Abstract
Alternatives to the retrieval techniques applied by NESDIS operations to derive the FGGF Level IIa soundings are examined. A physical iterative retrieval technique is compared to the operational statistical method, and the influence of using higher resolution subsets of the original infrared observations is examined. These alternatives are evaluated using TIROS-N observations from a January 1980 period over the conventional data-rich region of the United States. The evaluations involve colocation statistics and 700–300 mb thickness difference fields. The initial tests using operational (9 × 7 HIRS/2 fields of view) resolution show that the physical iterative retrieval makes substantial corrections to climatological first guesses, but only minor corrections to a first guess based on the operational soundings. Colocation statistics and 700–300 mb thickness difference fields indicate that the physical retrieval method does not offer significant improvements over the FGGE operational soundings. As in the operational soundings, there is a tendency for the sounding errors to be synoptically correlated with troughs ton warm and ridges too cold, thus reducing thermal gradients.
In an attempt to improve the thermal gradient information, the physical iterative method (using the operational sounding first guess) was also employed to retrieve soundings based on radiances obtained from higher (3 × 3 HIRS/2 fields of view) resolution. Four different subsets of the 3 × 3 sounding sets were studied with varying horizontal resolutions and with and without manual editing. Each set shows some improvement over the 9 × 7 retrievals, particularly through a reduction of the bias in the low and midtroposphere. Further analysis reveals that the improvement in retrieval accuracy is sounding-type dependent, with only the 3 × 3 clear retrievals showing definite improvement over 9 × 7 retrievals for this case. The 700–300 mb thickness fields obtained from the 3 × 3 FOV soundings also show synoptically correlated errors.
