These web-pages represent the initial results of a project started by Rob Hargreaves (now at Harvard/Smithsonian), and continued by Anu Dudhia. Since we haven't yet published anything on this, the following is a brief summary.
The Scaled Linear Retrieval applies the following operations to IASI L1C spectra
| xstd | = | g.y | Applies a standard linear retrieval for column amount xstd |
| s | = | f.y | Similarly determines scale factor s from IASI spectra |
| x | = | xstd/s | Evaluate scaled linear retrieval |
There are several different methods for evaluating the standard linear retrieval (see Walker et al, 2011) but these are generally based on the assumption of a fixed Jacobian, ie that the depth of an absorption feature depends only on absorber amount and not on the atmospheric state. The problem is often highlighted in (the unusual) situations when the atmosphere is warmer than the earth's surface, so that molecular signatures appear as emission rather than absorption lines, and consequently returning a 'negative' concentration.
The scaled linear retrieval is an attempt to allow for the influence of the atmospheric state on the Jacobian. This derives a scaling factor from the IASI spectra themselves rather than attempting to utilise any secondary products such as L2 temperature profiles. It can be thought of as applying a standard linear retrieval to a well-mixed molecule (such as CO2) whose actual concentration is known, hence yielding a correction factor which is assumed to also apply to the original target molecule.
The plots on these web-pages represent the median value of the results in 2°lat,lon boxes, having first screened the data to remove any pixels flagged as more than 50% cloud-contaminated and near the edges of the IASI swath (zenith angles beyond 45°). The pages currently just show results from IASI-A (both day and night passes combined).
Any queries on the data should be addressed to anu.dudhia@physics.ox.ac.uk