NASA LCLUC Program and MuSLI Activities

Garik Gutman, LCLUC Program Manager, Landsat Program Scientist

NASA Headquarters Washington, DC

2

Land-Cover/Land-Use Change Program

LCLUC is an interdisciplinary scientific theme within NASA’s Earth Science

program. The ultimate vision of this program is to develop the capability

for periodic global inventories of land use and land cover from space,

to develop the scientific understanding and models necessary to

simulate the processes taking place, and to evaluate the consequences

of observed and predicted changes

• Drivers of LCLUC

• Natural Drivers

• Anthropogenic Drivers

• Socio-Economic Drivers

• Landscape Modification

• Impacts of LCLUC

• Carbon Cycle

• Surface Hydrology

• Atmosphere

Carbon Cycle

and Ecosystems

Focus Research

Area

Terrestrial Ecosystems

Program

Ocean Biology

Program

Biodiversity

Program

Applications

Program Carbon Management

Coastal Management

Water Management

Agri. Management

Land-Cover/Land-Use

Change Program

Water and Energy Cycle

Focus Research Area

Terrestrial Hydrology

Internal NASA

Programmatic

Linkages

Atmospheric

Composition

Focus Area

Radiation

Science

Vuln./ Adapt.

4%

Synthesis Studies 5%

Climate Variability

and Change 6%

Water and Energy Cycle Impacts

7%

Ecosystems and Biodiversity Impacts

8%

Observations and Data/ Detection and

Monitoring of LCLUC 27%

Drivers of Change 11%

Predictive Land Use Modeling

14%

Carbon and Biogeochemical Cycle

Impacts 18%

LCLUC Program Content

Impacts - 33% (Carbon+Water+Eco) Monitoring – 27% LU Modeling – 14% Drivers – 11% LCLUC- Climate interactions - 6% Synthesis – 5% Vulnerability/Adaptation – 4% http://lcluc.hq.nasa.gov

4

Brochure

E-newsletter

Webinar series

200 projects since Program’s inception Each year ~40 3-yr projects Total in the Program >200 researchers

Solicitations in Time • LCLUC-08

– Climate change impacts on land use and contributions to large international non-NASA projects(“small” proposals)

– 66 step-1 received, 48 were encouraged to submit step-2 proposals, 42 were submitted, 18 were selected (9 impacts, 9 “small”)

• LCLUC-09

– Detecting/quantifying and explaining/attributing agricultural or urban land use

– 135 step-1 received, 67 were invited to submit step-2 proposals, 62 were submitted, 13 were selected (3 urban, 10 agriculture)

• LCLUC-10

– Land use in wetland regions (vulnerability, impacts, and adaptation) + LCLUC Synthesis

– 49 step-1 received, 29 were invited to submit step-2 proposals, 25 were submitted, 7 were selected (2 synthesis, 5 wetlands)

• LCLUC-11

– Early career scientists, land-use across political borders

– 90 step-1 received, 26 were invited to submit step-2 proposals, 26 were submitted, 10 were selected

• LCLUC-12

• Continental mapping of industrial forests from Landsat observations + LCUC Synthesis

• 24 step-1 received, 15 were invited to submit step-2 proposals, 16 were submitted, 10 were selected (6 synthesis, 4 industrial forests)

• LCLUC-13

• Changes in land use and land cover in mountainous areas +Synthesis

• 71 step-1 received, 33 invited, 31 submitted, 9 selected (3 Synthesis, 6 mountain land use)

• LCLUC-14 – Multi-source Land Imaging (MuSLI)

• LCLUC-15

• Changes in Agriculture and Urban land use in South Asia + LCLUC Synthesis (69 step-1 received)

• LCLUC-16

• Changes in Forest, Agriculture and Urban land use in Southeast Asia, Caucasus + LCLUC Synthesis

• LCLUC-17 – Multi-source Land Imaging (MuSLI)

Multi-Source Land Imaging (MuSLI) Landsat-Sentinel Team under LCLUC

• Sentinel-1a: launched in Apr

2014 • Sentinel-1b: launched in Apr

2016 • Sentinel-2a: 1.5 yr good

functioning • Sentinel-2b: to be launched in

spring of 2017 • Sentinel-3a: launched Feb

2016 • Landsat-7 & - 8 – nominal

operations

MuSLI NASA Project Scientist Jeff Masek, NASA Landsat-9 Project Scientist

•Prior efforts to synergistically use Sentinel data along with Landsat-8

•Joint NASA-UMD-CNES/CESBIO project •Sentinel-2 NASA Data Use Preparation team •Cross-Calibration (GSFC)

MuSLI ESA Project Scientist Benjamin Koetz, ESA Earth Observations Engineer

Salas: Rice crops product prototype

(South Asia)/Torbick, AG

Applied Geosolution, LLC

International Collaborators:

Hoekman, Wageningen U.

Le Toan, CESBIO

Operational algorithms and products for near real time maps of rice extent and rice crop growth stage using multi-source remote sensing

Huang: Inundation product prototype

(Americas)/Jones, USGS/ Lang, Fish

and Wildlife Service

University of Maryland

International Collaborators:

Creed, Western U., Canada

Towards Near Daily Monitoring of Inundated Areas over North

America through Multi-Source Fusion of Optical and Radar Data

Small: Global Urban Product/ Greg

Yetman, Columbia U./Nghiem, JPL

Columbia U. and JPL

International Collaborators:

Esch, DLR

Multi-source imaging of infrastructure and urban growth using

Landsat, Sentinel and SRTM

Friedl: Global veg. phenology

product prototype/ Gray, BU

Melaas, BU

Boston University

International Collaborator:

Eklundh, Sweden

Multisource Imaging of Seasonal Dynamics in Land Surface

Phenology: A Fusion Approach Using Landsat and Sentinel-2

Roy: Fire Burned Areas

(Africa)/Kovalskyy, SDSU/ Boschetti,

U. Idaho

South Dakota State U.

International collaborators:

Chuvieco, Spain; Tansey, UK

Prototyping a Landsat-8 Sentinel-2 global burned area product

Hansen: Global Crops Product

Prototype (Type and Area)/Potapov,

UMD

University of Maryland

International collaborators:

Defourny, Belgium

Integrating Landsat 7, 8 and Sentinel 2 data in improving crop type

identification and area estimation

Townshend: Global Tree cover and

water bodies product

prototype/Sexton, UMD / Feng, UMD/

Channan, UMD

University of Maryland

International Collaborators:

Schmullius, Germany

Multi-source imaging of time-serial tree and water cover at

continental to global scales

NASA LCLUC Multi-Source Land Imaging Projects: Mid-term

MuSLI Science Team: 40+ members

Salas , Applied Geosolutions

Torbick, AG

Lang, U Maryland

Jones, USGS

Huang, UMD (new PI)

Small, Columbia U.

Nghiem, JPL

Greg Yetman, Columbia U.

Friedl, Boston U.

Gray, BU

Melaas, BU

Roy, South Dakota State U.

Kovalskyy, SDSU

Boschetti, U. Idaho

Hansen, U. Maryland

Potapov, UMD

Townshend, U. Maryland

Sexton, UMD

Feng, UMD

Channan, UMD

Hoekman, Wageningen U.

Le Toan, CESBIO

Creed, Western U., Canada

Esch, DLR

Eklundh, Sweden

Chuvieco, Spain

Tansey, UK

Defourny, Belgium

Schmullius, Germany

Masek, NASA, MSLI Project

Scientist

Markham, NASA, calibration team

Helder, SDSU

Czapla-Myers (U. Az)

Schott, RIT

DIRSIG model, LST

Vermote, NASA GSFC

Atm. Corr. Team Claverie, U. MD

Woodcock, Boston U.,

clouds (USPI +1)

Dungan, NASA Ames, NEX

Ganguly, NASA Ames, NEX

PI and CO-Is Int. Collaborators Landcover Project Science Office

Koetz, ESA,

Sentinel-2 Projects

Coordinator

Dedieu & Hagolle,

CESBIO

Operational algorithms and products for near real time maps of rice extent and rice crop growth stage using multi-source remote

sensing PI: W. Salas, N. Torbick, AGS

Thuy Le Toan CESBIO, Dirk Hoekman, Wageningen

• Fuse SAR-optical for mapping agricultural conditions; technology transfer with partners

• S. Asia • Sentinel-1, PALSAR-2, Landsat-8,

Sentinel-2, Radarsat-2 • Fusion provides more and better

information; the approach strives to extract strengths of any / each satellite to complement another platform

• Pilot sites have beta products; begun cal / val coordination with ESA, regional partners (AsiaRice, IRRI, VAST, SERVIR, Ministries,…)

Example Tonle Sap, Cambodia Sentinel-1A Rice Inundation Dynamics Time Series

• Goals:

– Develop automated inundation mapping algorithms for Landsat, Sentinel-2 and Sentinel-1

– Generate near daily inundation products over US and southern Canada

• Progress:

– Algorithm for the automated quantification sub-pixel water fraction (SWF) developed for optical data streams (Landsat-5/7/8 and Sentinel-2) and tested over several sites

– Algorithm for automated classification of water for Sentinel-1 developed and tested over several sites

• Benefits of MuSLI inundation approaches:

– Provide needed spatial-temporal details: small water bodies, areas inundated for short periods, rapid inundation changes

– Enable advances in understanding aquatic systems: connectivity, function, carbon, and biodiversity

Towards Near Daily Monitoring of Inundated Areas over North America through Multi-Source Fusion of Optical and Radar Data

Chengquan Huang (PI), Ben DeVries, Wenli Huang, University of Maryland Megan Lang (Co-I), US Fish and Wildlife Service National Wetland Inventory; John Jones (Co-I), US Geological Survey

Irena Creed (International Collaborator), University of Western Ontario, London, ON, Canada

Time series sub-pixel water fraction (SWF) maps derived using Sentinel-1 (S1), Sentinel-2A (S2) and Landsat-8 (L8) imagery over the Everglades

Inundation change within above area

5 km

Multi-source imaging of infrastructure and urban growth using Landsat, Sentinel and SRTM

C. Small - Columbia Univ. USA S. Nghiem – NASA/JPL USA T. Esch – DLR Germany

• Develop continuous index to map built/impervious land cover

• Map changes in settlement extent between 2000 and 2015

• Global Extent – 20 urban-rural gradients in different biomes

• Optical: Landsat + Sentinel 2

• Radar: SRTM + Sentinel 1

• Multi-season – Spectral Stability

• Multi-source – Reflectance + Backscatter

• Multi-decade – Robust WRT sensor & time

• Developed spectral stability index

• Now characterizing reflectance-backscatter relations across biome

Stable Reflectance = Impervious

surfaces

High Backscatter = Corner reflectors

Both together = High Infrastructure Density

Multisource Imaging of Seasonal Dynamics in Land Surface Phenology: A Fusion Approach Using Landsat and Sentinel-2

Mark Friedl, PI; Josh Gray, Co-I; Boston University Lars Eklundh, Lund University, Patrick Hostert, Humboldt University, International Collaborators

• Goals:

– To quantify the timing and magnitude of land surface phenology events

(“phenometrics”) at moderate spatial resolution, and

– To generate gap-filled time series of spectral vegetation indices that

characterize the entire seasonal cycle of land surface phenology at fixed

time steps.

• Geographic area: Global (but focused on study sites)

• Data used: Landsat 8, S2 ( supplemented by Landsat 5, 7 & MODIS)

• Advantage of MuSLI approach: Time series density

• Up-to-date progress:

– Ongoing international collaboration with Lund University and Humboldt

University, Berlin (BU Team visiting Berlin & Lund in Nov., 2016)

– Manuscript in prep., AGU presentation, analyzing phenology algorithms

– Data set development focused on Cal/Val data for core test sites.

– Manuscript in prep describing Kalman Filter algorithm to fuse OLI and MSI.

– Analysis of phenology algorithms based on HLS data (see figure at right)

25

45

65

85SOS (days since Oct. 8, 2015)

0 5 10 15km

Above: Estimated day of year for SOS (start of season) based on HLS time series for scene near Lahore, Pakistan. Below: sample HLS time series for a multi-cropped pixel; estimated phenology dates identified by vertical lines.

David Roy, Haiyan Huang, Lin Yan, Hankui Zhang, Jian Li, (GSCE, South Dakota State University, USA, Luigi Boschetti (Idaho, USA) International Collaborators: Jose Gómez-Dans & Philip Lewis (UCL, London, U.K), Emilio Chuvieco (Alcala, Spain), Kevin Tansey (Leicester, U.K)

• Goal: Prototype global 30m burned area product to meet user needs for – improved carbon budget accounting – greenhouse gas and aerosols emissions – environmental management – post-fire assessment and remediation – people - environment - climate - fire research

• Geographic area: Africa + global sample

• Data used: Landsat-8 and Sentinel-2A/B

• MuSLI advantage:

– Landsat-8 and Sentinel-2A/B together provide needed temporal resolution for time series burn change detection

– Landsat-8 has improved quantization, signal/noise characteristics, and acquisition coverage over heritage Landsat missions

– Sentinel-2 has Landsat-8 like bands at 10m & 20m with higher acquisition coverage than Landsat

• Progress: Automated sensor-agnostic burned area algorithm for WELD processed time series prototyped

• Progress: Sentinel-2 processing under global WELD processing, and Landsat-8 to Sentinel-2 registration, developed and implemented

Landsat-8

global WELD tile

hh19vv12h3v2

Cape Town

Sentinel-2

global WELD tile

hh19vv12h3v2

Cape Town

• Progress: Four publications

1. Huang, H., Roy, D.P., Boschetti, B., Zhang, H.K., Yan, L., Kumar, S.S., Gomez-Dans, J., Li, J., 2016, Separability analysis of Sentinel-2A multi-spectral instrument (MSI) data for burned area discrimination, Remote Sensing, In Press.

2. Storey, J., Roy, D.P., Masek, J., Gascon, F., Dwyer, J., Choate, M., 2016, A note on the temporary mis-registration of Landsat-8 Operational Land Imager (OLI) and Sentinel-2 Multi Spectral Instrument (MSI) imagery, Remote Sensing of Environment, 186, 121-122.

3. Roy, D.P., Li, J., Zhang, H.K., Yan, L., 2016, Best practices for the reprojection and resampling of Sentinel-2 Multi Spectral Instrument Level 1C data, Remote Sensing Letters, 7(11), 1023-1032.

4. Yan, L., Roy, D.P., Zhang, H.K., Li, J., Huang, H., 2016, An automated approach for sub-pixel registration of Landsat-8 Operational Land Imager (OLI) and Sentinel-2 Multi Spectral Instrument (MSI) imagery, Remote Sensing, 8(6), 520.

Date 1

Date 2 Mapped burned

Prototyping a Landsat-8 Sentinel-2 Global Burned Area Product

South America 2015-2016 growing season soybean cover

Integrating Landsat 7, 8 and Sentinel 2 data in improving crop type identification and area estimation

PIs: M. Hansen, P. Potapov, University of Maryland International collaborators: Pierre Defourny, UCL and Carlos Di Bella, INTA

Goals • Develop and implement a system for defining

the required phenological sampling for mapping crop types.

• Determine if Landsat and/or Sentinel 2 acquisitions are sufficient in meeting required sampling frequencies for selected commodity crop type mapping (required best individual date image inputs).

• Compare the performance of single date images and seasonal metrics in mapping crop type at local (per sample block) and regional (all sample blocks at once) scales.

• Given sufficient temporal richness, validate area estimation of crop type for large scale industrial and fine scale smallholder cropping systems.

Scale - National

Data - Primarily Landsat 7 and 8 and Sentinel 2A

Advantage - Crops require more detailed phenologic profiles for characterization

Progress – Testing use of data in Tanzania for corn mapping, waiting for systematic acquisitions over large commodity crop landscapes

Multi-source imaging of time-serial tree and water cover at continental to global scales

PI: John Townshend, Global Land Cover Facility, Department of Geographical Sciences, University of Maryland Co-Is: Joseph O. Sexton, Min Feng, Saurabh Channan, Global Land Cover Facility, Department of Geographical

Sciences, University of Maryland International collaborator: Christiane Schmullius, University of Jena, ESA GLOBBIOMASS Project

• Goal: Develop methodologies for fusing optical and radar imagery into data streams of tree- and water-cover estimates

• Geographic area: Global • Data used: Landsat TM, ETM+ & OLI; Sentinel

1 & 2; PALSAR • Advantage: long-term and globally consistent

estimates without gaps due to clouds or sensor failure

• Progress: – Global estimates of tree cover based on Landsat

(2010) – Development of online portal for collaborative

data visualization and progress-reporting – Successful first test of incorporating Sentinel-2:

filled cloud gaps – Unsuccessful first test of incorporating Sentinel-1:

weak tree-cover signal

Global, Landsat-based tree-cover estimates for the 2010 GLS epoch visualized through an online portal for collaboration based on shared geospatial datasets.

Outstanding Issues

• Higher level products depend on preprocessing – Calibration

– Cloud detection

– Registration

– Atmospheric correction

• Errors in TOA reflectance due to cloud/cloud shadow contamination propagate to higher level products

• Mis-registration is an obstacle in many applications which use pixel-to-pixel colocation

Towards Higher Level Products

From Landsat and Sentinel Data

• Sensor reflectance products – assumed to be available to MuSLI PI’s – Are the L8 and S2 surface reflectance products and/or algorithms

available?

– Object: a seamless, sensor-independent reflectance record at high spatial, and high-temporal resolution to support land science

– Compatible atmospheric correction & cloud/shadow detection

– Corrections for BRDF (solar, view angle), band pass differences

– Adjustment to common frame, resolution (~30m), compositing period (~5 day) • Absolute geolocation of MSI L1C products may differ from Landsat-8 L1T products

by up to ~37m (2s), due to errors in the Landsat GLS2000 reference

• What is the status for each algorithm as compared to what’s done in Europe?

• 39 tests sites (22 from MusLI ST) were selected – Time series will be constructed

• 368 S2 tile and 360 Landsat path/row

17

Coordination and collaboration of MuSLI and Landsat Science Teams

• New call for MuSLI is coming out in Feb (NOI by Apr 1; Proposals by June 1 – one step only, no social science required)

• MuSLI team members will participate in at least one of Landsat ST meetings during the year

• MuSLI team members who are not on the Landsat team will be encouraged to share their results with Landsat team members

• Landsat team members will be invited to international MuSLI team meetings

Thank You