The Advanced Baseline Imager (ABI) is the primary instrument onboard GOES-R for imaging Earth’s weather, climate,
and environment and will be used for a wide range of applications related to weather, oceans, land, climate, and hazards
(fires, volcanoes, hurricanes, and storms that spawn tornados). It will provide over 65% of all the mission data products
currently defined. ABI views the Earth with 16 different spectral bands, including two visible channels, four nearinfrared
channels and ten infrared channels at 0.5, 1, and 2 km spatial resolutions respectively. For most of the
operational ABI retrieval algorithms, the collocated/co-registered radiance dataset are at 2 km resolution for all of the
bands required. This requires down-scaling of the radiance data from 0.5 or 1 km to 2 km for ABI visible and near-IR
bands (2 or 1, 3 & 5 respectively), the reference of 2 km is the nominal resolution at the satellite sub-point. In this paper,
the spatial resolution characteristic of the ABI fixed grid level1b radiance data is discussed. An optimum interpolation
algorithm which has been developed for the ABI multiple channel radiance down-scaling processing is present.
Satellite observation collocation algorithms are generally used to spatially match observations or products from different
satellite systems. The spatially matched and integrated satellite datasets are commonly used in integrated retrievals,
satellite instrument inter-calibration and satellite observation validation. Instrument physical based collocation
algorithms are developed at NOAA/NESDIS/STAR to support the development of the satellite observation integration
system. The algorithms are applied within the Geostationary satellite & Polar satellite (GEO-LEO) integration system
for IASI/SEVRI and will applied in the future CrIS/GOES-R observation integration system. In this paper, the details of
the algorithms for IASI/SEVERI and AIRS/SEVIRI collocation are described and some results for both are presented.
Traditional cloud clearing methods utilize a clear estimate of the atmosphere inferred from a microwave sounder to extrapolate cloud cleared radiances (CCR's) from a spatial interpolation of multiple cloudy infrared footprints. Unfortunately, sounders have low information content in the lower atmosphere due to broad weighting functions, interference from surface radiance and the microwave radiances can also suffer from uncorrected side-lobe contamination. Therefore, scenes with low altitude clouds can produce errant CCR's that, in-turn, produce errant sounding products. Radiances computed from the corrupted products can agree with the measurements within the error budget making detection and removal of the errant scenes impractical; typically, a large volume of high quality retrievals are rejected in order to remove a few errant scenes. In this paper we compare and contrast the yield and accuracy of the traditional approach with alternative methods of obtaining CCR's. The goal of this research is three-fold: (1) to have a viable approach if the microwave instruments fail on the EOS-AQUA platform; (2) to improve the accuracy and reliability of infrared products derived from CCR's; and (3) to investigate infrared approaches for geosynchronous platforms where microwave sounding is difficult. The methods discussed are (a) use of assimilation products, (b) use of a statistical regression trained on cloudy radiances, (c) an infrared multi-spectral approach exploiting the non-linearity of the Planck function, and (d) use of clear MODIS measurements in the AIRS sub-pixel space. These approaches can be used independently of the microwave measurements; however, they also enhance the traditional approach in the context of quality control, increased spatial resolution, and increased information content.
Today, most Numerical Weather Prediction (NWP) centers are assimilating cloud-free radiances. Radiances from the Atmospheric Infrared Sounder have been directly assimilated in NWP models with modest positive impacts. However, since only 5% percentage of AIRS fields of view (fovs) are cloud-free, only very small amounts of the data in the lower troposphere are assimilated. (Note that channels in the mid-upper stratosphere are always assimilated since they are never contaminated by clouds.) The highest vertical resolving power of AIRS is in the lower troposphere. To further improve forecast skill we must increase the use of channels in the lower troposphere. This can be accomplished by assimilating cloud-cleared radiances, which has a yield of about 50%. Since cloud-cleared radiance may have residual cloud contamination and forecast accuracy is very sensitive to the accuracy of the input observations, a technique has been developed to use the 1 km infrared channels on the Moderate Resolution Imaging Spectroradiometer (MODIS) to quality control the cloud-cleared radiances derived from an array of 3 x 3 high spectral infrared sounder AIRS 14 km fovs. This is accomplished by finding MODIS clear radiances values within the AIRS field of view. The MODIS clear radiances are compared to cloud-cleared AIRS radiances that have been convolved to the MODIS spectral resolution. Our studies have found that the cloud-cleared radiances error statistics are very similar to cloud-free (clear) when MODIS data are used to remove potential outliers in the population of AIRS cloud-cleared radiances.
A near real-time AIRS processing and distribution system is fully operational at NOAA/NESDIS/ORA. The AIRS system went though three separate production phases: design and development, implementation, and operations. The design and development phase consisted of two years of preparation for the near real-time AIRS data. The approach was to fully emulate the AIRS measurement stream. This was accomplished by using a forecast model to represent the geophysical state and computation of simulated AIRS measurements using the characteristics of the AIRS channels. The preparation included file format development and the creation of a program to subset the radiance and product data. The implementation phase lasted over a year and involved utilizing AIRS/AMSU/HSB simulated data quasi-operationally. This simulated data was placed into deliverable files and distributed to the customers for their pre-launch preparations. The operational phase consisted of switching the simulation system to real data and is the current system status. Details of what went right and wrong at each production phase will be presented. This methodology eased the transition to operations and will be applied to other advanced sounders such as IASI and CrIS.
Development and testing of the IASI processing and distribution system is currently ongoing at NOAA/NESDIS/ORA. Level 1C data for 8461 channels will be available to NESDIS/NOAA from EUMETSAT shortly after MetOp 1 launch (currently scheduled for October 2005). Prior to launch, a simulation system will provide pseudo near-real time data for system testing and refinement. This will allow for a smooth and immediate system transition to the actual data processing when it becomes available. The ingested EUMETSAT level1C data will be subset both spectrally and spatially and then placed into BUFR format for a number of products including: (1) Level 1C (calibrated, apodized, and navigated) brightness temperatures, (2) cloud-cleared radiances, and (3) PCA reconstructed radiances. The subset level 1C data will be delivered within three hours of observation. System validation will consist of comparing the products to collocated radiosonde observations and model forecasts.