Assessing VIIRS Land Biophysical EDRs

1
 Assessing VIIRS Land Biophysical EDRs 

• Jeffrey L. Privette
• Nov 4-6, 2003 

2
Team Members and Roles

• Boston University
• Ranga B. Myneni, Co-I and Boston Lead
• Yuri Knyazikhin, Co-I
• Crystal B. Schaaf, Co-I
• University of Arizona
• Alfredo R. Huete, Co-I and Arizona Lead
• Derrick Lampkin, PhD Student
• NASAS Goddard Space Flight Center
• Jeffrey L. Privette, PI and NASA Lead
• Mohan Nirala, UMBC Research Scientist

3
Primary EDR Interests

• Vegetation Index
• Observed Top-of-Atmosphere NDVI
• Observed Top-of-Canopy EVI
• Surface Albedo
• Daily
• Instantaneous Blue Sky
• Two algorithms (multi-regression, BRDF magnitude
  inversion)
• Land Surface Temperature
• Dual Split Window

4
Primary EDR Interests

• Vegetation Index
• Observed Top-of-Atmosphere NDVI MODIS
  Nadir-Normalized TOC NDVI
• Observed Top-of-Canopy EVI MODIS Normalized TOC
  EVI
• Surface Albedo
• Daily MODIS 16-day
• Instantaneous Blue Sky MODIS Black and
  White Sky
• Two algorithms (multi-regression, BRDF magnitude
  inversion)
• MODIS main algorithm full BRDF inversion
• Land Surface Temperature
• Dual Split Window MODIS Day/Night and Single
  Split Window

5
Associated Interests

• Geolocation IP
• Accuracy impacts multi-observation algorithms
  ancillary data
• 200 m, 3 sigma target not likely to be met
  (-NGST, July 2003)
• Collaboration with Wolfe, Schaaf, Ranson,
  Loveland
• Surface Reflectance IP
• Insufficient performance for EVI (-NGST, August
  2003)
• Collaboration with Lyapustin and Vermote
• The Products Formerly Known As VVI2P
• LAI and FPAR
• Internal expertise

6
EDR as CDRAnalysis Methods
?
7
Develop Proxy Data Transformations
Spectral - spatial transfer functions to
optimally relate MODIS and ASTER/ETM
characteristics to the VIIRS characteristics
Spectral Spatial Convolutions
A. Huete
8
Operational Generation of EDRs from Proxy-VIIRS
Data
9
Model-based Estimations of EDR Uncertainty
e( EDR ) fcn( e(algorithm model), e(
algorithm input ) ) Accuracy in the retrievals
can not exceed the summed accuracy of the data
and model

• VIIRS Model uncertainty zero for VI, but
  non-zero for albedo and LST
• Analysis performed following the MODIS LAI/FPAR
  algorithm (figures above)

R. Myneni
10
Deriving Further Biophysical Products (LAI, FPAR)
From VIIRS VI
MODIS LAI/FPAR main algorithmingests surface
reflectances
MODIS LAI/FPAR back-up algorithm ingests NDVI
Main algorithm
Back-up algorithm

• Adapt algorithms to ingest proxy-VIIRS surface
  reflectances and NDVI, respectively
• Characterize differences betweens the respective
  LAI and FPAR results
• VIIRS albedo also can generated internally by the
  main algorithm

R. Myneni
11
Sensitivity of VIIRS NDVI to Vegetation Dynamics
Resulting from Climate Change

• Identify areas significantly affected by climate
  change by determining the persistence index in
  an AVHRR NDVI time series (1981-2005) over
  northern latitudes
• Characterize signals of change in these areas,
  using MODIS data (2000-2005) as a VIIRS data
  proxy
• Estimate VIIRS VI sensitivity by evaluating the
  change signatures with their associated noise
  levels for different degrees of persistence index
• persistence index determines where NDVI
  increased consistently, as opposed to NDVI
  trending

R. Myneni
12
VI - LST Coherence and Synergy
VIIRS
MODIS
Ts
Tsoil max
Tsoil
?T
TvegTsoil min
Ta
VI
VImax
VImin
VI
?T ? Ts - Ta
VI-Ts diagram (Nemani Running, 1989 1993
Nishida 2003)
(Moran 1994)
A. Huete
13
Resources Offered, Requested

• Offer Radiometric Transformations and proxy data
  sets
• Offer/Request Implementation of VIIRS prototypes
  in MODAPS Subsetting Stream
• 26 200km x 200km Core Sites
• 274 7km x 7km ORNL ASCII Sites
• Request npp1 (GSFC) with MODIS feed

14
Contribution to Deliverables

• Algorithm Analysis Reports
• Lead on VI and LST, support of Albedo
• Cal/Val Plan
• Land-based field methods
• Lessons learned from EOS, SAFARI 2000, etc.
• Alignment with CEOS WGCV Land Product Validation
  (LPV) Subgroup
• CEOS Core Sites
• Standards, international cost-sharing
• Operations Concept
• Algorithm technology insertion to IDPS
• Cal/val data processing needs (e.g., subsetting)
• Periodic VIIRS diagnostic mode over some Core
  Sites