High Resolution Decadal Gridding of NASA Atmospheric Infrared Sounder (AIRS) Earth Monitoring Instrument.

This week's ebiquity lab meeting will comprise of presentation by PhD candidate, David Chapman.

David will talk on - High Resolution Decadal Gridding of the NASA Atmospheric Infrared Sounder (AIRS) Earth Monitoring Instrument.

Abstract: The NASA Atmospheric Infrared Sounder (AIRS) has been monitoring sun synchronous hyperspectral infrared radiation from Earth's surface and atmosphere operationally since September 2002, making AIRS one of the longest running IR sounders. AIRS has achieved very strong spectral calibration, and the spacecraft has never drifted more than 17KM from it's intended orbital track. AIRS has a 14KM Nadir spatial resolution, and 2378 IR spectral frequencies, and has collected nearly 50 Terabytes of raw orbital radiances over 9+ years. Our goal is to develop a footprint-resolution gridded global Brightness Temperature dataset with climate acceptable stability over the entire 10 year period of observation, using a physically accurate gridding methodology. Our proposed dataset could eventually simplify inter-comparison and assimilation with weather forecast models, and could correct for spatial-biases over terrestrial features at the inter-annual timescale. However, such a dataset requires a physically accurate gridding model, long term stability and statistical validation, as well as a scalable computational method. Towards this goal, we have investigated the Observation Coverage model at the 100km resolution and found significant 30-50% AIRS RMS noise improvement as verified by the 1km MODIS instrument for the 12u and 4u surface widow spectrum over a 2005 winter season grid average. We also have discovered inter-annual spatial biases over terrestrial features such as Mountain Ranges (~1.0K) and Land/Ocean boundaries (~0.5K) that depend exclusively on the gridding methodology. These interannual biases were measured as direct algorithm differences and not validated by MODIS. Surprisingly, the pre-launch "Tophat" optical dataset did not aid in any spatial resolution improvements. We are empirically deriving an alternative optical dataset based on correlation coefficients to nearby MODIS footprints over several days of observation. We have also begun to address the climate applicability question, and produced a prototype 8 year AIRS gridded dataset using the traditional forward methodology at 100KM resolution. We have found a very high yearly correlation coefficients of 0.89 of regional 4u Brightness Temperatures to the NOAA Oceanic Noni Index (ONI) over the 8 year period, as well as a 0.74 monthly correlation coefficient to the GISS global surface temperatures in the 12u with winter and summer correlations of 0.96 and 0.88 respectively over the 8 year period. A simple workqueue based processing system was sufficient to process our 100KM prototype, but cannot meet the data intensive requirements of the proposed 14km dataset. For this reason, we will investigate more scalable processing alternatives including Map Reduce to produce this dataset a small cluster.

Date: March 01 (Thursday)
Time: 01.00 pm - 02.00 pm
Place: ITE 325 - B