1

23/07/2013 1

Introduction to thermal infrared remote sensing ‐ Surface Energy Balance System Basics

Z. (Bob) SuITC, University of Twente Enschede, The Netherlands

z.su@utwente.nlwww.itc.nl/wrs

1. To understand basic concepts of surface radiation budget 

2. To understand basic measurements to derive surface radiation components

3. To be able to derive surface radiation components using different information 

4. To familiarize with retrieval of surface parameters

Learning Objectives

2

Wind

Land-Atmosphere Interactions - Terrestrial Water, Energy and Carbon Cycles

Lat

ent H

eat

(Pha

se c

hang

e)

Pre

cipi

tati

on

Biochemical Processes

Advection

VIS

UV

NIRTIR

MIR

SOLAR TERRESTRIAL

SOLAR AND TERRESTRIAL SPECTRUM

Wavelength (m)

THERMAL INFRARED RADIATION is a form of electromagnetic radiation with a wavelength between 3 to 14 micrometers (µm). Its flux is much lower than visible flux.

Source: J.-L. Casanova

3

RADIANT POWER: The rate of flow of electromagnetic energy, i.e., radiant energy, (watts, i.e., joules per second.)

RADIANT EMITTANCE: Radiant power emitted into a full sphere, i.e., 4π (steradians), by a unit area of a source, expressed in watts per square meter. Synonym radiant exitance, flux, radiant flux, E (W m-2)

RADIANCE: Radiant power, in a given direction, per unit solid angle per unit of projected area of the source, as viewed from the given direction. Note: Radiance is usually expressed in watts per steradian per square meter, L (W m-2.sr-1) = M/

IRRADIANCE: Radiant power incident per unit area upon a surface. Note: Irradiance is usually expressed in watts per square meter, but may also be expressed in joules per square meter. Synonym: power density, flux, radiant flux, M (watts/m2).

Some definitions

Source: J.-L. Casanova

Graphical representation of emittance, radiance and irridiance

Source: J.-L. Casanova

RADIANCE, L = E/ (Wm-2.sr-1)

IRRADIANCE, M (Wm-2)

EMITTANCE or EXITANCE, E (Wm-2)

4

All these magnitudes are usually expressed as spectral magnitudes, that is to say, referred to a wavelength

RADIANT POWER: P (W.m-1) or (W.cm-1)

RADIANT EMITTANCE: E (W.m-2.m-1) or (W.cm-1m-2)

RADIANCE: L (W.m-2.sr-1. m-1) or (W.cm-1 m-2.sr-1) = E/

IRRADIANCE: M (W.m-2. m-1) or (W.cm-1m-2).

Spectral magnitudes

Source: J.-L. CasanovaWavenumber (cm-1) = number of wavelengths per cm = 10,000/ (m) i.e. (m) = 10,000/ wavenumer (cm-1)e.g. 1 m ~ 10,000 cm-1; 10 m ~ 1,000 cm-1…

c1 = 1.19104·108 W µm4 m-2 sr-1

c2 = 1.43877·104 µm K, are the first radiation constant (for spectral radiance) and second radiation constant, respectively, B (λ, Ts) is given in W m-2 sr-1 μm-1 if λ the wavelength is given in µm.

The theory to express the thermal emission is based on the black body concept: a black body is an object that absorbs all electromagnetic radiation at each wavelength that falls onto it. A black body however emits radiation as well, its magnitude depends on its temperature, and is expressed by the Planck’s Law:

1exp

),(2

51

S

S

T

c

cTB

Thermal emission - the Planck’s Law:

5

Thermodynamic (kinetic) temperature

Brightness temperature

Directional radiometric surface temperature & Directional emissivity

Hemispheric broadband radiometric surface temperature & Hemispherical broadband emissivity

Terminology in Thermal Remote Sensing

Source: J. Norman, and F. Becker, 1995

Thermodynamic (kinetic) temperature (T) is a macroscopic measure to quantify the thermal dynamic state of a system in thermodynamic equilibrium with its environment (no heat transfer) and can be measured with a infinitesimal thermometer in good contact with it.By maximizing the total entropy (S) with respect to the energy (E) of the system, T is defined as

Thermodynamic (kinetic) temperature

TdE

dS 1

which is consistent with the definition of the absolute temperature by the ideal gas law

RTPV

where P is pressure, V volume, R the gas constant (R=kN0, k is Boltzmann constant, N0 number of molecules in a mole).

6

5223Nadir

Object

Sensor

The Fundamental

of Earth Observation

(Sensor - Object Radiative Relationship)

Sensor ResponseA. How much radiation is detected?

B. When does it arrive?

Object Properties:Its range, its combined temperature &

Emissivity (or reflectivity)at different times, at different spatial resolution, at different wavelengths, at different direction,

at different polarization

A: A Passive Sensor SystemA+B: An Active Sensor System

atmosphere

Brightness temperature (Tb,i) is a directional temperature obtained by equating the measured radiance (Rb,i) with the integral over wavelength of the Planck’s black body function multiplied by the sensor response fi.

Brightness temperature

2

1

1,

exp

,,

,

25

1,,

d

TC

CfTRR

ib

iibib

12

1

df i is the detector relative response.

7

The spectral radiance (RR) measured by a directional radiometer is the sum of the emitted black body radiance modified by the emissivity and the reflected radiation within an infinitesimal wavelength band.

Directional radiometric surface temperature & Directional emissivity

2/

0 ,

2

0

,,

sin,cos,,,

,,,

ddL

RR

b

BR

Bidirectional reflectance distribution function Sky radiance

If the incident sky radiance is isotropic, we have

2/

0 ,

2

0,sin,cos,,,

LddLb

Therefore for opaque bodies at thermal equilibrium,

,1,

A hemispheric radiometric surface temperature is used to provide an estimate of the emitted radiance (Rl) over a broad wavelength and the full hemisphere of view

Hemispheric broadband radiometric surface temperature & Hemispherical broadband emissivity

42/

0 ,

2

0cossin,,

2

1RBl TdddRR

8

Corn

Bare Soil

Water Body

Garlic

Green Grass

Vineyard

Reforestation

In-Situ measurements of Surface Energy Balance- EAGLE/SPARC Campaign 2004, Barrax, Spain

- Situated in the area of La Mancha, in the west of the province of Albacete, 28 km from Albacete- Geographic coordinates: 39º 3’ N; 2º 6’ W- Altitude (above sea level): 700 m

The Canopy from Different Perspectives

9

Factors Influencing the Emissivity

• The material (minerals, water etc.)• The surface geometry (roughness of the surface) • The wavelength of the radiation • The view angle

(Source: Sobrino et al., 2005)

800 1200 1600 2000 2400 28000.5

0.6

0.7

0.8

0.9

1.0

Water:

Soil:

Spe

ctra

l em

issi

vity

Wave number (cm-1)

Seawatr1Vegetation: 106

108 121 e3101 e4643 eply3c7

Spectral emissivity extracted from MODIS UCSB Emissivity Library

12.5 8.3 6.3 4.2 3.6 3.3

Wavelength (μm)

10

Solar R

adiation

Atm

ospheric therm

al radiation

Reflected solar radiation

Surface therm

al radiation

swuR

Surface Radiation Budget - Measurements

lwulwdswuswdn RRRRR

swdR lwdR lwuR

Net radiation

CNR1 net radiometer:

Spectral response- Pyranometer: 305 to 2800 nm- Pyrgeometer: 5000 to 50000 nm

Radiation components: 1. incoming short-wave radiation2. surface-reflected short-wave radiation3. incoming (long-wave) thermal infrared radiation 4. outgoing (long-wave) TIR radiation

Albedo (whiteness): ratio of scattered to incident electromagnetic radiation

nmm

mm

10001

505

1

2

3

4

Surface Radiation Budget - Measurements

11

EAGLE/SPARC Campaign 2004, Barrax, Spain

Barrax 2004 Radiation @ Vineyard

-200

0

200

400

600

800

1000

1200

196 197 198 199 200 201 202 203

SW_inc

SW_out

LW_inc

LW_out

Radiation Components, Barrax, 15 July 2004

-200

0

200

400

600

800

1000

1200

0 50 140

230

320

410

500

550

640

730

820

910

1000

1050

1140

1230

1320

1410

1500

1550

1640

1730

1820

1910

2000

2050

2140

2230

2320

Time

W/M

^2 SW_in

SW_outLW_inLW_out

EAGLE/SPARC Campaign 2004, Barrax, Spain

12

EAGLE/SPARC Campaign 2004, Barrax, Spain

Exercise:

1. Explain what are the influencing factors for the net radiation2. Explain the pattern of the radiation components 3. Calculate the albedo from the radiation components

Wind

How is the net radiation consumed?

LE

00

S

SLEHGRn

cfT ,,, 0

Right hand side:1. Soil heat flux (warming up of the soil layer)2. Sensible heat flux (warming up of the air layer)3. Latent heat flux (phase change, evaporation and transpiration)4. Heat storage and biological activities

13

Forest nurseryDutch site

garlicvineyard

corn

Bare soil

Wheat

~3 km

3

12

4

5

INRA sitesForest nursery

Dutch site

garlicvineyard

corn

Bare soil

Wheat

~3 km

3

12

4

5

Dutch site

garlicvineyard

corn

Bare soil

Wheat

~3 km

3

12

4

5

INRA sites

Locations of measurements 

Source: Su et al., (2008)

The soil heat fluxes were measured with 4 soil heat fluxes buried at a depth of 1cm below surface, with two placed in the middle of the row and two under the vine and shaded. 

Soil heat flux plates:SH1 and SH4 were in the shadeSH2 and SH3 were in the sunlit areas

Soil heat flux: measurements with soil heat flux plates

SH2SH3

SH1SH4

14

Soil heat flux: measurements with soil heat flux plates

Sensible heat flux: scintillometer measurements

•Scintillometer, 2 Sets• Radiation components• Soil Heat flux plates• Temperature profile (air & soil)

Transmitter

Receiver

15

Scintillometer Data, EAGLE/SPARC Campaign 2004, Barrax, Spain

Barrax 2004 Sensible heat f luxes

-200

-100

0

100

200

300

400

500

600

196 197 198 199 200 201 202 203

Vineyard

Corn-C2

Reforestation-F1

Sunflow er-SF1

Bare-B1

corn-3D

Scintillometer Data, EAGLE/SPARC Campaign 2004, Barrax, Spain

Barrax 2004 Fluxes @ Vineyard

-400

-200

0

200

400

600

800

196 197 198 199 200 201 202 203

Rnet

G_avg

H-LAS

LE_Rest.

16

Sensible & latent heat fluxes:Eddy Covariance Measurements

Canopy level measurements:

Eddy covariance system (Gill 3D sonic + closed path Licor

gasanalyser: CO2 and H2O + nitrogen reference gas + pneumatic mast + data loggers)

• Turbulence, sensible heat flux, H2O, CO2 fluxes

• CO2 concentrations

Eddy Correlation Data, 10 Min Average, July 2004

-500

-300

-100

100

300

500

700

DOY 195-203

-20

-10

0

10

20

30

40

H W m-2

LE W m-2

Fc μmol m-2 s-1

Flux data SPARC 2004

17

Energy balance components

-200

-1000

100200

300

400500

600700

800

196 197 198 199 200 201 202 203

H_ed

LE

Rn

G

H_las

Energy balance (60 minute average), Vineyard, Barrax 2004

Energy balance closure ??

(Sum of H and LE exceeds the available energy)

-200

0

200

400

600

800

196 197 198 199 200 201 202 203

H_ed+LE_ed

R_avai10 minutes average, Vineyard, Barrax 2004

-200

0

200

400

600

800

196 197 198 199 200 201 202 203

H_ed+LE_ed

R_avai60 minutes average, Vineyard, Barrax 2004

18

Solar R

adiation

Atm

ospheric therm

al radiation

Reflected solar radiation

Surface therm

al radiation

eTemperaturSurfaceT

emissivity

albedo

:

:

:

0

Surface Radiation Budget ‐ Parameterisation

401 TRRR lwdswdn

Pre-processing

Remote Sensing Data

Geographical Data

Regional Atmospheric State

Remote sensing retrievals Weather prediction model outputs

Radio sounding

ENVISAT NOAA/AVHRR, LANDSAT, METEOSAT, MODIS

MSG

DN Radiance/Reflectance at Sensor

Radiance/Reflectance at Surface

Albedo, Temperature, Emissivity, Vegetation Parameters, Roughness, Incoming Radiation

Atmospheric correction

Inversion of radiative transfer models

19

Incoming global radiation

Albedo

Vegetation cover, Leaf Area Index

Surface temperature

Emissivity

Roughness

Derived Surface Parameters/variables

Components of reflected and scattered solar beams received at the satellite

20

Input parameters:

– Group 1: date, time, location (latitude) and DEM (elevation, slope, aspect) to compute solar declination, eccentricity, solar zenith angle for a horizontal surface, solar azimuth angle, solar zenith angle on slopes

– Group2: weather condition ‐ Optical properties of the atmosphere (at the time of calculation)

Incoming global radiation

z

Case: a

z

Case: b

In Ida Idr

Idm

• Case a: use total optical depth

•

• where Isc is the solar constant, e0 the eccentricity factor, z solar zenith angle, m 

air mass,  the optical depth

)exp(cos0 meII zsc

Incoming global radiation

21

Incoming global radiation (cont.)

Case b: use water-vapor / horizontal visibility

where the direct solar radiation is

etc.References:M. Iqbal (1983) An introduction to solar radiation, Academic Press, Toronto. p.188-191.R. Bird and R.L. Hulstrom (1980) Direct insolation models, Trans. ASME J. Sol. Energy Eng., 103, 182-192. R. Bird and R.L. Hulstrom (1981) A simplified clear sky model for direct and diffuse insolation on horizontal surfaces, SERI/TR-642-761, Solar Energy Research Institute, Golden, Colorado.

ag

dadrndmdadrn

dn

IIIIIII

III

1

1

awgorzscn eII cos9751.0 0

Algorithms

‐ Calculate surface bidirectional reflectance of narrow band

‐ Derive surface broad‐band albedo from narrow surface 

reflectance

‐ Derive NDVI from band RED & NIR surface reflectance 

‐ (Ref. Lecture optical remote sensing)

Surface Albedo & NDVI

22

Algorithms

Land surface temperature derived by using a theoretical split‐window algorithm 

References: Sorino et al.(2004)

Surface temperature

The at-sensor radiance for a given wavelength (λ) s:

where

– ελθ is the surface emissivity,

– Bθ(λ, Ts) is the radiance emitted by a blackbody at temperature Ts of the surface,

is the downwelling radiance,

– λθ is the total transmission of the atmosphere (transmittance) and

is the upwelling atmospheric radiance. All these magnitudes also depend on the observation angle θ.

The radiative transfer equation for LST retrieval 

atmatmS

sensorat LLTBL )1(),(

atmL

atmL

23

According to Sobrino et al. (1996), the upwelling and downwelling atmospheric radiance can be substituted, respectively, by: 

)()1( aiiatm TBL

)()1( 53 aiiatm TBL

where Ta is the effective mean atmospheric temperature andi53 is the total atmospheric path transmittance at 53 degrees.

Substituting both atmospheric radiances in the radiative transfer equation, an algorithm involving temperatures can be obtained using a first‐order Taylor series expansion of the Planck’s law and writing the equation for i and j (i and j being two different channels observed at the same angle, Split‐Window method):

Split‐window equations to derive land surface temperature

Ts = Ti+A(Ti-Tj)-B0+(1-i)B1- B2

where A and Bi are coefficients that depend on atmospheric transmittances, i is the mean value of the emissivities of channels i and j, is the spectral variation, Ti and Tj are the brightness temperatures for two different channels with the same view angle.

This equation is the so-called split-window equation and gives a separation between the atmospheric and emissivity effects in the retrieval of surface temperature.

24

Split‐window and dual‐angle algorithms for AATSR sensor with values of the coefficients calculated by a minimization process

Notation: n:  Nadir view;f:  Forward view;SW:  Split‐Window Method (two spectral channels at the same observation angle)quad: algorithm that includes a quadratic dependence on (Ti‐Tj);(1):  AATSR Channel 1 (12 m);(2):  AATSR Channel 2 (11 m);W:  algorithm with water vapor content dependence;:  algorithm with emissivity dependence;:  algorithm including spectral or angular emissivity difference.

25

Algorithms

Estimation of surface emissivity using the theoretical model of Caselles and Sobrino (1989).

References:

V. Caselles and J.A. Sobrino (1989) Determination of frosts in orange groves from NOAA-9 AVHRR data, RSE, 29:135-146.

E. Valor and V. Caselles (1995) Mapping land surface emissivity from NDVI: Application to European, African, and South American Areas, RSE, 57:167-184.

Sobrino and Soria (2006)

Emissivity

A simplified method to emissivity, ᵋ, using atmospherically corrected data in the visible

and near infrared channels (Sobrino and Raissouni, 2000). Three different surface

types: bare soil pixels (NDVI<0.2), mixed pixels (0.2<NDVI<0.5) and fully vegetation

pixels (NDVI>0.5).

The NDVI is calculated with reflectivity values from the Red region (red) and Near

Infrared (nir) region:

AATSR channels centred in 0.67 m and 0.87 m used for red and nir respectively. 

Thresholds Method (NDVITHM)

nirred

nirredNDVI

26

The final expressions obtained for this method are for bare soil pixels (NDVI < 0.2), = 0.9825 ‐0.051 redΔ = ‐0.0001 ‐0.041 red

for mixed pixels (0.2<=  NDVI  <=0.5), = 0.971 + 0.018 fcΔ = 0.006 (1 ‐ fc )

and for vegetation pixels (NDVI > 0.5), = 0.990

with fc being the vegetation proportion, given by

where NDVImin=0.2 and NDVImax=0.5. The main constraint of this method is that it can not be used to extract water emissivity values because it is not possible to apply the NDVI and fc equations for water pixels.

Thresholds Method (NDVITHM)

2

minmax

min

NDVINDVI

NDVINDVI

cf

20

200 vgor ZZZ

Roughness

Total roughness

Vegetation roughness

Orographic roughness vp

Zsize

or 1

0

),(0 cvg fhfZ

(h - vegetation height, fc – fractional coverage)

27

Questions:

1. What is surface radiation budget 

2. List some methods to derive surface radiation components

3. What information is needed in order to derive surface 

radiation components using the above mentioned methods 

4. What are the essential surface parameters in determination of surface 

radiation balance 

Su, Z., Timmermans, W., Gieske, A., Jia, L., Elbers, J. A., Olioso, A., Timmermans, J., Van Der Velde, R., Jin, X.,Van Der Kwast, H., Nerry, F., Sabol, D., Sobrino, J. A., Moreno, J. and Bianchi, R., 2008, Quantification of land‐atmosphere exchanges of water, energy and carbon dioxide in space and time over the 

heterogeneous Barrax site, International Journal of Remote Sensing, 29:17,5215‐5235.

Su, Z., J. Wen, and L. Wan, 2003, A methodology for the retrieval of land physical parameter and actual evaporation using NOAA/AVHRR data, Journal of Jilin University (Earth Science Edition), 33(sup.), 106‐118. 

Su, Z., Timmermans，W. J., van der Tol， C., Dost， R. J. J., Bianchi， R., Gómez， J. A., House， A., Hajnsek， I., Menenti，M., Magliulo， V., Esposito，M., Haarbrink， R., Bosveld， F. C., Rothe， R., Baltink， H. K., Vekerdy， Z., Sobrino， J. A., Timmermans， J., van Laake， P., Salama， S., van der Kwast，H., Claassen， E., Stolk， A., Jia， L., Moors， E., Hartogensis， O., and Gillespie， A., 2009, EAGLE 2006 ‐multi‐purpose, multi‐angle and multi‐sensor in‐situ, airborne and space borne campaigns over grassland and forest. Hydrology and Earth System Sciences, 13, 833–845.

Jia, L., Z.‐L. Li, M. Menenti, Z. Su, W. Verhoef, Z. Wan, 2003, A practical algorithm to infer soil and foliage component temperatures from bi‐angular ATSR‐2 data, International Journal of remote Sensing, 24 (23), 4739–4760.

Li, Z.‐L., L. Jia, Z. Su, Z. Wan, R.H. Zhang, 2003, A new approach for retrieving precipitable water from ATSR‐2 split window channel data over land area, International Journal of Remote Sensing, 24(24), 5095–5117.

Sobrino, J., G. Soria, F. Prata, 2004, Surface temperature retrieval from Along Track Scanning Radiometer 2 data: Algorithms and validation, Journal of Geophysical Research, Vol. 109, doi:10.1029/2003JD004212.

Sobrino, J.A. , J.C. Jim´enez‐Mu˜noz,., P.J. Zarco‐Tejada, G. Sepulcre‐Cant´o, E. de Miguel, G. S`oria, M. Romaguera, Y. Julien, J. Cuenca, V. Hidalgo, B. Franch, C. Mattar, L. Morales, A. Gillespie, D. Sabol, L. Balick, Z. Su, L. Jia, A. Gieske, W. Timmermans, A. Olioso, F. Nerry, L. Guanter, J. Moreno, and Q. Shen, 2009, Thermal remote sensing from Airborne Hyperspectral Scanner data in the framework of the SPARC and SEN2FLEX projects: an overview, Hydrol. Earth Syst. Sci., 13, 2031–2037.

Norman, J., F. Becker, 1995, Terminology in thermal remote sensing of natural surfaces, Agricultural and Forest Meteorology, 77, 153‐166.

Wan, Z., Y. Zhang, Q. Zhang, Z.‐L. Li, 2004, Quality assessment and validation of the MODIS global land surface temperature, International Journal of Remote 

Sensing, 25(1), 261–274.

References/Further Readings

28

Field experimental data access:

SPARC2004, SEN2FLEX2005, EAGEL2006 data, REFLEX2012 data

ESA’s Principle Investigators portal http://earth.esa.int/