History of the NDVI and Vegetation Indices by historyman

VIEWS: 3,523 PAGES: 60

									History of the NDVI
   & Vegetation
      Indices
  Compton Tucker
 NASA/UMD/CCSPO
     What are the NDVI, EVI, etc.
     and why do we use them?
VIs are estimates of the visible light absorbed by plant
canopies --the photosynthetic capacity (Sellers 1985 & 1987,
Myneni et al. „95)

         This energy drives photosynthesis.
Thus S NDVI over time = ~ GPP
The NDVI is not “greenness”! -- what does this mean
anyway?; and is not biomass (biological mass)!
The NDVI does not “saturate” any more than photosynthesis
“saturates”. Both are limited by a lack of photons in the red.
The EVI is not as highly correlated to APAR as the NDVI
because the EVI is more weighted to the near infrared.
Vegetation Indices from Susan
     Index         Formula                                 Details                        Citation


Ustin
  Simple Ratio               R NIR
                             RR
                                                           Gre en vegetation cover.
                                                           Various wavelengths,
                                                           depending on sensor. (e.g.
                                                           NIR = 845nm, R=665nm)
                                                                                          Pearson, 1972




  Normalized          RNIR  RR                            Gre en vegetation cover.
                                                           Various wavelengths,
                                                                                           Tucker 1979
  Difference
Vegetation Index      RNIR  RR                            depending on sensor. (e.g.
                                                           NIR = 845nm, R=665nm)

   Enhanced                                                C1 =6; C2=7; L=1; G=2,5
Vegetation Index                                                                            Huete 1997




                     Rs Rv2 (NIRs NIRv2
 Perpendicular                                             Perpendicular distance from      Richardson
Vegetation Index                          )                the pixels to the soil line.    and Wiegand
                                                                                               1977

 Soil Adjusted          NIR R                             L = soil adjusted factor         Huete 1988
Vegetation Index                    L
                                  1
 Modified Soil         NIR R  L                          L = (1-2a x(NIR-aR) x NDVI)    Qi et al 1994
   Adjusted                                                Self adjusting L:f on to
                                                                     or
                                                           optimize f soil effects.
Vegetation Index                                           Higher dynamic range.

Transformed Soil               a NIR  aR  b            a=slope of soil line
                                                           b=intercept of soil line
                                                                                            Baret and
                                                                                            Guyot 1991
                       R  a ( NIR  b)  0.08(1  a 2 )
   Adjusted
Vegetation Index


                                NIR R      
    Soil and                                               More independent of surface      Huete et al

                      2.5  
Atmospherically                                            brightness                         1997
   Resistant
Vegetation Index            1 NIR 6R  7.B
                                             
   Who
 invented
the NDVI?
      Spectral Vegetation Indices
Birth and McVey 1968 “Measuring color of growing turf with a
reflectance spectrometer” Agronomy Journal 60(6):640-645.
Jordan 1969 “Derivation of leaf area index from quality of light on the
forest floor” Ecology 50(4):1271-1318.

Pearson and Miller 1972. “Remote mapping of standing crop biomass
for
estimation of the productivity of the shortgrass prairie. Proc. 8th Intl.
Symp.
Remote Sens. Environ., Univ. Michigan, pp. 1357-1381.

Rouse et al. 1974 “Monitoring vegetation systems in the Great Plains
with ERTS. Proc. 3rd ERTS-1 Symp. pp. 301-319. [NO LITERATURE
CITED, a few paragraphs on vegetation indices, and they used the TVI
not the NDVI!].

Tucker 1977 “Use of Near Infrared/Red Radiance Ratios for Estimating
Vegetation Biomass and Physiological Status”. X-923-109, Greenbelt,
NASA Goddard Space Flight Center 41 p. (preprint of Tucker 1979 RSE).
     Spectral Vegetation Indices
Birth and McVey 1968 “Measuring color of growing turf with a
reflectance spectrometer” Agronomy Journal 60(6):640-645. Other
papers:
       Birth, G. S. 1977 “Optical properties of blue cheese as affected
by             aging” J. Dairy Sci. 60:57-58.
      Birth, Davis, and Townsend 1976. “Scatter coefficient as a
measure of    pork quality” J. Animal Sci. 43:238-239.

Jordan 1969: 4 data points and no other publications on topic. Many
other publications on a range of tropical ecology topics though.

Rouse et al. and Deering et al. -- no subsequent journal articles on
NDVI (Verba volent, scripta manet), used the Transformed Vegetation
Index. TVI= SQRT(NDVI+0.5). Rouse et al. 1974: “However, previous
studies have shown that the combinations of these four parameters
can provide more useful parameters for specific comparisons.” What
previous studies? --

Japanese researcher in the 1960s working in photosynthesis published
     Spectral Vegetation Indices
>2,800 articles in referred literature @ Web of
Science*#
>46,000 citations of these articles*
Wide use in many earth science disciplines
Greatest use of any remote sensing measurement
Greatest use by many times of data from “AVHRR”
*as of June 1, 2006
#using criteria “vegetation index”, “normalized difference
vegetation index”, “NDVI”, & papers that cite Tucker 1979 or
Richardson and Wiegand 1977
Top 20 VI Article Journals, from
a total of 386 Journals, from ISI
 Rank                        Name                          Cited    Pubs
  1     Remote Sensing of Environment                      13,426   499
  2     International Journal of Remote Sensing            12,403   741
  3     Science                                            3,423    27
  4     IEEE Transactions on Geoscience & Remote Sensing   2,946    111
  5     Journal of Geophysical Research-Atmospheres        2,930    143
  6     Nature                                             2,838    21
  7     Photogrammetric Engineering and Remote Sensing     2,577    144
  8     Global Biogeochemical Cycles                       1,588    40
  9     Journal of Clim ate                                1,160    32
 10     Ecological Applications                            1,138    34
 11     Global Change Biology                              1,064    66
 12     Agricultural and Forest Meteorology                 926     40
 13     Clim atic Change                                    813     33
 14     Journal of Applied Meteorology                      600     24
 15     Agronomy Journal                                    427     35
 16     Geophysical Research Letters                        341     44
 17     Journal of Arid Environments                        204     32
 18     Transactions of the Am. Soc. Agricultural Eng.      196     29
 19     Ecological Modelling                                187     26
 20     Global Ecology and Biogeography                     125     23
           Top 10 cited VI papers
rank                         Published work                      citations
 1     Tucker 1979, Remote Sens. Environ. 8:127-150                780

 2     Schlesinger et al. 1990, Science 247:1043-1048              559

 3     Turner 1989, Annual Review Ecology and 20:171-197           558

 4     Myneni et al. 1997, Nature 386:698-702                      525

 5     Noilhan J. 1989, Monthly Weather Review 117:536-549         480

 6     Holben B.N., 1986, Intl. J.Remote Sens. 7:1417-1434         476

 7     Tucker et al. 1985, Science 227:369-375                     401

 8     Justice et al. 1985, Intl. J.Remote Sens. 6:1271-1318       384

 9     Walther et al. 2002, Nature 416:389-395                     384

10     Potter et al. 1993, Global Biogeochem. Cycles 7:811-841     375
HistCite




           From Web of
           Science
      11-20 VI most cited papers
11 Sellers P.J. 1996, Journal Climate 9:676-706                353

12 Huete A.R. 1988, Remote Sens. Environ. 25:295-309           344

13 Valentini R. 2000, Nature 404:861-865                       337

14 Serreze M.C. 2000, Climatic Change 46:159-207               303

15 Loveland T.R. 1991, Photogram. Eng. Remote Sens. 57:1453-   296
   1476
16 Fan S. 1998, Science 282:442-446                            290

17 Richardson, A.J. 1977, Photogram. Eng. Remote Sens.         290
   43:1541-1552
18 Tucker & Sellers 1986, Intl. J.Remote Sens. 7:1395-1384     288

19 Baret F. 1991, Remote Sens. Environ. 35:161-173             250

20 Deschamps P.Y. 1994, IEEE Transactions on Geoscience        250
2 very important questions we
            face:
Human-accelerated climate change
               &
Unprecedented biological diversity
             loss
      Measured Surface
  Temperature the past 150
               years
5 warmest years: 1998, 2002, 2003, 2004,
2005
We must protect the Earth




Apollo 12‟s Classic Earth Rise from
Mauna Loa Observatory &
Charles David Keeling
Atmospheric CO2
measurements
Where are we?
        How did I get here today?
   In situ hyperspectral grassland studies 1971-73

                                                           leaf radiation
Plant physiological basis for band selection, NDVI,          modeling
hand-held instruments 1973-1975                               1972-73




Extensive field testing: Colorado & Yellowstone 1972-74; Iceland,
Sweden, England, Scotland, Wales 1976; Beltsville USA 1977-79



  Senegal: Use NDVI from NOAA satellites 1981-1983; work
  continued and expanded by Danish and other scientists


Start large-scale environmental studies with many others
Graduate School-Colo. State
          Univ.
Leaf Reflectance &
        Absorption




Bean leaf x 3900
       M.S. and Ph. D Work
1975 "Shortgrass Praire Spectral Measurements", C.J. Tucker, L.D.
Miller, and R.L. Pearson, Photogrammetric Engineering and Remote
Sensing, 41(9):1157-1162.

1976 "A Hand-held Spectral Radiometer to Estimate Gramineous
Biomass," R.L. Pearson, L. D. Miller, and C.J. Tucker, Applied Optics,
16(2):416-418.

1976 "Sensor Design for Monitoring Vegetation Canopies," C.J. Tucker
and E.L. Maxwell, Photogrammetric Engineering and Remote Sensing,
42(11):1399-1410.

1977 "Leaf Optical System Modeled as a Stochastic Process," C.J.
Tucker and M.W. Garratt, Applied Optics, 16:635-642.

1977 "Asymptotic Nature of Grass Canopy Reflectance," C. J. Tucker,
Applied Optics, 16(8):1059-1067.

1977 "Spectral Estimation of Grass Canopy Variables", C. J. Tucker,
Remote Sensing of Environment, 6(1):11-28.
Close to Missoula in 1974
Field Work in Iceland July
          1976
Field Work in Iceland July
          1976
Field Work in Iceland July
          1976
Beltsville USA winter wheat
          biomass
Winter wheat biomass
      “harvest”
S NDVI vs. total dry biomass




            Explained 80% of
                biomass
             accumulation
Satellite Test of NDVI-Biomass
            Results
1 mm/yr/km North-South precipitation
             gradient
Senegal‟s Ferlo Area
during rainy season
Marked contrasts between the
    dry and wet seasons




      (~300 mm/yr @ Senegal)
Satellite&ground
study 1981-1983
NOAA AVHRR 8-km NDVI Data
          Set
       Radiation
 The
GIMMS
Project
 1982
       NDVI Maximum Value
           Composites




  8-km output bin from 4-km spatial resolution data at
                       subpoint
All orbits, all data processed, equal area map projection
        Scan Angle Restriction to +/- 40 degrees
Average NDVI 1981-2006




~40,000 orbits
of satellite
data
NDVI = (ir- red)
        (ir+red)
                                  Satellit
                                  e NDVI
                                   data
                                  source
                                     s
                                                     NOAA-16
                                             NOAA 14       MODISes
                           NOAA 11            AVHRR
                  NOAA 9    AVHRR                                              NPP
                  AVHRR                                   SPOT
       NOAA 7
       AVHRR                                                            NOAA-18
                                                  SeaWiFS
                                     NOAA 9                       NOAA-17

1980            1985       1990       1995         2000          2005       2010
NDVI AVHRR-MODIS matchup
Different AVHRR NDVI Data Sets
 • Pathfinder from NASA/GSFC DAAC
 • NOAA Global Vegetation Index or
   GVI
 • GIMMS data set of NASA/GSFC
 • Others also

  All share more or less identical input
                     data
      Different calibrations, different
   volcanic corrections, different solar
           zenith angle corrections
NDVI problems with PAL and GVI
        for 35oN - 35oS
   What could have been done
            better?
Why wasn‟t the AVHRR ever improved
for better NDVI data by NOAA?
•Narrow channel 1 and channel 2
•Global 1 km data -- it‟s only 15 gb a day!
•Operate 6 channels all the time instead of 5
•There‟s no “L” in NOAA!
 What needs to be done better?
The earth science community must focus on
climate science -- satellites are crucial for this
AVHRR-MODIS-VIIRS data are the most important
climate science measurements (land-ocean-
atmosphere)
Landsat observations are the next most important
land climate science observations
Canopy lidar was the 3rd most important land
climate science observation
Avoid detours in the present fiscal situation that
compromise climate measurements
  What about “Why can‟t the earth
 science community get organized
 like the space science community
    for future satellite missions?
In the space science community, every group gets
their turn for missions because there is NO NEED
FOR MEASUREMENTS OVER TIME
Climate change is THE earth science question
We all must support satellite and ground
measurements that provide key climate science
information thru time
Time series satellite data are invaluable for climate
Satellite Remote Sensing of
            Earth




    SeaWiFS Ocean Chlorophyll
           Land NDVI
5 SeaWiFS land bands
                                                                      Sahara Desert
               0.50


               0.45


               0.40


               0.35                                        Sahara RFL412                      Sahara RFL443
reflectance




                                                           Sahara RFL555                      Sahara RFL670                          Sahara Desert
               0.30                                        Sahara RFL875


               0.25


               0.20


               0.15


               0.10


               0.05

                    97          97       98       98       98          98       99       99       99          99       00       00       00          00       01       01       01          01       02
                 22/         2/       22/      22/      22/         2/       22/      22/      22/         2/       22/      22/      22/         2/       22/      22/      22/         2/       22/
              9/          /2       3/       6/       9/          /2       3/       6/       9/          /2       3/       6/       9/          /2       3/       6/       9/          /2       3/
                       12                                     12                                     12                                     12                                     12

                                                                                                             Time
Sahel Zone
Sudanian NDVI - EVI
EVI = 2.5*[ir-red]/[ir+(6*red)-7.5*(blue)+1]
Amazon forest 443 & 670 nm
          bands
Amazon NDVI vs r 670 nm
           Amazon EVI




EVI = 2.5*[ir-red]/[ir+(6*red)-7.5*(blue)+1]
EVI = 2.5*[ir-red]/[ir+(6*red)-7.5*(blue)+1]
NDVI vs. Red Reflectance
EVI vs. Red Reflectance
       Who invented the NDVI?




                        or


              Charlie
              Chan?

The Creature from the
Black Lagoon never           Larry, Mo, & Curly?

								
To top