Validation of ASTER and MODIS Surface-Temperature and
Vegetation Products with Surface-Flux Applications


Participants:
S. Tom Gower
John M. Norman
George R. Diak
Kenneth J. Davis

Abstract:
The objective of this proposed work is to provide a set of verification and supporting measurements for ASTER land-surface temperature and emissivity and MODIS surface temperatures and vegetation-related products. The ASTER/MODIS products to be directly validated in this program are: Land surface temperature (LST) and emissivity at the scale of ASTER and MODIS pixels; leaf area index (LAI); fraction intercepted photosynthetically active radiation (FIPAR); above-ground net primary production (ANPP) and the net ecosystem CO2 exchange (NEE); snow and land cover. The tasks will be accomplished by enhancing the measurement capabilities of an existing Ameriflux tower situated in northern Wisconsin and also by making detailed complementary in-situ measurements on scales ranging from 14,000 km2 to the flux footprint of the tower, to the pixel sizes of ASTER and MODIS. Insolation derived from GOES-Visible satellite measurements (Diak et al. 1996) will provide the input to calculate FIPAR over the flux footprint of the tower. Additionally, we will compare flux measurements made by tower-based instrumentation to values obtained from a calibrated remote sensing method and to those derived from vegetation characteristics, estimated both from in-situ measurements and remote-sensing data.

Introduction and Objectives:
The objective of this proposed work is to provide a unique and cost-effective set of verification and supporting measurements for ASTER land-surface temperature and emissivity and MODIS surface temperatures and vegetation-related products. The ASTER/MODIS products to be directly validated in this program are: Land surface temperature (LST) and emissivity at the scale of ASTER and MODIS pixels; leaf area index (LAI); fraction intercepted photosynthetically active radiation (FIPAR); above-ground net primary production (ANPP) and the net ecosystem CO2 exchange (NEE); snow and land cover. The 450-m Park Falls tower is positioned in the approximate center of a 360 m diameter circle of grass. By mounting an infrared thermometers (15 degree field of view Everest Interscience Model 4000) at the 400-m level of the tower, the brightness temperature of an ASTER-sized pixel area of grass can be monitored at near-nadir angles (<10 degrees). The infrared thermometer (IRT) is sensitive to the 8 - 14 m wavelength band while the ASTER surface temperature channels are 10.25-10.95 and 10.95-11.65 m. The spectral, directional emissivity of the surface will be measured at ground level with a wavenumber resolution of one cm-1 between 8 and 14 m, so that surface directional emissivities will be available for each of the relevant thermal bands to define accurate surface directional radiometric temperatures. Minor atmospheric corrections (< 0.5 C) will be made on the IRT's placed at the 400-m level using the temperature and water vapor profiles measured on the tower. Surface temperature measurements on a large nearby lake (Blockhouse Lake within 5 km distance, diameter ~ 1 km) will provide a second surface of different and known temperature and emissivity to permit improved determination of atmospheric corrections. Having verified the ASTER emissivity algorithms over the grass area near the Park Falls tower, we would trust these emissivity algorithms over the forested areas and thus convert brightness temperatures from ASTER to radiometric temperatures from ASTER for these forested areas. Aggregating these ASTER radiometric temperatures over MODIS pixels would provide a check on the accuracy of MODIS surface temperatures at near-nadir view angles. The approach we propose to use to validate MODIS vegetation products involves two activities: a) Intensive in situ measurements of LAI, FIPAR, NPP and vegetative cover on a 2.5 km by 2.5 km area (about 4 MODIS pixels) centered on the Park Falls tower, and b) extensive in-situ measurements spread over a five-county area in northern Wisconsin near the Park Falls tower based on 46 sampling locations that represent six different forest types over an area of 14,000 km2. The above tasks will be accomplished by enhancing the measurement capabilities of an existing Ameriflux tower situated in northern Wisconsin and also by making detailed complementary in-situ measurements on scales ranging from 14,000 km2 to the flux footprint of the tower, to the pixel sizes of ASTER and MODIS. Additionally, we will compare flux measurements made by tower-based instrumentation to values obtained from a calibrated remote sensing method and to those derived from vegetation characteristics, estimated both from in-situ measurements and remote-sensing data.

Field Activities:

Location
The approach we propose to use involves two activities:

a) Intensive in situ measurements of LAI, FIPAR, NPP and vegetative cover on a 2.5 km by 2.5 km area (about 4 MODIS pixels) centered on the Park Falls tower, and

b) extensive in-situ measurements spread over a five-county area in northern Wisconsin near the Park Falls tower based on 46 sampling locations that represent six different upland second and old-growth forest types over an area of 14,000 km2.

Measurements:
Numerous measurements will be made in the area surrounding the WLEF tall tower (Table 1). At the intensive site centered on the tower, we will use an x, y grid to establish plot centers that will be used to characterize vegetation cover, LAI, FIPAR, and aboveground net primary production (NPPA). Plots will be established every 100 m along 2.5 km east-west transect which will be positioned 200 m apart for 1.25 km on the north and south side of the tall tower. Plot location will be established using two Professional series GPS units. We will use a nested sampling design where extensive, easily measured parameters (vegetation cover, LAI, and FIPAR) will be measured every 100 meters along each east-west transect (intensive grid-100) and more labor intensive measurements (i.e. NPPA) will be measured every 200 meters along each east-west transect (intensive grid-200) and extrapolated to all the plots using algorithms. NPPA is positively correlated (n = 24, r2 = 0.75, p < 0.001) to LAI for forests near the study area (Fassnacht and Gower 1997). Belowground NPP will be measured in replicated northern hardwood, upland pine and lowland black spruce forests in a separate study. Emissivity measurements of the grass circle surrounding the tower will be made using surface-based interferometry, once early in the growing season, when the surface is virtually bare, a second time when the grass is mature and a third time after the grass has senesced.

Table 1: Summary of Proposed Field Measurements

Field measurements Time period Location
Vegetation cover 7/98, 4/99, 7/99 intensive grid-100
HSAI (LAI) 7/98, 4/99, 7/99 intensive grid-100
FIPAR 7/98, 4/99, 7/99 intensive grid-100
Biomass 7/98, 7/99 intensive grid-100
Aboveground NPP 7/98, 7/99 intensive grid-200
Belowground NPP 4/99 - 10/00 NH, UP, BS
Emissivity 5/99, 7/99, 10/99 grass
Sun photometer 1998 - 2000 WLEF tower
Snow cover 1999 - 2000 WLEF tower

HSAI (LAI) = hemi-surface area index which equals total leaf surface area
FIPAR = fraction intercepted photosynthetically active radiation
NPP = net primary production
NH = northern hardwood, UP = upland pine, BS = black spruce


Table 2: Summary of Remotely Sensed Data

Remotely Sensed Measurements Time period Location
ASTER every pass (+/- 16 days) WLEF tower + 20km
MODIS ????  
GOES-8 visible 8-12 per day (May - Oct. 98 - 00) WI
GOES-8 thermalclear mornings (May - Oct. 1999) WI
AVHRR bi-weekly composited NDVI WI


Time Table (Project duration three years)
Comparison of ground-based radiometric temperature and directional emissivity on several ASTER pixels for all acceptably clear ASTER images will be performed for the last two project years.

Comparison of MODIS biophysical properties with independent ground-based measurements at the intensive four-pixel site centered on the tower will be done.

Comparison between MODIS and in situ NPPA measurements will be done on an annual basis for the last two project years. Comparison between MODIS and in situ measurements of LAI, FIPAR will be done at least two times per year (leaf-on and leaf-off) for the last two project years. Belowground net primary production (NPPB) will be estimated for a subset of plots in a companion study (McIntire-Stennis-Gower and Mackay).

Comparison of remotely sensed radiometric temperature between aggregated ASTER pixels and MODIS pixels using emissivity estimates derived from ASTER for all acceptably clear ASTER images for last two project years. We will provide NPPA, NPPB, heterotrophic respiration (RH) and NEE estimates for the footprint of the tall tower for the last two project years to compare with fluxes of CO2 from the tower.

Comparison of the spatial distribution MODIS biophysical properties with independent ground-based measurements at the extensive, five-county site during mid-growing-season will be made for the last two project years.


Satellite and Meteorological Data:
MODIS (primary data and EOS products)
ASTER (primary data and EOS products)
GOES (visible and thermal infrared)
Landsat TM
Surface and upper-air meteorological data, meteorological analyses and forecasts

Derived Products:
Land surface temperature (LST) and emissivity at the scale of ASTER and MODIS pixels;
Leaf Area Index (LAI);
Fraction Intercepted Photosynthetically Active Radiation (FIPAR);
Aboveground Net Primary Production (NPPA) and the net ecosystem CO2
exchange (NEE);
Snow and land cover


Modeling Activities:
In addition to providing flux validation for land-surface process models, the Park Falls tower will also serve as a "calibration target" for flux diagnoses which we will make using an established remote sensing method (Anderson et al. 1997), that has already been extensively tested against surface-based flux measurements. This method uses the time-changes of satellite measured brightness temperatures (from GOES), vegetation indices derived from remote sensing spectral measurements (MODIS) and a modest amount of supporting meteorological and surface information to evaluate the land-surface fluxes water and heat and potentially carbon uptake by vegetation.


Relevant Publications:

Anderson, M. C., J. M. Norman, G. R. Diak and W. P. Kustas, 1997. A two-source time integrated model for estimating surface fluxes for thermal infrared satellite observations. Rem. Sens. Environ., 60, 195-216.

Chen, J., P. W. Rich, S. T. Gower, J. M. Norman and S. Plummer, 1997. Leaf area index of boreal forests: Theory, techniques, and measurements. J. Geophys. Res., 102 (D24), 29,429-29,443.

Diak, G. R., W. L. Bland and J. R. Mecikalski, 1996. A note on first estimates of surface insolation from GOES-8 visible satellite data. Agr. For. Meteor., 82, 219-226.

Fassnacht, K.S., S.T. Gower, M.D. MacKenzie, E.V. Nordheim, T.M. Lillesand. 1997. Estimating the leaf area of north central Wisconsin forests using the Landsat Thematic Mapper. Remote Sens. Environ. 61, 229-245.

Fassnacht, K. S. and S. T. Gower, 1997. Interrelationships among the edaphic and stand characteristics, leaf area index, and aboveground net primary production of upland forest ecosystems in north central Wisconsin. Can. J. For. Res. 27: 1058-1067.