phenology land product validation workshop lpv phenology subgroup; status following on the dublin...

27
Phenology Land Product Validation Workshop LPV Phenology Subgroup; Status Following on the Dublin meeting 2010, aim is to develop a plan on how to effectively use ground- to airborne-level phenological measurements to validate satellite-based land surface phenology products internationally-coordinated remote sensing land surface phenology validation and inter-comparison activity Around 80 members in the mailing list from different parts of the world

Upload: william-hill

Post on 13-Jan-2016

215 views

Category:

Documents


0 download

TRANSCRIPT

Page 1: Phenology Land Product Validation Workshop LPV Phenology Subgroup; Status Following on the Dublin meeting 2010, aim is to develop a plan on how to effectively

Phenology Land Product Validation WorkshopLPV Phenology Subgroup; Status

• Following on the Dublin meeting 2010, aim is to• develop a plan on how to effectively use ground- to airborne-level

phenological measurements to validate satellite-based land surface phenology products

• internationally-coordinated remote sensing land surface phenology validation and inter-comparison activity

• Around 80 members in the mailing list from different parts of the world

Page 2: Phenology Land Product Validation Workshop LPV Phenology Subgroup; Status Following on the Dublin meeting 2010, aim is to develop a plan on how to effectively

Phenology Land Product Validation WorkshopLPV Phenology Subgroup; Status

• What terminology we should use?

• Email was sent to the list server for input into the AGU poster

• Led on to the Semantics of Phenology

• Different scale, processes, sensors..

• Leaf phenology, Vegetation/ canopy phenology, Lands surface phenology/ seasonality, Landscape phenology …

• Overall the agreement was on ‘Land surface phenology’

• http://vip.arizona.edu/VIP_LSP_Semantics.php

Land surface phenology refers to the type of products that seek to quantify and summarize the dynamics of the vegetated land surface at temporal scales from annual to seasonal. Products should clearly mention about the sensor/ method ……

Page 3: Phenology Land Product Validation Workshop LPV Phenology Subgroup; Status Following on the Dublin meeting 2010, aim is to develop a plan on how to effectively

Phenology Land Product Validation WorkshopLPV Phenology Subgroup; Status

• In the context of LPV validation

Stage 1 Validation

Product accuracy is assessed from a small (typically < 30) set of locations and time periods by

comparison with in-situ or other suitable reference data.

Product Vs Ground

Stage 2 Validation

significant set of locations and time periods by comparison with reference in situ or other suitable

reference data.Spatial and temporal consistency of the product

and with similar products has been evaluated over globally representative locations and time

periods.

Product vs Product vs (more) Ground

Stage 3 Validation

Uncertainties are characterized in a statistically robust ……….

more(product) vs (more) ground

Stage 4 Validation

systematically updated ………

Page 4: Phenology Land Product Validation Workshop LPV Phenology Subgroup; Status Following on the Dublin meeting 2010, aim is to develop a plan on how to effectively

Phenology Land Product Validation WorkshopLPV Phenology Subgroup; Status

• Normal LPV activity

• Most often algorithm development groups collect field data

• Strength for us (phenology)

• Two groups ( ground data & Satellite data)

• Challenge for us (phenology)

• How to combine?

Page 5: Phenology Land Product Validation Workshop LPV Phenology Subgroup; Status Following on the Dublin meeting 2010, aim is to develop a plan on how to effectively

Phenology Land Product Validation Workshop

Agenda and Objectives

Review of available data:•Remote Sensing Phenology Products•Camera & In situ Networks•Ground/Citizen Measurements & Networks•Core Site Selection based on data availabilityPanel led Discussion:•Satellite and in situ data scaling issues, utility of citizen science for product assessment, address major questions and concerns…Define Pilot Projects:•Review of Sites - Preparation and Distribution of Data Bundles•Structure and Timeline of Projects•Responsible Parties Workshop Review:•Did we meet the workshop objectives?•Publication of meeting results.•Schedule an informal meeting at AGU 2012 for status update on Data Distribution and Pilot Projects?

Page 6: Phenology Land Product Validation Workshop LPV Phenology Subgroup; Status Following on the Dublin meeting 2010, aim is to develop a plan on how to effectively

Remote Sensing data product

Contribution from data providers/algorithm development team

Page 7: Phenology Land Product Validation Workshop LPV Phenology Subgroup; Status Following on the Dublin meeting 2010, aim is to develop a plan on how to effectively

Data from many moderate resolution remote sensing sensor, mainly vegetation indices at a compositing period

We broadly follow three steps to derive phenological matrices• Data filtering• Temporal smoothing (many methods)• Derived matrices ( many method and many matrices)

JÖNSSON and EKLUNDH, 2004

Page 8: Phenology Land Product Validation Workshop LPV Phenology Subgroup; Status Following on the Dublin meeting 2010, aim is to develop a plan on how to effectively

MODIS NACP Phenology Products

Retrieved Phenology Metrics1. Beginning of season2. End of season3. Length of season4. Base VI value5. Peak time 6. Peak value7. Amplitude8. Left derivative9. Right derivative10. Integral over season

- absolute11. Integral over season

- scaled12. Maximum value13. Minimum value14. Mean value15. RMSE of fitting

Page 9: Phenology Land Product Validation Workshop LPV Phenology Subgroup; Status Following on the Dublin meeting 2010, aim is to develop a plan on how to effectively

MODIS NACP Phenology Products

Availability and Status

Availability: From http://accweb.nascom.nasa.gov/

Products: phenology metrics derived from LAI/EVI/NDVI, and

original, smooth/gap-filled LAI, FPAR, EVI & NDVI.

Temporal Coverage: From 2001 to 2010.

Spatial Coverage: Full North America, partially South America.

Asia is under processing.

Online data services: Subset by geographic area Subset by data layer Reproject Mosaic Aggregation Re-format (to GeoTIFF).

Page 10: Phenology Land Product Validation Workshop LPV Phenology Subgroup; Status Following on the Dublin meeting 2010, aim is to develop a plan on how to effectively

MCD12Q2 C5 Product

• Global database– Annual since 2001, 500-m

• Includes 7 metrics– Onset of EVI increase

– Onset of EVI maximum

– Onset of EVI decrease

– Onset of EVI mimimum

– Min EVI

– Max EVI

– Sum of growing season EVI

• Validation:

– Opportunistic, largely in New England

– Current focus on PhenoCam Data

Timing

AnnualMetrics

Mark Friedl

Page 11: Phenology Land Product Validation Workshop LPV Phenology Subgroup; Status Following on the Dublin meeting 2010, aim is to develop a plan on how to effectively

USGS EROS Vegetation Dynamics

Availability: From http://phenology.cr.usgs.gov//

Products: Nine annual remote sensing phenological indicators

(served as raster data sets) are available at two spatial resolutions

(1000 m2 and 250 m2) based on NDVI

Temporal Coverage: AVHRR (1989-2011)

MODIS (2001-2011)

Spatial Coverage: conterminous U.S.

Method : Delayed Moving Average (DMA) method (Reed et al.,

1994).

Considerable QA checking done on USGS phenological data

Jesslyn Brown

Page 12: Phenology Land Product Validation Workshop LPV Phenology Subgroup; Status Following on the Dublin meeting 2010, aim is to develop a plan on how to effectively

Phenological metrics available at multiple resolutions

Jesslyn Brown

Page 13: Phenology Land Product Validation Workshop LPV Phenology Subgroup; Status Following on the Dublin meeting 2010, aim is to develop a plan on how to effectively

Phenological metrics available at multiple resolutions

Jesslyn Brown

Page 14: Phenology Land Product Validation Workshop LPV Phenology Subgroup; Status Following on the Dublin meeting 2010, aim is to develop a plan on how to effectively

Jesslyn Brown

Page 15: Phenology Land Product Validation Workshop LPV Phenology Subgroup; Status Following on the Dublin meeting 2010, aim is to develop a plan on how to effectively

The “VGT4Africa” phenology product• Algorithm developed by the Joint Research Centre (European

Commission)• Product generated by VITO (Belgium)• Based on the processing of a moving time-window of 1.5 year

of NDVI from the VEGETATION instrument• Updated within 3 days after every 10-day period (“dekad”)• Covers the whole African continent• Provides dekad dates for “start of growth”, “max NDVI” and

“half-senescence”• Availability: from VITO through ftp and EUMETCast, jan 2007

until present• Product description: Combal B. & Bartholomé E. 2006: Phenology. In: Bartholomé

edit: VGT4Africa user manual 1st edition, European Commission ref EUR 22344 EN: 165-212

• Method: Combal B. & Bartholomé E. 2010: Retrieving phenological stages from low resolution Earth observation data. In: Maselli & al.: Remote Sensing Optical Observations of Vegetation Properties, Research Signpost, Kerala, India, 115-129.

Bartholomé

Page 16: Phenology Land Product Validation Workshop LPV Phenology Subgroup; Status Following on the Dublin meeting 2010, aim is to develop a plan on how to effectively

Bartholomé

Start dates as observed on 3rd dekad of Dec 2011(note: actual time resolution of the product is the dekad, not the month)

Page 17: Phenology Land Product Validation Workshop LPV Phenology Subgroup; Status Following on the Dublin meeting 2010, aim is to develop a plan on how to effectively

VIP Data Explorer:30 Years of Multi-Sensor VI and Phenology Data

Availability: From vip.arizona.edu/viplab_data_explorer.php

Products: Vegetation index and phenology from AVHRR,

VEGETATION, MODIS (Sensor independent)

Temporal Coverage: 30+

Spatial Coverage: Global

Spatial resolution : 0.05 deg

Considerable data quality assessment

Kamel Didan

Page 18: Phenology Land Product Validation Workshop LPV Phenology Subgroup; Status Following on the Dublin meeting 2010, aim is to develop a plan on how to effectively

Kamel Didan

Page 19: Phenology Land Product Validation Workshop LPV Phenology Subgroup; Status Following on the Dublin meeting 2010, aim is to develop a plan on how to effectively

PHAVEOS – the Phenology And Vegetation EO Service

A service to provide:Vegetation maps of several biophysical variables relevant

to models of bio-geochemical cycles Leaf Area Index (LAI) fraction of Absorbed Photosynthetically Active Radiation (fAPAR)

MERIS Terrestrial Chlorophyll Index (MTCI) fraction of green land cover (fCover)

Continuous time series to support phenology studies and monitoring

Visualisation of individual maps and phenology curves for individual locations

Thomas Lankester

Page 20: Phenology Land Product Validation Workshop LPV Phenology Subgroup; Status Following on the Dublin meeting 2010, aim is to develop a plan on how to effectively

MERIS / MODIS

Sentinel 3 Sentinel 2 (LDCM)

Biophysical processing and mapping

Data sources

HiProGen and Overland

Daily Level 3 and Level 4 data dissemination

WebServerWeb client on user PC

Page 21: Phenology Land Product Validation Workshop LPV Phenology Subgroup; Status Following on the Dublin meeting 2010, aim is to develop a plan on how to effectively

Level 3 daily product examples

fCover LAI fAPAR

Page 22: Phenology Land Product Validation Workshop LPV Phenology Subgroup; Status Following on the Dublin meeting 2010, aim is to develop a plan on how to effectively

ftp://l3-server.infoterra.co.uk/pub/SNL/MTCI_L4_2009-2010_comparison.gif

Spring 2009 – 2010 comparison

Page 23: Phenology Land Product Validation Workshop LPV Phenology Subgroup; Status Following on the Dublin meeting 2010, aim is to develop a plan on how to effectively
Page 24: Phenology Land Product Validation Workshop LPV Phenology Subgroup; Status Following on the Dublin meeting 2010, aim is to develop a plan on how to effectively

Phenology Land Product Validation WorkshopCore Site Selection

Original Sites (2010 Dublin Workshop):• Do we keep the original sites?• Are more sites needed?• What are the essential variables and is it

necessary for every site to offer the same set of core variables/instruments?

Site Name Country Cover Type Lat Lon MET FLUX Pheno CameraRadiometer

PAR Pheno ObservationsTorgnon – Tellinod Italy Grassland 45.82 7.56 X X X X XTorgnon – Tronchaney Italy Larch Forest 45.82 7.56 X X X X XPark Falls USA Deciduous Broadleaf 45.95 -90.27 X X XHyytiala Finland Boreal Conifer 61.85 24.29 X X X X XHarvard USA Mixed Forest 42.54 -72.17 X X X XBartlett USA Mixed Forest 44.06 -71.29 X X XHowland USA Boreal Hardwood Trans 45.2 -68.74 X X XTakayama Japan Deciduous Broadleaf 36.15 137.42 X X X XTakayama Japan Evergreen Coniferous 36.14 137.37 X X X XBarrax Spain Cropland 39.05 -2.09Hubbard Brook USA Deciduous Hardwood 43.93 X X X XVaira Ranch USA Grassland 38.41 -120.95 X X X X…other suggestions? particularly Asian or Southern Hemisphere locations.

Page 25: Phenology Land Product Validation Workshop LPV Phenology Subgroup; Status Following on the Dublin meeting 2010, aim is to develop a plan on how to effectively

Phenology Land Product Validation Workshop

Panel Discussion

Working across scales:•Are site specific nested datasets (in-situ, phenocam, RS) and validation results applicable to validation of continental/global RS phenology products?• Do PhenoCams need to be validated with in-situ observations?

What standards need to be set for Phenology LPV:•Are standardized definitions needed for metrics? – Start of Season, End of Season•Are standardized methods needed to calculate metrics? – Curve fitting, Derivative peaks, etc. •What do we mean by Phenology Validation? Is it setting a realistic offset/error range between phenocam or in-situ and RS metrics? Is this application specific?•What are best practices for LPV using in-situ data?

Page 26: Phenology Land Product Validation Workshop LPV Phenology Subgroup; Status Following on the Dublin meeting 2010, aim is to develop a plan on how to effectively

Phenology Land Product Validation Workshop

Pilot Project Definition

Core Sites Selection and Considerations:•Do we agree upon the site selections?•Is all data freely available? Creation of formal data sharing agreement.

Data Collections/Bundles:•RS products – size of subset over each site, 100km?•Centralized Storage and Access•Ground/In Situ Site Data – centralized storage?Project Objectives:•Do we allow for a flexible structure and let researchers dictate site by site analysis OR do all projects follow a set protocol?•Timeline – What is a realistic expectation? The LPV 5yr Plan states Validation Protocol established by 2013. Responsible Parties:•Data Collections/Bundles – must be available by…?•Who will conduct the research? PhD Students, Post-Docs, Staff Scientists.

Page 27: Phenology Land Product Validation Workshop LPV Phenology Subgroup; Status Following on the Dublin meeting 2010, aim is to develop a plan on how to effectively

Phenology Land Product Validation Workshop

Workshop Review

Did we meet our objectives?Provide a synopisis of the majority available data sets.Review and discuss validation methods, current limitations and concerns.Selection of Core Sites.Agreement on data subsets, storage and access.Define Pilot Projects.Set a course for future Land Surface Phenology Validation

For the future:Do responsible parties understand their tasks (providing data, analysis, etc.)Write up of a Meeting Summary Publication – EOS.Summary Poster for AGU – Jadu and Matt with input from committee.Informal Meeting at AGU 2012 to discuss progress.