Abstract
Precipitable water (PW) is retrieved over the tropical and subtropical Pacific Ocean from TIROS-N Operational Vertical Sounder (TOVS) infrared and microwave brightness temperatures (BTs) by statistical regression onto collocated Radiosonde Observations (RAOBs). All basic models which use only TOVS observations explain about 75% of the variance. All OLR-enhanced models which also include Outgoing Longwave Radiation (OLR) as a stratification criterion explain about 80% of the variance. A radiative transfer model applied to RAOB soundings shows that collocation errors may explain about 10% of the variance. Given the resultant 80-85% variance explained by all basic models, the statistical method approaches the lower limit accuracy of microwave techniques. All OLR-enhanced models capture more variance and show more detailed features in the PW fields, but they are overly sensitive to calibration data sets and not easily transportable to other time periods. Independent regressions are performed for two data sets, and two best-fit models are chosen. Each model is dominated by different predictors, partly because of the high correlations between TOVS- channels and partly because of the climate changes during the time period. For climatological study of PW, a comprehensive model is developed from the calibration data set for winters of 1982-1990. The model explains 74% of the variance with 0.6 g/CM2 root mean square error. The model explains stationary variance better than transient variance. The model is biased toward the climate mean PW, overestimating in dry regions and underestimating in wet regions. Principal component analyses are performed on TOVS channel BTs and RAOB temperatures and analyze errors. The model errors and RAOB temperatures and dew point temperatures to analyze errors. The model errors are associated with certain atmospheric vertical structures. Wet errors are associated with trade wind inversions, while dry errors are associated with more atmospheric vertical structures. convective-like soundings. TOVS convective-like soundings. TOVS BTs have similar structures for both good and poor estimations so that one cannot anticipate the occurrence of erroneous estimations.
Yin, Min (1994). Information content and reliability of TOVS estimates of precipitable water. Master's thesis, Texas A&M University. Available electronically from
https : / /hdl .handle .net /1969 .1 /ETD -TAMU -1994 -THESIS -Y51.