Remote Sensing of Evapotranspiration

with MODIS

Remote Sensing of Evapotranspiration

with MODIS

Daniel Siegel

What is MODIS?What is MODIS?Moderate-Resolution Imaging Spectroradiometer

Launched in 1999 aboard the EOS AM (Terra); EOS PM (Aqua) followed in 2002

Monitors 36 spectral bands between 0.4 m and 14.4 m

Images entire Earth every 1-2 days at 1 km resolution

Moderate-Resolution Imaging Spectroradiometer

Launched in 1999 aboard the EOS AM (Terra); EOS PM (Aqua) followed in 2002

Monitors 36 spectral bands between 0.4 m and 14.4 m

Images entire Earth every 1-2 days at 1 km resolution

Why use MODIS?Why use MODIS?

ASTER and Landsat have 60 m resolution but available once a month at best

Geostationary satellites capture data with 15 min frequency but 5 km resolution

ASTER and Landsat have 60 m resolution but available once a month at best

Geostationary satellites capture data with 15 min frequency but 5 km resolution

Relevent MODIS Products

Relevent MODIS Products

MOD11 - Surface temperature and emissivity

MOD43 - Albedo

MOD15 - Leaf Area Index (LAI)

MOD13 - NDVI

Mod07 - Atmospheric stability; temperature and vapor pressure at 20 vertical levels

MOD03 - Lattitude, longitude, ground elevation, solar zenith angle, satellite zenith angle and azimuth angle

MOD11 - Surface temperature and emissivity

MOD43 - Albedo

MOD15 - Leaf Area Index (LAI)

MOD13 - NDVI

Mod07 - Atmospheric stability; temperature and vapor pressure at 20 vertical levels

MOD03 - Lattitude, longitude, ground elevation, solar zenith angle, satellite zenith angle and azimuth angle

NDVINDVI

NDVI RIR RredRIR Rred

First measured by the original Landsat in 1972

Measurement of a pixel’s “greenness”

Accessing MODIS DataAccessing MODIS Data

Level 1 and Atmosphere Archive and Distribution System (LAADS)

Warehouse Inventory Search Tool (WIST) submits orders via EOS ClearingHouse (ECHO)

HDF can interface with C, Fortran, Perl, MATLAB, IDL or Mathmatica

Level 1 and Atmosphere Archive and Distribution System (LAADS)

Warehouse Inventory Search Tool (WIST) submits orders via EOS ClearingHouse (ECHO)

HDF can interface with C, Fortran, Perl, MATLAB, IDL or Mathmatica

WISTWIST

Go Rd + Ld - s

sc

Go = Rn[c + (1-fc)(s - c)]

fc = percentage of ground covered by vegetation

= Measured by MODIS

= Variables

RnRd + Ld - sRnGo E

Surface Energy Balance System (Su 2002)

Surface Energy Balance System (Su 2002)

Calculating HCalculating H

= cannot be measured remotely

z0m and z0h z0m and z0h

Can vary by several orders of magnitude

Using LAI and wind speed, z0m can be calculated as afunction of canopy height following Massman (1997)

Zoh = zom/exp(kB-1) Wind speed

Limiting CasesLimiting Cases

Hdry = Rn - Go

Constraining the result between these values decreases the uncertainty considerably

Summary: Local Variables

Summary: Local Variables

Rd - Measured with a radiation sensor

Ld - Stephen-Boltzman equation using air temp

Wind speed and canopy height must be measured on site

Rd - Measured with a radiation sensor

Ld - Stephen-Boltzman equation using air temp

Wind speed and canopy height must be measured on site

ResultsResults

Triangle Method

(Jiang and Islam 2001)

Triangle Method

(Jiang and Islam 2001)

E (Rn G)

max

desdT TTa

f (Ta )

min 0

f (NDVI, soil moisture)

ResultsResults

Triangle Method Original Priestly-Taylor Eq

Complementary Model

(Venturini & Islam 2007)

Complementary Model

(Venturini & Islam 2007)

ET + ETpot = 2Etwet (Bouchet 1963)ET + ETpot = 2Etwet (Bouchet 1963)

EF = ET / (Rn-G)

From Priestly-Taylor

From Penman

Uses temp profile as surrogatefor humidity deficit

Benefits of Isolating EF

Benefits of Isolating EF

Rn is a large source of error because of atmospheric interference and cloud cover

Generally constant during daytime

Useful for mapping drought conditions

Rn is a large source of error because of atmospheric interference and cloud cover

Generally constant during daytime

Useful for mapping drought conditions

ResultsResults

Future ResearchFuture Research

Removing cloud-contaminaed pixels biases results, ignores diffuse radiation

Nocturnal transpiration

3°K error in in Ts causes 75% error in H

Removing cloud-contaminaed pixels biases results, ignores diffuse radiation

Nocturnal transpiration

3°K error in in Ts causes 75% error in H