1. Vegetation Indices
Radiometric measures of the amount, structure, and condition of vegetation,
Precise monitoring tool
phenology
inter-annual variations
Serve as intermediaries in the assessment of various biophysical parameters
green cover,
biomass,
leaf area index (LAI),
fAPAR
chlorophyll conc.

2. APPLICATIONS
Indicators of seasonal and inter-annual variations in vegetation (phenology)
Change detection studies (human/ climate)
Tool for monitoring and mapping vegetation
Serve as intermediaries is the assessment of various biophysical parameters: leaf area index (LAI), % green cover, biomass, FPAR, land cover classification

4. Relating transpiration and photosynthesis to NDVI, 1988

5. Testing limits of interpretation for NDVI, 1988

6. Relating canopy processes to NDVI, 1988

7. Atmospheric Influences on Spectral Response Functions
Total Radiance
Path Radiance
Sunlight
Water vapor absorption
Scattering by aerosols
Reflected Energy
Skylight
Total energy incident at the surface is comprised of direct and skylight energy.
Total energy at the sensor will be a function of direct reflectance from the surface (accounting for skylight), and the Path Radiance.
Atmosphere influences are not the same for Red and NIR

9. FPAR (Fraction of absorbed PAR):
BIOPHYSICAL MEASURES
Leaf Area Index (m2/m2):
FPAR (Fraction of absorbed PAR):
Incident Radiation
Ground
Leaf
PAR
absorption
(radiometric)
Leaf Area
(structural)

10. Theoretical basis for
spectral vegetation indices:

11. Aerosol Correction from Collection 1 to Collection 4
Collection 4 (Current re-processing)
No aerosol correction

12. Annual Cloudcover Percentage for MODIS
Figure: MODIS 2003 NPP QC outputs showing regional impacts of cloud/aerosol effects on optical/IR satellite retrievals. Unlike microwave remote sensing, optical remote sensing is strongly constrained by atmospheric conditions and solar illumination. Tropical forests of South America and Southeast Asia, for example, are largely obscured by frequent cloud cover, smoke and other atmospheric aerosols. At high latitudes reduced solar illumination, enhanced shadowing and atmospheric aerosols from boreal fire activity significantly degrade optical remote sensing retrievals as well. Data processing techniques such as spatial and temporal compositing can partially mitigate these effects, but at the price of decreased spatial and temporal fidelity. New remote sensing methods are needed that integrate synergistic information from multiple sensors for improved information extraction and global monitoring capabilities. For example, microwave remote sensing provides day-night and all-weather monitoring capabilities, as well as sensitivity to landscape structure and moisture conditions. Integration of these data with optical remote sensing based estimates of vegetation photosynthetic properties would improve global assessment and monitoring of soil-vegetation interactions with the atmosphere that could benefit regional weather forecast accuracies. The wide array of current and planned sensors and integrated satellite platforms such as those available under NASA EOS and future NPOESS programs offer unprecedented coverage of the globe and rich opportunities to exploit synergy among different sensors for improved biospheric monitoring.

13. Persistent clouds

14. Compositing Algorithm
Provide cloud-free VI product over set temporal intervals,
Reduce atmosphere variability & contamination
Minimize BRDF effects due to view and sun angle geometry variations
Depict and reconstruct phenological variations
Accurately discriminate inter-annual variations in vegetation.
Physical and semi-empirical BRDF models
Maximum VI (MVC) or constrained VI (CMVC)

15. Normalized Difference Vegetation Index (NDVI)
The NDVI is a normalized ratio of the NIR and red bands,
NDVI is a functionally equivalent to and is a non-linear transform of the simple ratio. 3

16. Enhanced Vegetation Index (EVI)
rNIR - rred
EVI =
L + rNIR + C1 rred + C2 rblue
L = canopy background adjustment,
C1 and C2 are for aerosol correction and feedback
L=1, C1 and C2 are 6 and 7.5
G = gain factor of 2.5
Reduces both atmosphere and canopy background contamination.
Increased sensitivity at high biomass levels (less saturation)
Linear *G

17. 1 km VI’s Tapajós ‘Forest’ Day 113 - 128
NDVI
EVI
1 km VI’s Tapajós ‘Forest’
Day
EVI
NDVI

18. NDVI - Soil Sensitivity1
0.9
LAI=2.7
0.8
0.7
LAI=1.6
0.6
0.5
NDVI
LAI=1.1
0.4
0.3
0.2
LAI=0.5
0.1
Soil
0.05
0.1
0.15
0.2
0.25
0.3
0.35
0.4
Soil Albedo

19. Soil Adjusted Vegetation Index
SAVI isolines overlap vegetation isolines over a wide range of LAI values.
SAVI becomes insensitive to soil noise within this range of LAI
The perfect range depends on the choice of the ‘L’ value in the SAVI formulation.

20. SAVI Soil Albedo 0.7 0.6 0.5 0.4 0.3 0.2 0.1 Soil LAI=0.5 LAI=1.1
0.05
0.1
0.15
0.2
0.25
0.3
0.35
0.4
Soil Albedo

21. Inter-relationships among Spectral VI’s
VI’s behave similarly to each other under a ‘constant’ set of conditions,
If soils are not varying then NDVI and SAVI are well correlated,
If aerosol differences are minimal, then similarly there is minimal difference between EVI and SAVI.

22. NDVI & EVI Relationships
MODIS Data ( )
RT-Model

23. EVI & SAVI Relationships
MODIS Data ( )
RT-Model
EVI

24. VI - FPAR Relationships
NDVI
EVI

25. …The Algorithm
Weighted average scheme

26. Long term stability

27. Global NDVI at 500 m DOY

28. MODIS Standard Vegetation Index Products Products
The MODIS Products include 2 Vegetation Indices (NDVI, EVI) and QA produced at 16-day and monthly intervals at 250m/ 500m, 1km, and 25km resolutions
The narrower ‘red’ MODIS band provides increased chlorophyll sensitivity (band 1),
The narrower ‘NIR’ MODIS band avoids water vapor absorption (band 2)
Use of the blue channel in the EVI provides aerosol resistance
The at launch MOD13 algorithm will allow the individual processing of two vegetation indices at different spatial and
temporal resolution. The Level 3 HDF filespec will therefore be split in 5 files/products (MOD_PR13A, MOD_PR13B,
MOD_PR13C, MOD_PR13D, MOD_PR13E) that each have commonalities with respect to spatial and spectral
resolutions. The standard DAAC production run will process the NDVI at 250 m resolution for 16-day intervals. The
enhanced VI (EVI) and NDVI will both be produced at 1km and CMG and both 16 days and monthly intervals. The
output products will have datafields for the NDVI and EVI with corresponding QA, reflectance data, angular
information and spatial statistics (mean and std of each VI and for the CMG scales.

29. Radiometric Measures
Vegetation Indices
SR (Simple Ratio), MSR (Modified SR)
SAVI (Soil Adjusted VI), MSAVI, ARVI, GEMI
NDVI (Normalized Difference Vegetation Index)
EVI (Enhanced Vegetation Index)
Biophysical Measures
Leaf Area Index (Area of leaves per unit ground area, m2/m2)
FPAR (Fraction of incident PAR that is absorbed)

30. VI Equations SVI Formulations Enhanced Vegetation Index:
Simple Ratio = NIR/Red
Normalized Difference = (NIR-Red)/(NIR+Red)
Enhanced Vegetation Index:
-where r is atmospherically-corrected, surface reflectances, L is the canopy background adjustment, G is a gain factor, and C1 , C2 are coefficients for atmospheric resistance.

31. Annual average EVI
Amazon vegetation seasonal analyses and land conversion effects on biologic activity.

32. 1 km VI’s Tapajós ‘Forest’ Day 113 - 128
NDVI
EVI
1 km VI’s Tapajós ‘Forest’
Day
EVI
NDVI

34. Vegetation seasonality in the Brazilian ‘Cerrado’ Region

35. Histograms of VI’s at 250 m, 500 m, and 1 km resolutions
NDVI
EVI
South America (August 12 to August 27, 2000)

36. Analyses, Uncertainty & Validation
Heterogeneous surfaces (scaling issues)
Dynamic Range
Tropical forests & clouds
Arid/ semiarid regions
Snow - vegetation problems
Cloud shadows

37. Scaling issues & heterogenous surfaces
Tropical forest conversion (Ikonos)
Brasília – CCD/CBERS-1
CB2 - Minas Gerais - nov/03

38. 100%
75%
50%
25%
75%
50%
25%
25% crop

39. Southwest Megadrought Analysis per land cover type (MODIS 3-year)

40. Inland water bodies (Caspian Sea)
Traced to over/under corrected surface reflectance over water bodies
Hypersensitivity of NDVI to the proportional relation between Red & NIR

41. Spatial issues (Blocky retrieval)
Traced to Aerosol correction

42. Snow-Vegetation Surfaces
Snow has Red > NIR causing numerator of VI’s (NDVI, SAVI, EVI) to become negative (or decrease in case of mixed pixels).
Snow also has Blue > Red causing denominator of EVI equation to decrease and, at times, become negative

43. Inter-relationships among biophysical products (VI, LAI/FPAR)
Global FPAR 2004
Global LAI

44. MODIS Phenology Logic

45. 10/08/2002
UNBC Seminar

46. MOD12Q2: Global Vegetation Phenology
From Mark Friedl, Boston Univ.
First global products for vegetation phenology based on MODIS EVI data released for
Identifies key transition
dates in growing season
Onset EVI increase
Onset EVI maximum
Available globally
Legend provides julian date for
Onset EVI decrease
Onset EVI minimum