developing a spectral database of various land...

05/22/2012 TEMPO Science Team Meeting, Hampton, VA May 21-22, 2014

Developing a spectral database of various land cover types to characterize land

surface albedo for TEMPO Wasit Wulamua, Jack Fishmana, Kelly Chanceb

aSaint Louis University, St. Louis, MO 63103 bHarvard-Smithsonian Center for Astrophysics, Cambridge, MA, USA 02138

2 TEMPO Science Team Meeting, Hampton, VA May 21-22, 2014

Albedo/BRDF

!  Land surface albedo/BRDF (Bi-directional reflectance function) ! An important variable that characterizes solar

illumination changes, vegetation growth and agricultural practices

!  Provides useful Chappuis-band contributions for the retrieval of O3 and chlorophyll fluorescence, e.g.

Surface reflectance must be known

chlorophyll fluorescence (Joiner, et al 2013)


Methods for albedo estimation !  Physical approaches based of BRDF,

!  e.g., MODIS snow-free albedo products !  BRDF is modeled by atmospheric correction and estimation of aerosol optical depth,

transmittance, etc !  Difficult to translate albedo from MODIS (0.5 – 1km) or other scales to TEMPO scale (4 km

* 4 km) due to the lack of homogeneous pixels at coarser resolution. !  Direct estimation - data driven approach

!  Directly links broadband albedo to satellite observed TOA reflectance !  Cannot estimate spectral albedos/BRDF

!  Band reconstruction methods by fitting the satellite-derived BRDF !  BRDF database in a sensor (e.g., VIIRS, RTTOV) is derived from MODIS BRDF/albedo

(Wang et al. 2013; Vidot et al. 2014) !  A conversion coefficient from field spectra to MODIS and then to VIIRS and RTTOV is

estimated using USGS and ASTER spectral library !  Limitations

"  Requires representative sets of hyperspectral spectra typically found in the landscape at different spatial and temporal scales

"  USGS and ASTER spectral data are mostly measured in the laboratory "  Does not provide sufficient water, snow and ice reflectance "  Assumes that there are some homogeneous pixels with coarser spatial resolution that correspond to

the finer resolution data for each of the land cover (impact of scale is a problem!)


Ongoing efforts for albedo estimations

!  Harvard Smithsonian ! Band reconstruction method from Field-MODIS-TEMPO

" Representative spectra found in Midwest is needed " Not just lab spectra but albedo/BRDF measured in the field,

airborne

!  Saint Louis University ! Developing a database of spectral albedo/BRDF ! Field-airborne data collection, more realistically reflect

the land-cover and land-use !  Impact of scale when spectra are converted from field

to airborne, and then to MODIS and TEMPO


Spectral profiles of several instruments

MODTRAN PSR-3500 Specim Camera

AVIRIS TEMPO PANDORA

Wavelength range

0.2-100 um 0.35-2.5 um 0.35- 2.5 um 0.4-2.5um 0.29-0.69 um

0.29-0.52 um

Spectral Resolution

> 0.1 cm-1

(0.08 nm @ 400 nm)

3nm @700nm 10nm @ 1500nm 7nm @ 2100nm

1.5nm@ VNIR 12nm@SWIR

10 nm 0.6 nm 0.6 nm

Spectral Sampling

1.5nm @700nm 3.8nm @ 1500nm 2.5nm @ 2100nm

~@ VNIR 5.6nm@SWIR

1 nm 0.2 nm


Test Site - MoBOT 6

Drought bed Controlled bed

Vitis riparia (river bank grape); Vitis rupestris (sand grape ) Nine genotypes


Leaf and Canopy spectra – different


Map modified from Price, D. T., McKenney, D. W., Papadopol, P., Logan, T. & Hutchinson, M. F. (26th Conference on Agricultural and Forest

Meteorology, 2004).

Rare earth minerals - the AVD

Examples of diatreme outcrops sampled in the AVD. Variation in mineralogy and resistance to weathering dictates the outcrop profile spanning preferentially resistant (left and right) to laterized.

At top right is the spectroradiometer set up for laboratory spectral measurements. Below the portable spectroradiometer being used to take VIS-NIR reflectance measurements of an AVD outcrop.

A representative group of spectra from various rare earth related minerals in Missouri (measured in the lab)

spec

tral a

lbed

o/B

RD

F

wavelength (nm)


Laboratory and field measured spectral albedo/BRDF from AVD samples displaying different VIS-NIR absorption features.

Lab and field spectra - different

Highlights the impact of scale Note: most of USGS spectra and ASTER library were measured in the lab


We need to know the scale effect 13

A very large hall?


We need to know the scale effect 14

Compared to what?


AmeriFlux network

!  https://www.dropbox.com/s/yvofvsnpdqbs3im/Screenshot%202014-05-08%2022.22.27.png


AVIRIS data availability


AVIRIS spectral albedo


Future work – multi sensor fusion

Field (0.13m res.) Mobile lab (0.5 m res.)

Airborne (AVIRIS, Ball, SLU?) 4 – 20m res.

Tempo (4km res.)

Numeric simulations


Future work – multi sensor fusion

Kalman filter

Defining'a'slope'across'the'Chappuis'core'region:''(is$just$a$slope$enough?)$ 2R.Chatfield,$NASA$Ames'

'4

These are simple ratios of upward to downward: all low altitude (0-500 m), minimal aerosol effect

Higher$Albedo,2Vegetation$+$$2Soil/Road/DrA(λ)2

slope

Chappuis(“horn”4

Chappuis(“horn”4

Strong0H2O0line4

Strong0H2O0line4

MODIS'line'

centers4 Real0Curvature0Here0at05500nm4

Huge%Credit%to%,S.<Schmidt/Pilewskie0group0at0CU0Boulder,00esp.0Bruce0Kindel(

Solar'spectrum'flux'radiometer'(CU@Ames)'provides'the'best@understood,'most'direct'evidence'about'

spectral'shape4“Half$of$SSFRGs$signal$(not$an$imager$but$irradiance$radiometer)$comes$from$within$a$cone$that$has$a$ground$crossJsection$about$the$same$as$the$flight$altitude”$Sebastian$Schmidt2

Flight$path$north$of$Houston,$2006,$mostly$vegetated2

Credit%to%,S.<Schmidt0/Pilewskie0group0at0CU0Boulder,00esp.0Bruce0Kindel(

Examples'of'naïve'albedo'from'SSFR'data:''ratio'of'upward'flux'to'downward'flux:'this'spotlights'our'

concern:'λ@variation4

These are simple ratios of upward to downward: all low altitude (0-500 m), minimal aerosol effect

Vegetation$A(λ)22What$happens$with$drought,$Ozone$damage?2Yellowing?2

Soil/Road/Dry$A(λ)2 ?$Red$Soil$?$2

Slopes'can'change'very'rapidly'over'<'6'km,'but'there'are'also'broad'regional'paPerns4

Variogram shows strong increase of variability of slope up to 10 km, …then similar variability up to 40 km

Higher$Albedo,2Vegetation$+$$2Soil/Road$$2

$Δ A(λ)$/ Δ λ2

slope

Map of slopes, unitless/(1000 nm) August/ September 2013

'Analysis'Overview4

General'direction'of'tasks:'4•'Identify'“types”'and'“mixtures”'using'aircraft'data'to'check'and'extend'“pure@land@use”'(lab)'albedo.'What%might%we%be%missing?'4•'Chose'“components”'method,'paying'aPention'to'advice'from'MODIS'land@use'scientists.4•'Produce'sample'dA/dλ'descriptions'with'2–3'degrees'of'freedom'per'description,'or'as'TEMPO'instrument'error'covariances'mandate'(do'small'10@nm'features'maPer?)44MAIAC:Repeat'analysis'Albedos'in'MODIS'bands'at'0.5–1'km'resolution,'with'careful'multi@footprint'filtering'and'rationalization'(in'time'and'space'domains).'Lyapustin'data'will'be'available'to'GEO@CAPE'(Aerosol)'group'by'late'March.44R.Chatfield,[email protected](44

ADDITIONAL'SLIDES4

Usefulness$of$Some$RT$“Modeling”2

SSFR and

spectrum fitting

Houston NNW

flight analysis,

2900 m

at 19.19 UT

H2O to zenith

should be minimal

for samples in this

dataset.

Measured Downward

(zenith) X 0.1 to fit scale

Measured upwards

(nadir) (red dots)

and “shift-Chappuis fit”

Chappuis spectrum

Water spectrum (VPL plot)

Downward – best-fit upward shifted +0.02 to fit scale

(corresponds to

~0.10 to ~0.20 “albedo” in dataset ),

Extra slides


G. Schaepman-Strub et al. / Remote Sensing of Environment 103 (2006) 27–42

The above reflectance and reflectance factor definitions leadto the following special cases:

• ωi orωr are omitted when either is zero (directional quantities).• If 0b (ωi or ωr)b2π, then θ,ϕ describe the direction of thecenter axis of the cone (e.g. the line from a sensor to thecenter of its ground field of view—conical quantities).

• If ωi =2π, the angles θi,ϕi indicate the direction of theincoming direct radiation (e.g., the position of the sun). Forremote sensing applications, it is often useful to separate thenatural incoming radiation into a direct (neglecting the sun'ssize) and hemispherical diffuse part. One may also include aterrain reflected diffuse component that is calculated with atopographic radiation model such as TOPORAD (Dozier,1980). Consequently, the preferred notation for the geometryof the incoming radiation under ambient illuminationconditions is θi,ϕi,2π. Note that in this case, θi,ϕi describethe position of the sun and not the center of the cone (2π),except if the sun's position is at nadir. In the case of anisotropic diffuse irradiance field, without any directirradiance component (closest approximated in the case ofan optically thick cloud deck), θi,ϕi are omitted.

• If ωr =2π, θr and ϕr are omitted.

Finally, according to Nicodemus et al. (1977), the angularcharacteristics of the incoming radiance are named first in theterm, followed by the angular characteristics of the reflectedradiance. This leads to the attributes of radiance and reflectancequantities as illustrated in Table 2.

It should be noted that the nine standard reflectance termsdefined by Nicodemus et al. (1977) “are applicable only tosituations with uniform and isotropic radiation throughout theincident beam of radiation”. They then state that “If this is nottrue, then one must refer to the more general expressions”. Thisimplies that any significant change to the nine reflectanceconcepts when the incident radiance is anisotropic lies in themathematical expression used in their definition. Based on thisimplication, Martonchik et al. (2000) adapted the terminologyand reflectance names to the remote sensing case, whichinvolves direct and diffuse sky illumination. In the following,we give the mathematical description of the most commonlyused quantities in remote sensing, thus the general expressionsfor non-isotropic incident radiation. When applicable, wesimplify the expression for the special case of isotropic incidentradiation. The nine possible combinations of beam geometriesof the incident and reflected radiant fluxes are indicated asCases 1 to 9, corresponding to the illustrations in Table 2.

2.2.1. The bidirectional reflectance distribution function(BRDF)—Case 1

The bidirectional reflectance distribution function(BRDF) describes the scattering of a parallel beam of incidentlight from one direction in the hemisphere into another directionin the hemisphere. The term BRDF was first used in theliterature in the early 1960s (Nicodemus, 1965). Beingexpressed as the ratio of infinitesimal quantities, it cannot bedirectly measured (Nicodemus et al., 1977). The BRDFdescribes the intrinsic reflectance properties of a surface and

Table 2Relation of incoming and reflected radiance terminology used to describe reflectance quantities

Incoming/Reflected Directional Conical

Directional BidirectionalCASE 1

Directional–conicalCASE 2

Conical

Hemispherical

Conical–directionalCASE 4

Hemispherical–directionalCASE 7

Hemispherical–conicalCASE 8

BiconicalCASE 5

Hemispherical

Directional–hemisphericalCASE 3

Conical–hemisphericalCASE 6

BihemisphericalCASE 9

The labeling with ‘Case’ corresponds to the nomenclature of Nicodemus et al. (1977). Grey fields correspond to measurable quantities (Cases 5, 8), the others(Cases 1–4, 6, 7, 9) denote conceptual quantities. Please refer to the text for the explanation on measurable and conceptual quantities.

30 G. Schaepman-Strub et al. / Remote Sensing of Environment 103 (2006) 27–42

developing a spectral database of various land...

Documents