• Blumstein, D., Tournier B. , Cayla F. R. , Phulpin T. , Fjortoft R. , Buel C. , and Ponce G. , 2007: In-flight performance of the infrared atmospheric sounding interferometer (IASI) on METOP-A. Atmospheric and Environmental Remote Sensing Data Processing and Utilization III: Readiness for GEOSS, M. D. Goldberg et al., Eds., International Society for Optical Engineering (SPIE Proceedings, Vol. 6684), DOI:10.1117/12.734162.

    • Search Google Scholar
    • Export Citation
  • Chahine, M. T., and Coauthors, 2006: AIRS: Improving weather forecasting and providing new data on greenhouse gases. Bull. Amer. Meteor. Soc., 87 , 911926.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Clough, S. A., and Iacono M. J. , 1995: Line-by-line calculation of atmospheric fluxes and cooling rates II: Application to carbon dioxide, ozone, methane, nitrous oxide and the halocarbons. J. Geophys. Res., 100 , 1651916535.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Goldberg, M. D., 2007: Global space-based inter-calibration system (GSICS). Atmospheric and Environmental Remote Sensing Data Processing and Utilization III: Readiness for GEOSS, M. D. Goldberg et al., Eds., International Society for Optical Engineering (SPIE Proceedings, Vol. 6684), DOI:10.1117/12.735246.

    • Search Google Scholar
    • Export Citation
  • Gunshor, M. M., Schmit T. J. , and Menzel W. P. , 2004: Intercalibration of the infrared window and water vapor channels on operational geostationary environmental satellites using a single polar-orbiting satellite. J. Atmos. Oceanic Technol., 21 , 6168.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Gunshor, M. M., Schmit T. J. , Menzel W. P. , and Tobin D. , 2007: Intercalibrating geostationary imagers via polar orbiting high spectral resolution data. Atmospheric and Environmental Remote Sensing Data Processing and Utilization III: Readiness for GEOSS, M. D. Goldberg et al., Eds., International Society for Optical Engineering (SPIE Proceedings, Vol. 6684), DOI:10.1117/12.735855.

    • Search Google Scholar
    • Export Citation
  • Guo, Q., Xu J. , and Zhang W. , 2005: Stray light modelling and analysis for the FY-2 meteorological satellite. Int. J. Remote Sens., 26 , 28172830.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Hillger, D. W., and Schmit T. J. , 2007: The GOES-13 science test: Imager and sounder radiance and product validations. NOAA Tech. Rep. NESDIS 125, 75 pp.

    • Search Google Scholar
    • Export Citation
  • Puschell, J. J., and Coauthors, 2006: In-flight performance of the Japanese Advanced Meteorological Imager. Earth Observing Systems XI, J. J. Butler, Ed., International Society for Optical Engineering (SPIE Proceedings, Vol. 6296), DOI:10.1117/12.683505.

    • Search Google Scholar
    • Export Citation
  • Schmit, T. J., and Herman L. D. , 1992: Comparison of multi-spectral data: GOES-7 (VAS) to NOAA-11 (HIRS). Preprints, Sixth Conf. on Satellite Meteorology and Oceanography, Atlanta, GA, Amer. Meteor. Soc., 258–259.

  • Smith, W. L., Woolf H. M. , Abel P. G. , Hayden C. M. , Chalfant M. , and Grody N. , 1974: NIMBUS-5 Sounder Data Processing System. Part I: Measurement characteristics and data reduction procedures. NOAA Tech. Memo. National Environmental Satellite Service (NESS) 57, 99 pp.

    • Search Google Scholar
    • Export Citation
  • Tjemkes, S. A., König M. , Lutz H-J. , van de Berg L. , and Schmetz J. , 2001: Calibration of Meteosat water vapor channel observations with independent satellite observations. J. Geophys. Res., 106 , 51995209.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Tobin, D. C., Revercomb H. E. , Moeller C. C. , and Pagano T. S. , 2006a: Use of Atmospheric Infrared Sounder high-spectral resolution spectra to assess the calibration to Moderate resolution Imaging Spectroradiometer on EOS Aqua. J. Geophys. Res., 111 , D09S05. doi:10.1029/2005JD006095.

    • Search Google Scholar
    • Export Citation
  • Tobin, D. C., and Coauthors, 2006b: Radiometric and spectral validation of Atmospheric Infrared Sounder observations with the aircraft-based Scanning High-Resolution Interferometer Sounder. J. Geophys. Res., 111 , D09S02. doi:10.1029/2005JD006094.

    • Search Google Scholar
    • Export Citation
  • Wang, L., Cao C. , and Ciren P. , 2007: Assessing NOAA-16 HIRS radiance accuracy using simultaneous nadir overpass observations from AIRS. J. Atmos. Oceanic Technol., 24 , 15461561.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Wanzong, S., Menzel W. P. , and Callan G. , 1998: Intercalibration of GOES, Meteosat, GMS, HIRS, and AVHRR infrared radiances. Preprints, Ninth Conf. on Satellite Meteorology and Oceanography, Paris, France, Amer. Meteor. Soc., 714–717.

  • Weinreb, M. P., Jamison M. , Fulton N. , Chen Y. , Johnson J. X. , Bremer J. , Smith C. , and Baucom J. , 1997: Operational calibration of Geostationary Operational Environmental Satellite-8 and -9 imagers and sounders. Appl. Opt., 36 , 68956904.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • View in gallery

    Sample tropical spectrum (black) with spectral gaps filled with the adjusted USSA (gray).

  • View in gallery

    GOES-12 imager (green) and Meteosat-8 (blue) SRFs shown with the brightness temperature difference ΔTbb of AIRS real SRFs convolved with the USSA spectrum subtracted from AIRS mock SRFs convolved with the USSA spectrum for each AIRS channel. (right to left) GOES bands are 2, 3, 4, and 6 and Meteostat-8 bands are 4–11.

  • View in gallery

    The CIMSS-32, convolved with mock AIRS SRFs and then gap filled with the adjusted USSA, were then convolved with GEO SRFs and differenced with the CIMSS-32 convolved with the GEO SRFs. The GOES-12 IRW (circles), GOES-12 water vapor band (squares), and the Meteosat-8 shortwave window (asterisks) are shown. The differences are less than 0.1 K for all 32 atmospheres for all the bands except for the water vapor bands and the Meteosat 8.7-μm band (not shown).

  • View in gallery

    GOES-13 imager band 6 (13.3 μm) SRF (blue) and the shifted SRF (green) shifted −4.7 cm−1 (to approximately 13.4 μm), with a representative brightness temperature Tbb spectrum in the background (red). By shifting the spectral response this amount, the bias, or mean Tbb difference for all 19 cases, becomes 0.0 K (rounded from 0.01 K) with a std dev of 0.7 K.

  • View in gallery

    Time series of GOES-12 water vapor band differences with AIRS, illustrating the effect of decontamination.

All Time Past Year Past 30 Days
Abstract Views 0 0 0
Full Text Views 96 82 4
PDF Downloads 38 28 2

Intercalibration of Broadband Geostationary Imagers Using AIRS

View More View Less
  • 1 Cooperative Institute for Meteorological Satellite Studies, University of Wisconsin—Madison, Madison, Wisconsin
  • | 2 NOAA/NESDIS, Center for Satellite Applications and Research, Advanced Satellite Products Branch, Madison, Wisconsin
  • | 3 Cooperative Institute for Meteorological Satellite Studies, University of Wisconsin—Madison, Madison, Wisconsin
© Get Permissions
Full access

Abstract

Geostationary simultaneous nadir observations (GSNOs) are collected for Earth Observing System (EOS) Atmospheric Infrared Sounder (AIRS) on board Aqua and a global array of geostationary imagers. The imagers compared in this study are on (Geostationary Operational Environmental Satellites) GOES-10, GOES-11, GOES-12, (Meteorological Satellites) Meteosat-8, Meteosat-9, Multifunctional Transport Satellite-IR (MTSAT-IR), and Fenguyun-2C (FY-2C). It has been shown that a single polar-orbiting satellite can be used to intercalibrate any number of geostationary imagers. Using a high-spectral-resolution infrared sensor, in this case AIRS, brings this method closer to an absolute reckoning of imager calibration accuracy based on laboratory measurements of the instrument’s spectral response. An intercalibration method is presented here, including a method of compensating for AIRS’ spectral gaps, along with results for approximately 22 months of comparisons. The method appears to work very well for most bands, but there are still unresolved issues with bands that are not spectrally covered well by AIRS (such as the water vapor bands and the 8.7-μm band on Meteosat). To the first approximation, most of the bands on the world’s geostationary imagers are reasonably well calibrated—that is, they compare to within 1 K of a standard reference (AIRS). The next step in the evolution of geostationary intercalibration is to use Infrared Atmospheric Sounding Interferometer (IASI) data. IASI is a high-spectral-resolution instrument similar to AIRS but without significant spectral gaps.

Corresponding author address: Mathew M. Gunshor, CIMSS, 1225 W. Dayton St., Madison, WI 53706. Email: matg@ssec.wisc.edu

Abstract

Geostationary simultaneous nadir observations (GSNOs) are collected for Earth Observing System (EOS) Atmospheric Infrared Sounder (AIRS) on board Aqua and a global array of geostationary imagers. The imagers compared in this study are on (Geostationary Operational Environmental Satellites) GOES-10, GOES-11, GOES-12, (Meteorological Satellites) Meteosat-8, Meteosat-9, Multifunctional Transport Satellite-IR (MTSAT-IR), and Fenguyun-2C (FY-2C). It has been shown that a single polar-orbiting satellite can be used to intercalibrate any number of geostationary imagers. Using a high-spectral-resolution infrared sensor, in this case AIRS, brings this method closer to an absolute reckoning of imager calibration accuracy based on laboratory measurements of the instrument’s spectral response. An intercalibration method is presented here, including a method of compensating for AIRS’ spectral gaps, along with results for approximately 22 months of comparisons. The method appears to work very well for most bands, but there are still unresolved issues with bands that are not spectrally covered well by AIRS (such as the water vapor bands and the 8.7-μm band on Meteosat). To the first approximation, most of the bands on the world’s geostationary imagers are reasonably well calibrated—that is, they compare to within 1 K of a standard reference (AIRS). The next step in the evolution of geostationary intercalibration is to use Infrared Atmospheric Sounding Interferometer (IASI) data. IASI is a high-spectral-resolution instrument similar to AIRS but without significant spectral gaps.

Corresponding author address: Mathew M. Gunshor, CIMSS, 1225 W. Dayton St., Madison, WI 53706. Email: matg@ssec.wisc.edu

1. Introduction

Intercalibration of satellite sensors is needed to improve the use of space-based global observations for weather, climate, and environmental applications. There is a desire to use satellites, traditionally used to enhance weather forecasting, for global climate monitoring and other applications related to the environment. Satellites are also playing a larger role in numerical weather prediction (NWP;Chahine et al. 2006). The purpose of intercalibration is to quantitatively relate the radiances from different sensors viewing the same target, which provides valuable information for a wide range of satellite activities.

There are numerous quantitative products, many produced operationally, which are made at least in part using either polar or geostationary satellite radiances. Application areas include weather, climate, hazards, land, ocean, and the cryosphere. Absolute validation is difficult, because a method to measure earth-emitted radiances precisely from a broadband instrument does not exist. Differences in satellite view angle, scan time, field of view (FOV) size and shape, and navigation accuracy all contribute to noncalibration differences between measured radiances. In addition, broadband imagers have different spectral responses, differing even between detectors for the same band on the same instrument. Past studies intercalibrating two broadband instruments required accounting for spectral response function (SRF) differences with the use of forward model calculations (Gunshor et al. 2004; Tjemkes et al. 2001). The present study is able to avoid that issue by using high-spectral-resolution infrared data to simulate the spectral response of the broadband imagers.

An international effort is being undertaken (Goldberg 2007) to intercalibrate the world’s environmental satellites: Global Space-based Intercalibration System (GSICS). One of the goals of this undertaking is “to improve the use of space-based global observations for weather, climate, and environmental applications through operational intercalibration of the space component of the World Weather Watch’s (WWW) Global Observing System (GOS) and Global Earth Observing System of Systems (GEOSS)” [available online at http://www.star.nesdis.noaa.gov/smcd/spb/calibration/icvs/GSICS/index.php]. Intercalibration of broadband geostationary imagers with a high-spectral-resolution instrument such as Atmospheric Infrared Sounder (AIRS) represents a major step in reaching the goals set forth by GEOSS and the World Meteorological Organization (WMO).

Geostationary simultaneous nadir observations (GSNOs) are collected for Earth Observing System (EOS) AIRS on board Aqua and a global array of geostationary imagers. The imagers compared in this study are on (Geostationary Operational Environmental Satellites) GOES-10, GOES-11, and GOES-12 from the United States; (Meteorological Satellites) Meteosat-8 and Meteosat-9 from the European Organisation for the Exploitation of Meteorological Satellites (EUMETSAT); Multifunction Transport Satellite-IR (MTSAT-1R) from Japan; and Fengyun-2C (FY-2C) from China (Table 1). This is an expansion upon Gunshor et al. (2007). It has been shown that a single polar-orbiting, or low earth orbiting (LEO), satellite can be used to intercalibrate any number of geostationary (GEO) imagers (Gunshor et al. 2004). Using a high-spectral-resolution IR sensor—in this case, AIRS—with absolute calibration to within 0.2 K in most bands (Tobin et al. 2006b) refines the “LEO–GEO” method, which is closer to an absolute estimate of imager calibration accuracy.

This method of comparison with AIRS data—as the standard implies—is that it is pertinent to know how well the individual AIRS spectra are calibrated. This has been researched by a number of investigators for many types of atmospheric conditions. For example, Tobin et al. (2006b) have compared AIRS data to time coincident Scanning High-Resolution Interferometer Sounder (S-HIS) aircraft data at comparable spectral and spatial resolutions and found very good agreement. The “double observation minus calculation difference” was smaller than 0.2 K for the longwave, midwave, and shortwave spectral regions. This has been duplicated with other high-spectral-resolution data, such as Infrared Atmospheric Sounding Interferometer (IASI) and U.S. National Polar-orbiting Operational Environmental Satellite System (NPOESS) Atmospheric Sounder Testbed—Interferometer (NAST-I), demonstrating the high quality of AIRS data that is being compared with the many geostationary imager data. In early in-flight measurements, IASI compared favorably with AIRS, to within 0.1 K in most bands (Blumstein et al. 2007).

2. Method

The technique for intercalibration was developed at the Cooperative Institute for Meteorological Satellite Studies (CIMSS) during the past decade (Schmit and Herman 1992; Wanzong et al. 1998; Gunshor et al. 2004). The following paragraphs describe the process of obtaining and processing a single intercomparison. There are four steps to satellite intercalibration: data collection, data transformations, study area selection, and calculations. The intercalibration method used here is similar to the one described in Gunshor et al. (2004). The fundamental steps of that method are still used, though altered to accommodate high-spectral-resolution AIRS data in use as the polar-orbiting instrument.

a. Data collection

Collocation of the polar orbiter and geostationary data in space and time is required. GSNOs are used to minimize satellite zenith angle differences between the instruments and should employ a minimal temporal offset. Data are selected within ±10° (latitude and longitude) of the geostationary subsatellite point, and AIRS radiometer scan angles must be ± 10° to minimize satellite zenith angle differences. This encompasses the 18 central AIRS FOVs. To maximize scene consistency between the two instruments, ideally GEO data are collected within ±15 min of LEO overpass time at the geostationary subsatellite point. For imagers that only provide equatorial data once an hour, this requirement of the GSNO must be waived. The nature of the sun-synchronous orbit typically employed for low earth orbiting platforms, such as Aqua, is that at a single location on Earth, there are essentially only two overpass times—a “day” and a “night” pass from which the overpass time does not significantly vary. Thus, a disadvantage of this technique is that if the GEO does not sample often enough, it may not be possible to obtain data within ±15 min of the overpass. The worse-case scenario is a range within plus or minus half the number of minutes of the GEO’s sampling rate. For instance, if a GEO collects data once an hour, a LEO pass within the GSNO range will always be within at least 30 min of the GEO data. Sampling time differences are at least partially mitigated by spatial smoothing; thus, it is more important to collect comparisons than it is to adhere within strict limits to define a GSNO.

The most difficult part of comparing two broadband instruments is accounting for differences in their spectral responses. In Gunshor et al. (2004), this was done with a forward model calculation on an atmospheric profile generated from a global numerical model. The results were then somewhat vulnerable to the accuracy of the model and to assumptions regarding the surface emissivity. That method also requires an additional data collection step. This step has been removed in the technique as it applies to AIRS, which is an advantage the AIRS method has over the prior method.

b. Data transformations

There are a number of data transformations necessary before the comparison can occur. Data, in the form of radiances, from each satellite are smoothed to an effective 100-km resolution to mitigate the effects of different FOV sizes and sampling densities, and AIRS samples at an effective 13-km FOV. Nominal nadir viewing FOV sizes for the GEOs are in Table 1. Image smoothing consists of applying a running average; the resultant image has the same number of pixels as the original; however, each pixel is the average of the approximate 100 km × 100 km surrounding it. Another option would be to resample or average the data to a 100 km × 100 km grid, thereby simulating instruments with 100-km footprints.

Scaled radiances, commonly referred to as counts, are converted to radiances with a linear conversion; this process is described for the GOES variable format (GVAR) data stream by Weinreb et al. (1997), and a similar process is followed for the other instrument data streams. Typically, users obtain data as radiances, and it is necessary that all calculations performed on the data concerning spatial smoothing or averaging are computed on the radiances or scaled radiances, before converting to equivalent brightness temperatures.

AIRS is known to have some poorly performing (or bad) channels. There is a consistent set of bad channels as well as channels that can be bad from granule to granule. There is a test using the appropriate AIRS channel properties file that does a reasonable job eliminating bad channels. Additionally, there are other simple data quality checks to be sure no channels are used that have radiances at or below 0 or radiances unreasonably high. For a single AIRS granule, the same channel subset is used for every field of view. This is not a consistent channel subset for all cases that form the overall averages seen later in the results. Also, AIRS has spectral gaps that are part of the instrument design. This means that not every GEO band is sufficiently covered spectrally by AIRS. The subject of spectral gaps will be covered in section 3.

AIRS data are convolved with GEO spectral response functions (SRFs) and, in this manner, are believed to be an accurate simulation of the geostationary instrument. The term “convolved” is widely used in this application and will be used throughout this paper, though it may be more accurate to describe the process mathematically as a weighted average:
i1520-0426-26-4-746-e1
where CR is the convolved radiance, R is the AIRS radiance at wavenumber υ, and υ1 and υ2 are the lower and upper wavenumber limits, respectively, of the GEO SRF S. In this way, we calculate an average AIRS radiance, weighted by the geostationary imager’s SRF, for each AIRS FOV. In theory, and if the GEO SRF is accurate, this convolved radiance would be the same as what the geostationary imager would given the same instrument characteristics as AIRS (FOV shape, FOV size, altitude, angle, etc.).

c. Study area

The study area is located between 10° latitude north and south of the equator and 10° longitude east and west of the geostationary subsatellite point. These are the outer limits of a potential study area, however, and not the specific study area for a given case, that varies depending on the orbital characteristics of AIRS and the overlapping spatial coverage of the two instruments. The study area for an individual case is a narrow strip of data 18 AIRS FOV wide that fits within those criteria and is less than ±10° scan angle for AIRS (with a mean scan angle of 0°). The GEO view angles vary between 0° and approximately ±15°, and the mean view angle varies by case between approximately 2° and 10°. A single intercalibration case will consist of a subset of one AIRS granule and a subset of one geostationary image corresponding to one scan time.

d. Calculations

AIRS data, specifically radiances or scaled radiances, are convolved with the GEO spectral response. This is a common technique for simulating broadband radiances from a high-spectral-resolution instrument, such as AIRS (Tobin et al. 2006a). The mean radiance is computed by averaging the spatially smoothed data within the study area for both instruments (AIRS data are convolved with GEO SRF before being spatially smoothed.). The mean measured radiance difference between GEO and AIRS is attributed to calibration differences between the two instruments. These radiances are converted to brightness temperature using the inverse Planck equation, using band correction coefficients for the specific GEO band of interest. The results section shows the mean intercalibration difference for multiple such cases, presented as brightness temperature differences.

3. Spectral gaps in AIRS

Accounting for spectral gaps in AIRS is an important issue. In Tobin et al. (2006a), a method was presented where a “convolution error” was calculated for several synthetic atmospheric spectra, and it appears this is a reasonable fix. It is important to consider other possible methods of dealing with this issue, since other high-spectral-resolution instruments in the future will also have spectral gaps, such as the Cross-track Infrared Sounder (CrIS), which will fly on the NPOESS spacecraft. A method is presented here of filling the spectral gaps with generic spectral information from a calculated atmosphere that has been adjusted to fit each AIRS FOV. This spectrum was originally calculated at 0.1 cm−1 resolution using the line-by-line radiative transfer model (LBLRTM) developed at Atmospheric and Environmental Research (AER) Inc. (Clough and Iacono 1995) and must be convolved with AIRS SRFs to match AIRS channel centers. This is done by calculating Gaussian SRFs for AIRS bands at AIRS channel centers (taken from the AIRS channel properties file). In the AIRS spectral gaps, channel centers are calculated by using a mean of the widths between channel centers on either side of the gap and applying that across the gap from the longwave side. This was done for any individual gap greater than 5 wavenumbers.

The U.S. Standard Atmosphere, 1976 (USSA) calculated spectrum is convolved with Gaussian mock AIRS SRFs with channel centers ranging from 649.6 to 2999.25 μm. This simulates an AIRS-like view of the calculated spectrum, with imaginary AIRS channels in the AIRS spectral gaps and extending into the shortwave region of the spectrum. These convolved radiances are then converted to brightness temperatures as monochromatic radiances. Since radiances are wavelength dependent, the gaps are filled using brightness temperatures.

For each AIRS FOV, prior to spatial smoothing, the gaps are filled across the spectrum. For each gap greater than 5 wavenumbers, the temperatures at the endpoints of the gap are compared to the temperatures at the same wavelengths for the theoretical spectrum described above. The theoretical data are adjusted at the gap endpoints to fit the measured AIRS spectrum. Then the points in the gap are filled with temperatures using a weighted average between the two endpoints, weighted by distance across the gap. For instance, at the midpoint of the gap, half of the temperature difference at one endpoint is added to half of the temperature difference at the other endpoint and that sum total is added to (or subtracted from, as the case may be) the theoretical spectrum at that point. The data used to fill the gaps are called the “adjusted U.S. Standard Atmosphere,” or “adjusted USSA,” and are converted back into radiances (again, monochromatically) for the rest of the calculations. An example of the finished product can be seen in Fig. 1.

This method appears to be effective, though there is room for improvement in its application. Since the measured atmosphere varies from the U.S. Standard Atmosphere in temperature or moisture content, adjusting the AIRS spectra in the manner described does not always accurately take these changes into consideration. It has been shown in preliminary results that using a cloudy spectrum in cloudy scenes may work better than the USSA (Y. Tahara 2008, personal communication). Also, the method may be less useful if one endpoint, or both, of a spectral gap falls on an atmospheric absorption line.

4. Error analysis

The known sources of error in intercalibration are caused by temporal collocation differences, spatial collocation differences (partially due to spatial resolution differences), and geometric alignment differences (satellite zenith angle and azimuth angle differences). The errors that are specific to this type of intercalibration include spectral convolution errors, errors in the measurement of the spectral response of various GEO bands, and errors associated with the method used to fill AIRS spectral gaps. The biggest errors are caused by the AIRS spectral gaps and errors in GEO spectral response. The temporal collocation differences can also have a measurable effect on the results. These error sources may interact unpredictably for some scene types. It is believed that spatial collocation differences and geometric alignment differences are minimal—since in the method presented here, a 100 km × 100 km spatial smoothing is applied to each FOV and because a fairly large area of collocation is used in the comparisons instead of a pixel-to-pixel-type approach.

a. Spectral convolution and spectral gap-filling error analysis

Using the spectrum calculated from the USSA at 0.1 cm−1 resolution, it was determined that Gaussian (mock) AIRS SRFs can be used in place of the real ones. It is preferable to use mock SRFs for consistency. Since mock SRFs will be used in the spectral gaps, it is important to know that Gaussian approximations are effective and that a consistent set of mock SRFs can be used for all cases in all bands. A test was performed by convolving the USSA spectrum with the mock SRFs (only where AIRS channels exist) and the real SRFs, thus two simulated atmospheres were created. Both of these were then converted to brightness temperature. There are differences for some AIRS channels: the maximum being 4.6 K but the mean is 0.02 K, with a standard deviation of 0.6 K (Fig. 2). For some individual AIRS channels, the difference between real and mock SRFs is significant. However, when convolving these two simulations of the USSA with GOES and Meteosat-8 SRFs, those differences disappear. For the purposes of intercalibrating a broadband sensor, such as the GOES imager or Meteosat Spinning Enhanced Visible and Infrared Imager (SEVIRI) on Meteosat-8 and Meteosat-9, the mock SRFs are an adequate approximation. Even in bands where the differences between mock and real are relatively large, such as in the 13-μm region (Fig. 2), the results upon convolving with GOES-12 or Meteosat-8 SRFs (Table 2) show there is no measurable difference. The only broadband channel where there appears to be a difference is in Meteosat-8 band 7 (8.7 μm), where the SRF is almost entirely inside an AIRS spectral gap. Table 2 includes a column where the original USSA spectrum, at 0.1 cm−1 spectral resolution, was convolved with the GOES-12 and Meteosat-8 bands and a brightness temperature was calculated. This shows what sort of error the AIRS spectral gaps introduce for these bands—if one does not try to account for it. For GOES-12 and Meteosat-8 bands that do not have significant AIRS spectral gaps, nearly identical brightness temperatures were calculated as those for the mock and real AIRS SRFs. Such bands include bands 4 and 6 of the GOES-12 imager and bands 6, 9, 10, and 11 of Meteosat-8. For this table, convolutions were not done when the SRF exceeded the endpoints of the spectrum, as is the case for the shortwave window on both GOES-12 and Meteosat-8.

An analysis performed on a set of 32 atmospheres used originally at CIMSS for radiative transfer modeling studies in relation to Nimbus-5 sounder data (Smith et al. 1974), denoted as the CIMSS-32, shows how well this method works on calculated atmospheres that have been converted to high spectral resolution (at 0.1 cm−1). For this study comparing AIRS to GEOs in an equatorial environment, only the atmospheres taken from the tropics are pertinent; however, results from other atmospheres are discussed as well, since GSNO is not the only method by which intercalibration can be done. For those considering simultaneous nadir observations (SNOs between two LEOs) in colder atmospheres, these other results may be of interest. The first of the 32 is the U.S. Standard Atmosphere, 1976. Atmospheres 25 and 27–32 are essentially tropical atmospheres, within 10° latitude of the equator. Without going into much detail, most of the other atmospheres are from relatively colder locations (Table 3).

Using the same method described in the methods section for preparing the U.S. Standard Atmosphere to fill the spectral gaps, each of the CIMSS-32 calculated spectra was convolved with the mock AIRS SRFs to produce spectra that looked like AIRS with AIRS channel centers. However, the spectral gaps were left in. Then the gap-filling method was used to fill the gaps of these AIRS-like CIMSS-32 with the adjusted USSA. These gap-filled AIRS-like CIMSS-32 spectra were then convolved with GOES-12 and Meteosat-8 SRFs and a subsequent brightness temperature was calculated. These brightness temperatures were then compared to the brightness temperatures calculated by convolving GOES-12 and Meteosat-8 bands with the original CIMSS-32 prior to adding AIRS gaps. The GOES-12 and Meteosat-8 bands are close enough spectrally to the other geostationary imager bands that the results in this test with these two can be used as a proxy for the others.

The results of this test are encouraging where there are not significant AIRS spectral gaps. Small gaps pose no problem whatsoever to this gap-filling method, and the U.S. Standard Atmosphere is a good enough proxy atmosphere to fill those gaps for all 32 atmospheres tested in such bands as the IR window (IRW), for example (Fig. 3). In the water vapor region, where AIRS spectral gaps comprise a significant portion of the GEO SRF, the results are mixed. In the tropical atmospheres, for GOES-12 imager band 3, there are differences up to 0.3 K (Fig. 3). While this is a nontrivial difference, it is deemed acceptable considering the alternative of not accounting for the gaps at all introduces errors on the order of several degrees in this band. It is clear though, from Fig. 3, that the U.S. Standard Atmosphere is not the appropriate atmosphere to use to fill the gaps of all of the CIMSS-32. For this study where the comparisons are done at the equator, it will suffice. For studies with SNOs in other latitude regions, such as Wang et al. (2007), this may be a serious consideration.

The Meteosat-8 shortwave band (band 4) is less straightforward to analyze. In this band, the shortwave gap is not so much a gap as it is an extension of the AIRS spectrum. Since this band extends in spectral coverage beyond the shortest-wave AIRS band, an assumption has been made about the shortwave to extend the spectrum. The assumption is that the slope in the spectrum from the last AIRS channel, near 2665.2 cm−1, out to approximately 2999.5 cm−1 remains constant in brightness temperature space. It is likely that shortwave window channel intercalibration is only useful at night, when there is no solar-reflected component to obscure the measurements. With these calculated spectra, we can see how well the method might work at night. The results are encouraging (Fig. 3) and suggest that the method does a more than adequate job extending the shortwave side of the spectrum beyond AIRS spectral coverage.

The test showed that there are relatively large errors associated in all 32 atmospheres using this method for Meteosat-8 band 5 (approximately 0.5 K in the tropical atmospheres) and band 7 (greater than 1 K in most of the tropical atmospheres), though it was better in the tropical atmospheres than the others. For all of the other bands—besides the GOES-12 water vapor, as mentioned already—the errors associated with this method are very small, less than 0.1 K for all atmospheres in all the other bands.

b. Time dependency of matches

When collecting data for this study, the initial criterion for matching AIRS with a geostationary imager is based on the nominal image time. There is room for error here related to the difference in time when the two instruments actually scanned the lines used in the comparison. This time difference for individual cases is somewhat correlated to the temperature difference for those cases. By comparing the actual scan times at the GOES-12 subpoint for AIRS to GOES-12, it was discovered that the actual range in differences in the scan times between the two instruments ranged from 8 s to approximately 24.5 min. Table 4 examines the GOES-12 11-μm results more closely, by limiting the comparisons to only those that were within 15, 10, and 5 min. All of these steps show an improvement in the mean temperature difference and standard deviation over the results without sorting for time, though there is little difference between the 15- and 10-min increments. There are 174 infrared window band comparisons for GOES-12 but limiting the comparisons to an actual 15-min time difference reduces the number of comparisons n to 91 or 52% of the original sample size; limiting the comparisons to a 10-min time difference reduces n to 73 or 42% of the original sample size, and limiting the comparisons to a 5-min time difference reduces n to 46 or 26% of the original sample size. Reducing the sample size greatly may not be ideal in all situations, but limiting the comparisons to only those with relatively small time differences provides a more accurate assessment of the calibration of the instrument. By looking at comparisons with only a 5-min or smaller time difference between AIRS and GOES-12, all of the 11-μm band cases with brightness temperature differences greater than ±0.8 K are removed. Because the mean difference between GOES-12 and AIRS in this band is very small, the time difference between comparisons is a likely explanation for any of the individual cases that had very-high brightness temperature differences, which included one case with a difference of more than 3 K. By limiting the comparisons to a 5-min time difference for the other bands on GOES-12 yields interesting results (Table 5). The 3.9- and 13.3-μm band mean and standard deviation also improve, but this not true for the 6.7-μm water vapor band.

c. GEO spectral response uncertainty errors

In Tobin et al. (2006a), it was shown that most of the difference between AIRS and a broadband imager [Moderate Resolution Imaging Spectroradiometer (MODIS)] can be removed by shifting the SRF of the MODIS band without otherwise changing its shape. A similar exercise was done after the GOES-13 science test (Hillger and Schmit 2007).

In 19 cases compared to AIRS, the 13.3-μm band on GOES-13 had a −2.4-K mean brightness temperature difference and a standard deviation of 0.6 K. By shifting the GOES-13 SRF −4.7 cm−1 (Fig. 4), the mean brightness temperature difference was reduced to 0 K, with a standard deviation of 0.7 K. The value of −4.7 cm−1 was arrived at empirically using an iterative method. As shown in the results section below, there appear to be calibration problems with this band on GOES-12 and both Meteosats.

ITT Industries, the manufacturers of the GOES imagers, re-evaluated the 13.3-μm band SRFs. Updated SRFs were made available for that band. The new SRFs for that band have a modified shape that effectively shifted the central wavenumber by approximately −1 cm−1. ITT concluded that the spectral response for this band could be shifted by −4.7 cm−1 without being implausibly beyond the estimated instrument spectral uncertainty. The National Oceanic and Atmospheric Administration’s (NOAA) calibration experts continued to investigate this matter, with hopes of coming to a resolution prior to GOES-13 becoming operational. They have found that shifting the SRF eliminated the scene temperature bias, and that the bias is most likely not due to detector nonlinearity or other factors (X. Wu 2008, personal communication). This type of exercise could be done with all of the world’s geostationary imagers, though that falls outside the scope of this paper, which is to demonstrate how to measure these types of errors that exist in the data with the official or operational spectral response. It appears, though, that the major reason for differences in comparisons to AIRS is uncertainty in the GEO SRF. This is not an error in this method. This is a calibration error caused by the various methods employed in laboratory measurements of detector spectral response and also possibly by on-orbit contamination due to the build up of ice on the detectors, for example.

d. Scene uniformity

A measure of scene uniformity can be determined by the standard deviation about the mean of the radiances in the comparison area. The geostationary radiances were used for this purpose because of their finer spatial resolution over AIRS. Comparing GOES-12 scene uniformity to the GEO–AIRS brightness temperature difference showed there was not a statistically significant correlation between the two for any bands. The results are believed to be independent of scene uniformity.

5. Results

With the knowledge that the method works very well for GOES bands 2, 4, and 6 and reasonably well for band 3 and for Meteosat bands other than 5 and 7, the Tables 6a–f are presented. Results are not shown for Meteosat bands 5 (6.2 μm) or 7 (8.7 μm), because the gap filling method does not work well enough for those bands. The results calculated for the 6.2-μm band were particularly poor (mean brightness temperature differences approximately −1.4 K) and would unfairly characterize the instrument’s performance in that band. Surprisingly, the results calculated for the 8.7-μm band were very good (mean brightness temperature differences approximately 0.2 K), but they may not necessarily be indicative of a well-calibrated channel. Most of that band is not covered spectrally by AIRS. The results in Table 6f, the water vapor bands, are shown here with less confidence than those of the other bands.

The results cover cases collected from January 2006 through early October 2007, though not all of the instruments were operational during that entire period. GOES-11 cases only cover June 2006 through October 2007. Meteosat-8 cases run from January 2006 to 10 April 2007. Meteosat-9 cases run from 12 April 2007 to October 2007. The results for the shortwave window bands are nighttime cases only.

a. Diurnal signature in the shortwave window bands

The shortwave window band is susceptible to solar reflection, and this has an effect on the results (Table 7). Therefore, the results in Table 6a included only nighttime cases. There are two overpass times for Aqua at a given point on the equator: one is during daylight hours and the other is during nighttime hours. The results generally improve when only using nighttime cases in the shortwave band, as both the mean temperature difference and standard deviation move closer to 0 K.

b. MTSAT’s shortwave window band

Puschell et al. (2006) discussed problems with MTSAT’s shortwave window band. There is crosstalk between the water vapor band and the shortwave band. Two corrections were implemented: one in mid-March of 2006 and the other in mid-July of 2006. Prior to the first correction being implemented, mean temperature differences between AIRS and MTSAT in this region were slightly worse than 9 K (MTSAT’s colder than AIRS). For the results shown here, only cases after 20 July 2006 are used for that band and the mean temperature difference is improved to −0.1 K, with a standard deviation of 0.9 K.

c. FY-2C’s midnight problem

FY-2C has an instrument problem that affects calibration around satellite midnight. Stray light is introduced by direct solar irradiations to the telescope tube (Guo et al. 2005). Light from the telescope mounts affects the imagery between approximately 1500 and 2000 UTC. The effect is most dramatic in the visible and shortwave bands, but it affects the quality of all bands during those times. One of the AIRS overpass times at the FY-2C subpoint is between 1800 and 1900 UTC and that data cannot be used. Fortunately, nighttime shortwave band cases can be used. Only the nighttime cases are used for the other bands in the results shown here as well to avoid the contaminated times. By eliminating the contaminated cases, the mean temperature differences were not always moved closer to 0 K, though the standard deviations were reduced in each of the bands: from 3.7 to 2.9 K for the water vapor band; from 2.7 to 1.7 K in the IR window; and from 3.7 to 1.8 K in the “dirty” window band.

d. GOES-12 decontamination

The GOES-12 imager went through a decontamination (a heating of certain internal optical components) that lasted from just after 1200 UTC on 2 July to just after 1200 UTC on 4 July 2007. There was a case collected on 1 July 2007 and the next one was 8 July 2007, bracketing the decontamination period. The 6.7-μm band differences change sharply between these two days (Fig. 5). The mean brightness temperature difference before this period and the mean after that period are roughly the same but opposite in sign (0.4 before, −0.4 K after). The sudden change in the results of the comparisons between AIRS and GOES-12 in this region of the spectrum can be attributed to the decontamination (Table 8). This is perhaps a result of ice deposition on the detectors, which would have the effect of altering the detector’s spectral response.

6. Summary

To the first approximation, most of the bands on the world’s geostationary imagers are reasonably well calibrated—that is, they compare to within 1 K of a standard reference (AIRS). Using AIRS as a common reference, typically, the GOES series of satellites are similar to each other in most bands; the Meteosat Second Generation series of instruments are also similar to each other, as are MTSAT-1R and FY-2C. The most glaring exception to this is in the 13.3-μm band. In that band, the Meteosat series are the most different from AIRS and from each other. The GOES-12 imager is still nearly 1-K different from AIRS in that band, even after decontamination. Preliminary results, not shown, with the GOES-13 imager suggest that it, too, is different from both AIRS and GOES-12 in the 13.3-μm band (Hillger and Schmit 2007). This highlights a fundamental truth about intercalibration, which is that the most important time to perform intercalibration is following launch, before a satellite becomes operational. This can potentially reveal problems that are not found during engineering postlaunch tests and those problems can often be addressed before the instrument is expected to fulfill an operational role. The other unfortunate finding of these results is that FY-2C is not as well calibrated as the other geostationary imagers. In these comparisons to AIRS, FY-2C has larger mean brightness temperature differences and larger standard deviations than any of the other instruments.

The method of filling AIRS spectral gaps with the adjusted U.S. Standard Atmosphere spectrum works well for most bands for comparisons to AIRS near the equator. The exception to this is when the AIRS spectral gaps fall too heavily on GEO SRFs, such as for the water vapor bands and the 8.7-μm band on Meteosat. For the other GEO bands, the method appears to work well. For the water vapor bands, the results shown here are not meant to be accurate assessments of the calibration of those instruments. Results were not even generated for the 8.7-μm band as a result that uncertainty. In future comparisons, the standard tropical atmosphere will be tested more thoroughly for use in the gap filling, since these comparisons are done in tropical regions. Preliminary tests suggest it will not affect the results in most bands but could improve the results in the water vapor bands by approximately 0.1 K.

The next step in the evolution of GEO intercalibration is to use IASI data. IASI is part of the Meterological Operation-A (MetOp-A) satellite package operated by the European Organisation for the Exploitation of Meteorological Satellites (EUMETSAT). IASI is a high-spectral-resolution instrument similar to AIRS but without significant spectral gaps. Efforts are already underway at EUMETSAT, and will begin soon at CIMSS and elsewhere, to use IASI to intercalibrate the world’s geostationary imagers. Preliminary results are being circulated among the World Meteorological Organization’s (WMO) Global Space-based Intercalibration System (GSICS) Research Working Group members.

Acknowledgments

The authors thank Tim Hewison (EUMETSAT), Marianne Koenig (EUMETSAT), Fred Wu (NOAA), and the rest of the GSICS community for their input, advice, questions, and collaboration. Jim Nelson (CIMSS) is thanked for his support on software and hardware issues. Hal Woolf (CIMSS) is thanked for providing the CIMSS-32 atmospheres, spectral response functions for many of the instruments, and for his editorial comments. The SSEC Data Center is thanked for providing easily accessible, real-time satellite data in McIDAS format. Walter Wolf (NOAA) is thanked for providing AIRS data in near-real time. The GIMPAP project at NOAA is to be thanked for their financial support of this project. This research was performed primarily under NOAA Grant NA06NES4400002. The views, opinions, and findings contained in this report are those of the author(s) and should not be construed as an official NOAA or U.S. government position, policy, or decision.

REFERENCES

  • Blumstein, D., Tournier B. , Cayla F. R. , Phulpin T. , Fjortoft R. , Buel C. , and Ponce G. , 2007: In-flight performance of the infrared atmospheric sounding interferometer (IASI) on METOP-A. Atmospheric and Environmental Remote Sensing Data Processing and Utilization III: Readiness for GEOSS, M. D. Goldberg et al., Eds., International Society for Optical Engineering (SPIE Proceedings, Vol. 6684), DOI:10.1117/12.734162.

    • Search Google Scholar
    • Export Citation
  • Chahine, M. T., and Coauthors, 2006: AIRS: Improving weather forecasting and providing new data on greenhouse gases. Bull. Amer. Meteor. Soc., 87 , 911926.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Clough, S. A., and Iacono M. J. , 1995: Line-by-line calculation of atmospheric fluxes and cooling rates II: Application to carbon dioxide, ozone, methane, nitrous oxide and the halocarbons. J. Geophys. Res., 100 , 1651916535.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Goldberg, M. D., 2007: Global space-based inter-calibration system (GSICS). Atmospheric and Environmental Remote Sensing Data Processing and Utilization III: Readiness for GEOSS, M. D. Goldberg et al., Eds., International Society for Optical Engineering (SPIE Proceedings, Vol. 6684), DOI:10.1117/12.735246.

    • Search Google Scholar
    • Export Citation
  • Gunshor, M. M., Schmit T. J. , and Menzel W. P. , 2004: Intercalibration of the infrared window and water vapor channels on operational geostationary environmental satellites using a single polar-orbiting satellite. J. Atmos. Oceanic Technol., 21 , 6168.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Gunshor, M. M., Schmit T. J. , Menzel W. P. , and Tobin D. , 2007: Intercalibrating geostationary imagers via polar orbiting high spectral resolution data. Atmospheric and Environmental Remote Sensing Data Processing and Utilization III: Readiness for GEOSS, M. D. Goldberg et al., Eds., International Society for Optical Engineering (SPIE Proceedings, Vol. 6684), DOI:10.1117/12.735855.

    • Search Google Scholar
    • Export Citation
  • Guo, Q., Xu J. , and Zhang W. , 2005: Stray light modelling and analysis for the FY-2 meteorological satellite. Int. J. Remote Sens., 26 , 28172830.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Hillger, D. W., and Schmit T. J. , 2007: The GOES-13 science test: Imager and sounder radiance and product validations. NOAA Tech. Rep. NESDIS 125, 75 pp.

    • Search Google Scholar
    • Export Citation
  • Puschell, J. J., and Coauthors, 2006: In-flight performance of the Japanese Advanced Meteorological Imager. Earth Observing Systems XI, J. J. Butler, Ed., International Society for Optical Engineering (SPIE Proceedings, Vol. 6296), DOI:10.1117/12.683505.

    • Search Google Scholar
    • Export Citation
  • Schmit, T. J., and Herman L. D. , 1992: Comparison of multi-spectral data: GOES-7 (VAS) to NOAA-11 (HIRS). Preprints, Sixth Conf. on Satellite Meteorology and Oceanography, Atlanta, GA, Amer. Meteor. Soc., 258–259.

  • Smith, W. L., Woolf H. M. , Abel P. G. , Hayden C. M. , Chalfant M. , and Grody N. , 1974: NIMBUS-5 Sounder Data Processing System. Part I: Measurement characteristics and data reduction procedures. NOAA Tech. Memo. National Environmental Satellite Service (NESS) 57, 99 pp.

    • Search Google Scholar
    • Export Citation
  • Tjemkes, S. A., König M. , Lutz H-J. , van de Berg L. , and Schmetz J. , 2001: Calibration of Meteosat water vapor channel observations with independent satellite observations. J. Geophys. Res., 106 , 51995209.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Tobin, D. C., Revercomb H. E. , Moeller C. C. , and Pagano T. S. , 2006a: Use of Atmospheric Infrared Sounder high-spectral resolution spectra to assess the calibration to Moderate resolution Imaging Spectroradiometer on EOS Aqua. J. Geophys. Res., 111 , D09S05. doi:10.1029/2005JD006095.

    • Search Google Scholar
    • Export Citation
  • Tobin, D. C., and Coauthors, 2006b: Radiometric and spectral validation of Atmospheric Infrared Sounder observations with the aircraft-based Scanning High-Resolution Interferometer Sounder. J. Geophys. Res., 111 , D09S02. doi:10.1029/2005JD006094.

    • Search Google Scholar
    • Export Citation
  • Wang, L., Cao C. , and Ciren P. , 2007: Assessing NOAA-16 HIRS radiance accuracy using simultaneous nadir overpass observations from AIRS. J. Atmos. Oceanic Technol., 24 , 15461561.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Wanzong, S., Menzel W. P. , and Callan G. , 1998: Intercalibration of GOES, Meteosat, GMS, HIRS, and AVHRR infrared radiances. Preprints, Ninth Conf. on Satellite Meteorology and Oceanography, Paris, France, Amer. Meteor. Soc., 714–717.

  • Weinreb, M. P., Jamison M. , Fulton N. , Chen Y. , Johnson J. X. , Bremer J. , Smith C. , and Baucom J. , 1997: Operational calibration of Geostationary Operational Environmental Satellite-8 and -9 imagers and sounders. Appl. Opt., 36 , 68956904.

    • Crossref
    • Search Google Scholar
    • Export Citation

Fig. 1.
Fig. 1.

Sample tropical spectrum (black) with spectral gaps filled with the adjusted USSA (gray).

Citation: Journal of Atmospheric and Oceanic Technology 26, 4; 10.1175/2008JTECHA1155.1

Fig. 2.
Fig. 2.

GOES-12 imager (green) and Meteosat-8 (blue) SRFs shown with the brightness temperature difference ΔTbb of AIRS real SRFs convolved with the USSA spectrum subtracted from AIRS mock SRFs convolved with the USSA spectrum for each AIRS channel. (right to left) GOES bands are 2, 3, 4, and 6 and Meteostat-8 bands are 4–11.

Citation: Journal of Atmospheric and Oceanic Technology 26, 4; 10.1175/2008JTECHA1155.1

Fig. 3.
Fig. 3.

The CIMSS-32, convolved with mock AIRS SRFs and then gap filled with the adjusted USSA, were then convolved with GEO SRFs and differenced with the CIMSS-32 convolved with the GEO SRFs. The GOES-12 IRW (circles), GOES-12 water vapor band (squares), and the Meteosat-8 shortwave window (asterisks) are shown. The differences are less than 0.1 K for all 32 atmospheres for all the bands except for the water vapor bands and the Meteosat 8.7-μm band (not shown).

Citation: Journal of Atmospheric and Oceanic Technology 26, 4; 10.1175/2008JTECHA1155.1

Fig. 4.
Fig. 4.

GOES-13 imager band 6 (13.3 μm) SRF (blue) and the shifted SRF (green) shifted −4.7 cm−1 (to approximately 13.4 μm), with a representative brightness temperature Tbb spectrum in the background (red). By shifting the spectral response this amount, the bias, or mean Tbb difference for all 19 cases, becomes 0.0 K (rounded from 0.01 K) with a std dev of 0.7 K.

Citation: Journal of Atmospheric and Oceanic Technology 26, 4; 10.1175/2008JTECHA1155.1

Fig. 5.
Fig. 5.

Time series of GOES-12 water vapor band differences with AIRS, illustrating the effect of decontamination.

Citation: Journal of Atmospheric and Oceanic Technology 26, 4; 10.1175/2008JTECHA1155.1

Table 1.

Nominal subpoints of the geostationary imagers in this study over the equator. Meteosat-8 and Meteosat-9 are remapped to 0°, but Meteosat-8 is located at approximately 3.5°W. Nominal FOV sizes at nadir of the infrared bands are included. GOES imager oversamples 4-km FOVs in the east–west by 1.7 km, with the exception of 8-km FOVs, the GOES-10 and GOES-11 water vapor band, and the GOES-12 13.3-μm bands. Meteostat-8 and Meteostat-9 have been remapped to 3 km. MTSAT-1R samples at 2 km but the data have been remapped to 4 km.

Table 1.
Table 2.

The USSA radiance spectrum was convolved (⊗) with mock (Gaussian) AIRS SRFs, which introduce AIRS spectral gaps to the USSA, and then convolved with (top) GOES-12 imager and (bottom) Meteosat-8 SRFs and converted to Tbb (USSA ⊗ w/mock AIRS SRFs Tbb). The same was done with a set of actual AIRS SRFs (USSA ⊗ w/real AIRS SRFS Tbb). The original USSA radiance spectrum was also convolved with the GOES-12 imager (top) and Meteosat-8 (bottom) SRFs and converted to Tbb (original USSA). The differences in the final column show why it is necessary to account for AIRS spectral gaps in some GEO channels. Channels in bold type are channels for which it appears the spectral gaps are most manageable.

Table 2.
Table 3.

The location and dates of CIMSS-32. The first atmosphere listed is really the USSA, and thus the location and date are not literal. Total precipitable water (TPW) is included.

Table 3.
Table 4.

GOES-12 IRW band comparisons to AIRS spanning 8 Jan 2006 to 5 Oct 2007. Comparisons limiting the scan time difference between GOES and AIRS yielded improved results compared to allowing all of the cases, which spanned in time difference from approximately 8 s to 24.5 min.

Table 4.
Table 5.

GOES-12 comparisons to AIRS spanning 8 Jan 2006 to 5 Oct 2007 for all bands. Comparisons limited to a 5-min scan time difference between GOES and AIRS yielded improved results over allowing all of the cases, which spanned in time difference from approximately 8 s to 24.5 min.

Table 5.
Table 6.

Here, ΔT (GEO–AIRS) is the mean Tbb difference over N cases collected from early January 2006 through early October 2007. Std dev is the dev about the mean. The min and max mean scene Tbb (Tbb-Min and Tbb-Max, respectively) (for all FOVs in a single comparison) found in those N cases. (f) The water vapor band results, but those results reflect less confidence in the method, based on the tests referenced in Fig. 3.

Table 6.
Table 7.

Results showing the effects of daytime vs nighttime comparisons in the shortwave window bands. FY-2C is not included because no FY-2C daytime cases were used in this study.

Table 7.
Table 8.

Results before and after GOES-12 decontamination event (2–4 Jul 2007).

Table 8.
Save