Comparison of Nimbus-7 SMMR and GOES-1 VISSR Atmospheric Liquid Water Content

View More View Less
  • a Centre de Recherches en Physique de l'Environment, Issy-les-Moulineaux, France
  • | b California Space Institute, Scripps Institution of Oceanography, La Jolla, California
  • | c Centre de Recherches en Physique de l'Environment, Issy-les-Moulineaux, France
© Get Permissions
Full access

Abstract

Vertically integrated atmospheric liquid water content derived from Nimbus-7 Scanning Multichannel Microwave Radiometer (SMMR) brightness temperatures and from GOES-1 Visible and Infrared Spin-Scan Radiometer (VISSR) radiances in the visible are compared over the Indian Ocean during MONEX (monsoon experiment). In the retrieval procedure, Wilheit and Chang&apos algorithm and Stephens' parameterization schemes are applied to the SMMR and VISSR data, respectively. The results indicate that in the 0–100 mg cm−2 range of liquid water content considered, the correlation coefficient between the two types of estimates is 0.83 (0.81– 0.85 at the 99 percent confidence level). The Wilheit and Chang algorithm, however, yields values lower than those obtained with Stephens's schemes by 24.5 mg cm−2 on the average, and occasionally the SMMR-based values are negative. Alternative algorithms are proposed for use with SMMR data, which eliminate the bias, augment the correlation coefficient, and reduce the rms difference. These algorithms include using the Witheit and Chang formula with modified coefficients (multilinear regression), the Wilheit and Chang formula with the same coefficients but different equivalent atmospheric temperatures for each channel (temperature bias adjustment), and a second-order polynomial in brightness temperatures at 18, 21, and 37 GHz (polynomial development). When applied to a dataset excluded from the regressionn dataset, the multilinear regression algorithm provides the best results, namely a 0.91 correlation coefficient, a 5.2 mg cm−2 (residual) difference, and a −2.9 mg cm−2 bias. Simply shifting the liquid water content predicted by the Wilheit and Chang algorithm does not yield as good comparison statistics, indicating that the occasional negative values are not due only to a bias. The more accurate SMMR-derived liquid water content allows one to better evaluate cloud transmittance in the solar spectrum, at least in the area and during the period analyzed. Combining this cloud transmittance with a clear sky model would provide ocean surface insulation estimates from SMMR data alone.

Abstract

Vertically integrated atmospheric liquid water content derived from Nimbus-7 Scanning Multichannel Microwave Radiometer (SMMR) brightness temperatures and from GOES-1 Visible and Infrared Spin-Scan Radiometer (VISSR) radiances in the visible are compared over the Indian Ocean during MONEX (monsoon experiment). In the retrieval procedure, Wilheit and Chang&apos algorithm and Stephens' parameterization schemes are applied to the SMMR and VISSR data, respectively. The results indicate that in the 0–100 mg cm−2 range of liquid water content considered, the correlation coefficient between the two types of estimates is 0.83 (0.81– 0.85 at the 99 percent confidence level). The Wilheit and Chang algorithm, however, yields values lower than those obtained with Stephens's schemes by 24.5 mg cm−2 on the average, and occasionally the SMMR-based values are negative. Alternative algorithms are proposed for use with SMMR data, which eliminate the bias, augment the correlation coefficient, and reduce the rms difference. These algorithms include using the Witheit and Chang formula with modified coefficients (multilinear regression), the Wilheit and Chang formula with the same coefficients but different equivalent atmospheric temperatures for each channel (temperature bias adjustment), and a second-order polynomial in brightness temperatures at 18, 21, and 37 GHz (polynomial development). When applied to a dataset excluded from the regressionn dataset, the multilinear regression algorithm provides the best results, namely a 0.91 correlation coefficient, a 5.2 mg cm−2 (residual) difference, and a −2.9 mg cm−2 bias. Simply shifting the liquid water content predicted by the Wilheit and Chang algorithm does not yield as good comparison statistics, indicating that the occasional negative values are not due only to a bias. The more accurate SMMR-derived liquid water content allows one to better evaluate cloud transmittance in the solar spectrum, at least in the area and during the period analyzed. Combining this cloud transmittance with a clear sky model would provide ocean surface insulation estimates from SMMR data alone.

Save