated Lands (3)

Snow and Ice

Barren

Water Bodies


Provisional Land Cover Product June 01


Fire and Thermal AnomaliesFire and Thermal Anomalies

• Fire is an important ecosystem process and source of trace gas emissions.

• Identifies the location, timing and energy of fires

• Near-real time alerts of volcano and fire activity (experimental product)

• Burned area estimation and for use with global models (post-launch product) MODIS Fires, NW Australia, Oct 2, 2000


MODIS burned area composite NFDR Fuel model EF and CF from Ward and Hardy (1991)x x

Idaho/Montana September 5, 2000

0

500000

1000000

15000002000000

2500000

30000003500000

4000000

45000005000000

CO2 Fireemissions fromthis scene (Gg)

1995 Fossil FuelCO2 (Gg)

% Area burned in this scene relative to the

total in the U.S. in 2000

14%

86%

MODIS Fire Emissions


Left burned area algorithm results -> burning over days 249-290 (500m)

Right temporal composite of MODIS day and night active fires detected over same period (1km)

Keypurple - beginning of the time series (day 249) red - end of time series (day 290)white - insufficient data in the time series to make a burning decision

MODIS Burned Area Product Evaluation

MODLAND/Roy


MODIS Production and DistributionMODIS Production and Distribution

• MODIS production commenced Feb. 2000• Collection 1: Beta

– L1: 2/00 to 5/01– L2+ 2/00 to 3/01

• Collection 3: “Provisional” – L1: 2/00 to 9/02 (from 11/00 onward validated)– L2+: 11/00 to 9/02

• Collection 4: “Validated”– Ocean from 6/02 to 10/02 at 6 to 8X– Starting 11/02 for L1 and L2+ Land and Atmosphere at 3 to 4X– Reprocessing – 2/00 to 10/02– Extend time series back in time (preliminary):

A) 11/01 to 10/02; B) 11/00 to 10/01; C) 2/00 to 10/00


Software Development ProcessSoftware Development Process

Insert Testing

Science Software Transfer

Science Team Member

Develops Code

DAAC

Operations

Integration and Test Science

Dependency Testing

Science Data

Support Team


Science Code DeliveryScience Code Delivery

• Alpha/Beta (2-3 years before launch)– simulated data (MAS data)– data system test, sizing– interface testing

• Version 1 (1 year before launch)• Version 2 (at launch)

– Beta product quality

• Version 3 (first reprocessing)– Provisional product quality

• etc. (validated)


Processing Driven by Production RulesProcessing Driven by Production Rules

• Required and optional inputs– Upstream products, ancillary data– How long to wait

• Processing Period– 5 min., 1, 8, 16, 32 or 96 days, calendar

months, calendar years)

• Start of period (reset at beginning of year, 2-3 day overlap at end of year)

• On-line and near-line retention time


MODAPS DesignMODAPS Design

• Loader– Determines when products can be generated (generates individual tasks)

• Scheduler– Schedules all tasks

– Allocates system resources, monitors task

– Stage inputs, output product disposition

• Archiver– Moves data from disk to tape and back

• Importer/Exporter– Ingests L1 and Ancillary Data from DAAC

– Distributes L2+ data to DAACs and users

• GUIs (Graphical User Interfaces)– X-windows (Operator) and Web based (Operators and other users)


MODAPS HardwareMODAPS Hardware

• 80 Processors – 10 ECS Gflops– 6.3 Gflops available in ops. (excluding overhead)

– Compare to 4.2 Gflops load in ‘96 baseline

– Most processes’ load lower than baseline

• 44 TB disk – 800 MB/s aggregate access• 100 TB tape – 60 MB/s aggregate access

– Theoretical rate: 5.1 TB/day

– Actual for 8/16-day retrieves: 0.5 to 0.75 TB/day

– Space for about 100 days of data


MODAPS HardwareMODAPS Hardware

2 Fi

bre

Cha

nnel

s10

0Mby

tes/

sec

14 F

ibre

Cha

nnel

s10

0Mby

tes/

sec

HiP

PI c

hann

el80

0MB

/sec

SGI Origin 200080 250Mhz R10K processors

40GB memory

DAACs

100TB ADICTape Library

12 drives44TB RAID in44 filesystems

8 Fi

bre

Cha

nnel

s10

0Mby

tes/

sec

HiP

PI c

hann

el80

0MB

/sec

SGI Origin 300064 400Mhz R12K processors

32GB memory

DAACs

100TB ADICTape Library

12 drives35TB RAID in35 filesystems

2 Fi

bre

Cha

nnel

s10

0Mby

tes/

sec

Gigabit Ethernet100MB/sec

Science Team,Development and

Test Systems

Gigabit Ethernet100MB/sec

Science Team,Development and

Test Systems

4 Pentium III cpu’sDatabase ServerRunning Linux

4 Pentium III cpu’sDatabase ServerRunning Linux


Design IssuesDesign Issues

• Number of files vs. product size

• Disk vs. Tape

• Production rule complexity

• System reliability

• Interfaces (leaky pipes)

• Product Format– Map projection, Tiling – for global products

– Usability – heritage vs. improvements

– HDF-EOS, Metadata

• Network

• Long term archive


MODLAND GridsMODLAND Grids

• Fine resolution grids for the L2G, 3 and 4 products are based on two map projections: – Integerized Sinusoidal (ISIN) => Sinusoidal (SIN) Grid– Lambert Azimuthal Equal-Area (LAEA) (polar grids) – Almost all of the fine resolution products will be made in the

ISIN/SIN• Exception: Sea-ice products are made in the in the LAEA

projection with the grid centered at the north and south poles (EASI Grid implementation)

– Grid cell size varies by product and is either (approx.) 0.25 km, 0.5 km or 1 km – actual size depends on the projection

– Each grid is broken into non-overlapping tiles which cover approx. 10 x 10 deg. area

• Coarse resolution global Climate Modeling Grid (CMG) products are made in a geographic projection with grid cell sizes of 0.25 or 0.5 deg.


ISIN/SIN GridISIN/SIN Grid


MODIS Land ValidationMODIS Land Validation

• Uses a hierarchical measurement concept – focused on EOS Land Validation Core Sites – combining satellite and airborne data with ground based in-situ

measurements– augmented by established field sites and measurement networks,

e.g., LTERS, FLUXNET, AERONET

• Interacting with other Terra, NASA Earth Science Enterprise, and international instrument teams (GLI, ATSR, MERIS)

• Participates in community intensive field campaigns (LBA – South America, SAFARI 2000 – southern Africa)

• Collaborating with the data providers to make validation data openly available


Multiple Observations – Simple Case

Multiple Observations – Simple Case

Obs 2Obs 1

Obs 3 Obs 4Obs 2

Obs 1

a. About 50 % Overlap b. 25 % Overlap

Cell Cell

Multiple observations covering a single grid cell:• about 50% of overlap distributed into two observations, and • 25% of overlap distributed over four observations.

Observation Coverage (%)

Percent of Observations

75 to 100 16

50 to 75 57

25 to 50 37

0 to 25 0

Distribution of coverage for each of the observation with the largest intersection with a cell. The observations and grid cell are the same size and have the same orientation.


L2G StackL2G Stack

L2G product with samespatial dimensions ascorresponding L3 tile

obs 3

obs 1

obs 2

obs i

Stack ofobservations peroutput grid cell

Observation i

Pointer information:granule pointer,line, sample, etc.

Geophysical parameters:e.g., land surf. reflectance,thermal anomalies, etc.

Viewing geometry:view zenith,solar zenith, etc.


MODIS Land QAMODIS Land QA

• Objectives– identify suspect and poor quality products, including propagation

of errors originating from upstream products and ancillary data– provide feedback to appropriate entities so that problems can be

rectified– flag product quality (via metadata stored at the DAACs)

• QA approach– lead by the LDOPE (Land Data Operational Product Evaluation)

facility – activities:

• Product specific QA tools • LDOPE QA product metadata database• Global browse images


Global BrowseGlobal Browse

http://modland.nascom.nasa.gov/browse


MODLAND SubgroupMODLAND Subgroup

MODIS Science Team Meeting, Jan 26 2001


MODIS Snow CoverMODIS Snow over AsiaMODIS Snow over Asia

modis land data processing – current situation and progress

Documents

modis aqua

modis production

productsmodis land products

global carbon cycle

products biology

aqua launch

sensed products

area global water cycle