ERS ATSR data land/atmosphere studies and applications

About ATSR

ERS ATSR in environmental hazards: fires, volcanic eruptions

ERS ATSR data use in earth resources studies

ERS ATSR data in climate studies

ERS ATSR in environmental hazards: fires, volcanic eruptions

Fire detection

Detection of burning biomass is very important for both climatic change assessment and earth resources monitoring. Burning biomass affects the CO₂ immission in the atmosphere, related to the greenhouse effect. Global fire monitoring can be carried out by satellite.

Detection of exceptional fires over South East Asia (August-October 1997)

A near real-time service has been set up in ESRIN for the monitoring of fires occurred over Borneo and Sumatra in the period August-October 1997. ATSR fire products were available after a couple of days. Additionally NO₂ total columns, supplied in GOME products, were available on the server because in concomitance with burning biomass. anomalous high NO₂ level can be observed. The ATSR fire product is a colour composite image based on the BTs derived from the 3.7 and 11 micron spectral channels. Night images are used because during the day the 3.7 micron spectral channel is saturated over hot surfaces (about 312 deg. K). Over night images saturated pixels are marked as hot spots. In the fire product, areas without fires appear in grey colour or dark whereas areas with fires appear yellowish and hot spot pixels are in red colour. Note that cloud cover is not derived.

(Click to see the images)

References

Arino O., J.M. Melinotte, J.M. Rosaz, and E. Monjoux, 1997, "ESA Fire Product", Proc. 7^th ISPRS Conference on Physical Measurement and Signatures in Remote Sensing, 7-11 April 1997, Courchevel, France. (Also submitted to Rem. Sensing of Environment)

Buongiorno A., "The ESA/ESRIN ATSR Browse Generation System (ABS)", Earth Observation Quarterly, 52, June 1996.

Monitoring of volcanoes

Active volcanoes located in populated areas are a hazard so have to be continuously monitored. Satellites allow for it thanks to repeated acquisitions. Volcanic activity causes high ground surface radiance emission that can be detected from satellite. Moreover lava effusion and lava flow can be detected and the volume of the lava can be estimated from the energy lost by the cooling lava. In this case a sequence of satellite images covering the period including the eruptive event is needed.

ATSR radiance data from the 1.6 and 11 micron spectral channels have been used by Woster and Rothery (3rd ERS Symposium, Florence, 1997) for studying some volcanoes during eruptive events.

In the case of Fernandina volcano (Galapagos), an estimate of the lava volume has been carried out during the eruption that occurred in the period January-April 1995. This estimate is comparable to that obtained by using SPOT imagery and a DEM.

In the case of Unzen volcano (Japan), a correlation between lava effusion rate and ATSR radiance, in the 1.6 micron spectral channel, has been found over the period 1992-1994, following the eruption of May 1991. From TM imagery and airborne thermal infrared imagery it has been found that the high radiance levels correspond to periods of high fumarolic activity which increasingly heats the Unzen dome surface.

In the case of Lascar volcano (Chile), an explosive eruption in April 1993, has been monitored by means of a selection of ATSR images, spanning from 1992 to 1995. TM imagery was also available and in-situ inspections were carried out. A decrease of the 1.6 micron spectral channel radiance has been observed before the explosive eruption, in agreement with TM observations. This suggests that the radiance value, at the 1.6 micron specral channel, could be used as an indicator for predicting eruptive events of this type of volcanoes.

ERS ATSR data in earth resources studies

Tropical forest monitoring

Tropical forests are an important part of terrestrial ecosystems. Tropical bioma accounts for 60% of the terrestrial net primary productivity and play a key role in CO₂ budget.

A Tropical Ecosystem Environment Observation by Satellite (TREES) project was started in the 1990 by the European Community jointly with ESA, aimed at carrying out tropical forest inventory. Initially it was based on NOAA AVHRR data; lately ERS ATSR data are being used for forest mapping updating as well as ERS SAR data.

Other studies

Additional vegetation studies, based on ATSR data, concern vegetation mapping of semiarid areas (Edwards et al., 3rd ERS Symposium, Florence, 1997). Further studies concern lake biota monitoring, respectively, of the Lake of Ontario (Canada) (Johnson et al, 3rd ERS Symposium, Florence, 1997), and of the Lake of Baikal (Siberia) (Le Core et al, 3rd ERS Symposium, Florence, 1997).

ERS ATSR data in climate studies

Cloud studies

Clouds play an important role in the global climate system as they affect global energy budget. In particular, tropical cirrus clouds are important as about 20% of tropical areas are covered permanently by such clouds. In order to estimate cloud energy balance some cloud parameters are needed. They can be derived from satellite data; satellites allow the monitoring of large cloudy areas also at very high height while aircraft measurements provide detailed analyses of limited cloudy areas.

In a study by Watts and Baran (3rd ERS Symposium, Florence, 1997) the derivation of cloud parameters from ATSR data has been investigated. A selection of images over an area of the Western Tropical Pacific has been used as well as in-situ measurements from the CEPEX (Central Equatorial Pacific Experiment, 1993/1994) experiment. Some of their findings are listed below.

The cloud top height can be derived by using ATSR reflectance from nadir/forward views in the 1.6 micron spectral channel. Matching algorithms are found in the literature.

The cloud top temperature can be derived from the Brightness Temperature (BT) relative to the 11and 12 micron spectral channels. Its derivation is restricted to those pixels having the 0.87 micron channel reflectance higher than a given threshold value.

The cloud phase (ice or water) can be discriminated on the basis of the 0.87 and 1.6 micron spectral channels reflectance.

The ice crystal shape (or habit) can be retrieved from the nadir/forward view reflectances relative to the 0.87 micron spectral channel. Ocean glitter can complicate the interpretation of the results in the case of thin clouds.

The ice crystal size can be derived from the 1.6 micron spectral channel and cloud optical depth can be derived from the 0.87 micron specral channel.

Comparisons with CEPEX measurements have been attempted. In comparing satellite and aircraft datasets care is needed. In fact satellites measure mainly cloud top properties whereas airplans carry out measurements below the top therefore cloud parameters refer to different sampling.

Aerosol studies

Stratospheric and desert aerosol retrieval over ocean

In addition to clouds, atmospheric aerosols play an important role in the global climate system. Some of their properties (e.g. type and concentration) can be retrieved from satellite measurements over ocean, in clear sky conditions. Aerosol type and concentration can be varied in the model so that the "simulated" radiances - at the top of the atmosphere - fit those measured from satellite.

In a study carried out by Dundas et al. (3rd ERS Symposium, Florence, 1997) ATSR radiance data as well as a radiative transfer model by the RAL (Rutherford Appleton Laboratory) have been used. Two types of aerosols have been studied: stratospheric and desert aerosols. Stratospheric aerosols are distributed uniformly in the region between 20-25 km whereas desert aerosols are confined in a region between 1-5 km (troposphere).

A diagram suitable for the retrieval of aerosols' has been found through model simulations. The diagram is based on the following expressions:

(1) [BT₁₁(nadir) - BT₁₁(forward)] versus (2) [BT₁₁(nadir) - BT₁₂(nadir)]

The subscripts 11 and 12 refer, respectively, to the 11 and 12 micron spectral channels. There is a linear relationship between (1) and (2).

Different aerosol concentrations correspond roughly to lines with similar slope in the diagram: the bottom line corresponds to the absence of aerosols. A diagram as such can be used, in principle, for retrieving aerosol concentration from cloud-free pixel radiances.

An ATSR global dataset as well as a selection of images over a North Atlantic area has been considered for aerosol retrieval. The global dataset, spanning from 1991 (after Mt. Pinabuto eruption) to 1993, has been chosen to explore the potential of the method for detecting the amount of stratospheric aerosols of volcanic origin. The regional dataset has been chosen for monitoring the desert dust in the troposphere.

By drawing a diagram as that mentioned above general consistence has been observed between satellite derived values and theoretical curves. In general, points resulting from the 1991 dataset lay in the higher portion of the diagram whereas points resulting from the 1992-1993 dataset are located in the lower part. This is consistent with the expected reduction of the volcanic aerosol, after the Pinabuto eruption.

Earth Surface Temperature retrieval from ATSR data

See also ERS ATSR Sea Surface Temperature (SST)

(Click to see ATSR SST picture)

In order to derive SST and Land Surface Temperature (LST) from satellite measurements an algorithm is needed due to the interfering effect of the atmosphere. In the derivation of LST a complication arises due to spatial and temporal variations of the ground surface emissivity.

Algorithms are based on a linear combination of Brightness Temperatures (BT) derived from different spectral channels (split-window) or else - as it is the case with the ATSR - from different views (dual-view), corresponding to different atmospheric paths. The coefficients of the linear combination can be derived statistically or by means of an atmospheric radiative transfer model. By adopting a statistical approach surface temperatures, as measured at the ground, are fitted to BTs measured from satellite. This is not a general approach because of the limited number of cases that is considered. By adopting a "modellistic" approach the BTs are "simulated" therefore a wide set of geophysical situations can be considered.

Sobrino et al. (1996) have implemented a dual-view algorithm by using a radiative transfer model (LOWTRAN 7). They have also derived a split-window algorithm, having a similar formulation, for comparisons. The dual-view algorithm is based on the ATSR 11 micron spectral channel whereas the split-window algorithm is based on the 11 and 12 micron spectral channels (nadir view). In the algorithms the ground surface emissivity and the atmospheric water vapour content are important parameters which can have constant or varying values. From preliminary analyses it has turned out that a water vapour estimate derived from BT measurements increases the accuracy of the results. Moreover the emissivity angular variations have to be taken into account especially in the case of land surfaces. The algorithms have been applied to two datasets, respectively, over sea and over land. The dataset over land refers to a flat homogeneous vegetated area for which LST retrieval is feasible. Computations have shown that the dual-view algorithm has accuracy comparable or better than the split-window algorithm.

An ERS ATSR Land Surface Temperature product is under implementation and should become available in the future months.

References

Sobrino, J.A., Z-L. Li, M.P. Stoll, and F. Becker, Multi-channel and multi-angle algorithms for estimating sea and land surface temperature with ATSR data, Int. J. Remote Sensing, 17, 11, 2089-2114, 1996.

Kerr Y.H., C. Guillou, J.P. Laguarde, F. Nerry, C. Ottlé and O. Arino (1998), Global land surface temperature retrieval from NOAA AVHRR data, XXIII General Assembly, European Geophysical Society, Nice 20-24 April.

Land surface studies

Land Surface Temperature (LST) studies

In the derivation of LST a complication arises due to spatial and temporal variations of the ground surface emissivity. LST retrieval is usually easier at night than during the day as at night it is less affected by the atmospheric conditions near the ground surface; on the other hand at night clouds are less effectively filtered out from satellite images.

Prata et al. (3rd ERS Symposium, Florence, 1997) have been carrying out studies on the retrieval LST from ATSR-1/2 data. In-situ measurements have been carried out at a spatial scale comparable to that of the ATSR instrument over test-sites belonging to the Australian Continental Integrated Ground-truth Site Network (CIGSN).

Two algorithms (dual-view) have been proposed, respectively, in the case of a vegetated surface and in the case of a bare soil. In the case of a bare soil the algorithm includes a model of the ground surface emissivity angular variation which is expected to affect the results. A split-window algorithm with fixed coefficients, implemented for AVHRR data, has also been used for comparison.

In general, LST biases and rms errors are higher in the case of a bare soil. Dual-view and split-window algorithms give comparable results. Highest biases are of about 0.5 deg. K and highest rms errors of about 2 deg. K.

(Typical LST rms figures obtained by using AVHRR data range between 1.5 and 4 deg. K)

Sea Surface Temperature (SST) in climatic databases

In climate modelling SST historical datasets are used. They are based on buoy temperature measurements, carried out at some depth, below the ocean surface (bulk SST), typically 1 m. On the other hand SST, as derived from ATSR, are representative of the top few microns of the ocean surface (skin SST). There have been attempts to convert ATSR SST (at 0.5 deg. Lat/Long spatial resolution, as supplied in ATSR products) to bulk SST (B. Candy, 3rd ERS Symposium, Florence, 1997).

ATSR visible/near infrared studies

Calibration of ATSR-2 visible channels using desert scenes

ERS ATSR-2 visible/near infrared reflectance measurements (calibrated) have been analyzed by Smith (3rd ERS Symposium, Florence, 1997), over an area of the Lybian desert, during a period of 18 months, in order to investigate the instrument stability.

The area has been selected because of its predominant clear sky conditions although no ground measurements and information about the atmospheric aerosol content are available. A 10-year record of AVHRR reflectance measurements has been considered for comparison. ATSR reflectance at 0.63 micron (AVHRR Ch1) and at 0.84 micron (AVHRR Ch2) has been worked out. ATSR and AVHRR Ch1 show comparable reflectance; on the other hand the observed discrepancy between ATSR and AVHRR Ch2, is attributed to the different characteristics of the two instruments in this spectral range. A set of ATSR images has been selected in order to filter out clouds, sand storms, and wet sand as well as some effects due to satellite and sun geometry. Yet a dependence on the scattering angle [based on the sun and the satellite viewing geometry] is observed. A model for taking into account for the scattering angle dependence has been worked out then the ATSR-2 instrumental drifts over 18-months have been derived. Drifts of the different spectral channels span from 0.005 to 0.018 (year^-1).

Note that, for a given scattering angle, comparable reflectance values, both, in the nadir and in the along track views are observed. By considering that these values refer to different times, actually to different seasons, it can be argued that the observed scene is stable over the considered time interval.

[Remark. The conversion from ATSR-2 digital countings to reflectance is carried out by means of a look-up table available at the RAL website.]

Land surface shortwave irradiance from ATSR data

Among the validation studies carried out by Prata et al. (3rd ERS Symposium, Florence, 1997) there are studies of the ERS ATSR 1.6 micron spectral channel (near infrared). A very good linear correlation (r=0.97) has been found between shortwave (broadband) ground surface irradiance, measured from radiometer mounted on a tower, and ATSR coincident measurements of earth radiance at 1.6 micron. This allows an estimate of the ground surface shortwave (clear sky) albedo from satellite. A relationship as such has been derived also from tests based on the LOWTRAN-7 radiative transfer model. This is an indirect confirmation of ATSR 1.6 micron spectral channel performance.

About ATSR

ERS ATSR data geophysical information overview

ATSR SST product

ATSR SST validation results

ATSR instrument characteristics

(Document prepared by: M.G. Scarpino - Eurimage S.p.A. c/o ESA/ESRIN ERS Data Utilization Section APP-AED)

Keywords: ESA European Space Agency - Agence spatiale europeenne, observation de la terre, earth observation, satellite remote sensing, teledetection, geophysique, altimetrie, radar, chimique atmospherique, geophysics, altimetry, radar, atmospheric chemistry