IASI:CrIS Bias Study
Wed, Aug 1, 2018
1 IASI and CrIS Bias study
2 Methods
Two types of analysis are undertaken, the SNOs and random full-swath sampling.
2.1 SNOs
There are SNOs for four combinations of sensors: MetOp-A IASI and MetOp-B IASI each paired with Suomi-NPP CrIS and NOAA-20 CrIS (henceforth referred to as IASI.1 and IASI.2 paired with CrIS.1 and CrIS.2). CrIS.1 data are available since 2012 and CrIS.2 data available from 2018. IASI.1 data are available since 2007 and IASAI.2 since 2012. Most of the anaylsis has been performed with CrIS in standard resolution mode but since 2017 full spectral resolution (FSR) spectra are available.
SNOs are obtained for pairs of observations that occur within 20 minutes of each other and are less than 20 km apart. At this separation SNO pairs are gathered at the rate of a few hundred per day.
2.1.1 Temporal and Geographic coverage.
The SNOs occur in a very narrow latitude band centered at 72-deg, there tends to be a seasonal difference of the dynamic range of the northern compared to the southern latitude SNOs as can be discerned from the land/sea coverage. Thus, window channel SNOs tend to have a relatively small dynamic range compared to the AIRS:CrIS SNOs that encompass tropical scenes.
Figure 1: IASI:CrIS SNO geographic and temporal coverage
A typical spectrum of the mean bias between the sensors for the SNOs gathered during Feb to May 2018 is shown in the following figure. For this period there are between 20,000 and 30,000 SNO pairs, depending upon which pairing. The standard error of the mean bias is very small and typically of the order a few milli-Kelvin so is not shown for clarity. The distributions of the pair-wise differences are somewhat Gaussian, espcially in opaque channels, in the window regions they tend to have high kurtosis. Therefore to a first approximation the 95% condifence intervals are reasonably well defined and are of the order a few mill-Kelvin.
Figure 2: IASI:CrIS SNO mean bias
2.2 Random full-swath RTPs
As with SNOs there are in principle four sets of intercomparison that can be made depending upon how to pair the two sensors.
In this analysis, a randomly selected subset of observations from all possible fields of view are gathered, such that equal areas of the Earth are sampled an equal number of times. The important characteristic of these observations is that the same area may have been sampled at very different times, from the same sensors and when compared with the other sensor. In addition, the CrIS and IASI orbits are synchronized to different local solar times; approx equatorial crossing times of ascending phase are 13:30 for CrIS and 21:30 for IASI. Furthermore the CrIS.1 and CrIS.2 are in the same orbit but about 170 degrees out of phase and IASI.1 and IASI.2 are 174 degrees out of phase.
2.2.1 Source data:
- /asl/rtp/rtp_iasi1/random/<YYYY>/iasi1_era_d*_random_fs.rtp_[1,2]
- /asl/rtp/rtp_cris_ccast_lowres/random_fs/<YYYY>/cris_lr_era_d*_random_fs.rtp
2.2.2 Method:
- ~/projects/combined_sensors/stats/iasi_cris_random_means.m
2.2.3 Observations
- 2017 Q1 (Jan - Mar)
- Significant variation of bias IASI minus CrIS over latitude. (see figure) The obs have
not been subset further than by latitude.
