Remote Sensing of LAI Conghe Song Department of Geography University of North Carolina Chapel Hill, NC 27599.

Remote Sensing of LAIRemote Sensing of LAI

Conghe SongConghe Song

Department of GeographyDepartment of Geography

University of North Carolina University of North Carolina

Chapel Hill, NC 27599Chapel Hill, NC 27599

Small Leaves do the Big JobSmall Leaves do the Big Job

26126roplastlight/Chlo

22 O6OHCO6H6CO

CO2

H2O

Losing water is the price to pay for gaining CO2

stoma

Underside

Big-Leaf Scaling upBig-Leaf Scaling up

Net Radiation

Q

Qr

La

Ls

)()( asrn LLQQR

Radiation Budget:

Energy Budget: Net radiation is used for heat storage in the soil and plant body, evaporate water (latent heat), heating the air near the surface (sensible heat), and photosynthesis

Latent Heat (Evapotranspiration)Latent Heat (Evapotranspiration)

Transpiration Evaporation

T=f(Rnet, gs, ga, VPD, T)

Uniform Canopy vs. Gappy Canopy (two-leaf scaling up)

+ = Evapotranspiration

H2O

H2O

Carbon Assimilation

Plant Resp.

Soil Resp.

NetPSN

Tower Measurement: Net Ecosystem Exchange (NEE) Measured data cannot separate Net PSN from plant (stems, leaves and roots) and soil (dead biomass) respiration. A model is needed to estimate the plant and soil respiration in order to get Net PSN.

Spectral Information of Remotely Sensed ImagerySpectral Information of Remotely Sensed Imagery

RedNIR

RedNIR

RR

RRNDVI

NDVI: [-1.0, 1.0]

Red

NIR

R

RSRVI

Red

NIR

ChallengesChallenges1. Very difficult to get ground truth data: (1) Destructive Sampling: Based on forest inventory approach, cut a tree

and measure its total leaf weight, take a sample of leaves to estimate the area to weight ratio (specific leaf area) and convert to the total leaf area for the tree. You will need to do multiple trees so that you can develop an allometric relationship between leaf area and diameter at breast height. The allometric relationship changes with species. Therefore, you have to do this for each species.

(2) Optical measurements: Based on the light intensity inside and outside a vegetation canopy measured with optical instruments, and the estimate LAI under certain assumptions.

2. Many factors confound with remote sensing signals that can cause errors in estimating LAI (1) landscape heterogeneity: different types of vegetation, land-cover types. (2) leaf angle distribution, different degrees of leaf clumping. (3) variation of soil reflectance

Empirical Studies: LAI vs NDVI/SAVIEmpirical Studies: LAI vs NDVI/SAVI

Turner et al. 1999. RSE, 70: 52-68.

Empirical Study between LAI and NDVI/SR

Chen and Cihlar, 1996. RSE, 55: 153-162.

NDVI vs. LAI from Ikonos Imagery

Colombo et al. 2003, RSE,86: 120-131.

Empirical StudiesEmpirical Studies

y = 4.3299x - 0.3806

R2 = 0.2777

0

2

4

6

8

10

12

14

1 1.5 2 2.5

SRVI

Le

af

Are

a I

nd

ex

Hardwoods

Pines

y = 13.866x + 3.8225

R2 = 0.2048

0

2

4

6

8

10

12

14

0 0.1 0.2 0.3 0.4 0.5

NDVIL

ea

f A

rea

In

de

x

Hardwoods

Pines

Song and Dickinson, in press, IJRS.

LAI and Spatial SignalsLAI and Spatial Signals

y = 0.0001x + 3.5477

R2 = 0.6238

0

2

4

6

8

10

12

14

0 10000 20000 30000 40000 50000 60000

Image Variances (Pixel Size=4x4m)

Lea

f A

rea

Ind

ex

hardwoods

conifer

y = -7.5334x + 17.85

R2 = 0.567

2

3

4

5

6

7

8

9

10

11

12

1 1.2 1.4 1.6 1.8 2

V2/V3

Lea

f A

rea

Inde

x

Hardwoods

Conifer

Song and Dickinson, in press, IJRS.

Physical Based Model Inversions from Remote SensingPhysical Based Model Inversions from Remote Sensing

MODIS/MISR LAI Product:

(1) Global vegetation classified into 6 biomes: Broadleaf Forests, Needleleaf Forests, Shrubs, Savannas, Broadleaf Crops, Grasses and Cereal Crops.

(2) Use 3-D radiation transfer algorithms to invert for LAI and built a look-up-table (LUT) to retrieve LAI operationally from remotely sensed signals.

(3) Independent validations found that MODIS LAI provide a good estimates when LAI is low, while it can be significantly overestimated (Cohen et al., 2006).

Validation of MODIS LAI version 3Validation of MODIS LAI version 3

Cohen et al. 2003. RSE, 88:233-255

Validation of MODIS LAI version 4Validation of MODIS LAI version 4

Cohen et al., 2006. IEEE Transactions on Geoscience and Remote Sensing, 44(7): 1843-1857,