Retrieval Algorithm Of Aerosol Infrared Radiative Properties Over Land From AIRS

Atmospheric aerosol, as a major factor of uncertainties in climate change, can affect the radiative energy balance of earth-atmosphere system through its direct, indirect, and semi-direct radiative effects, it plays an important role in the global and regional radiative forcing by modulating the climate system balance. At the thermal infrared wavelength region, the observed radiance in the top of atmosphere (TOA) is affected by the mineral dust, the absorption of atmospheric gases and water vapor. Aerosol detection methods using visible wavelengths are well established, however, the study at infrared wavelength is still limited. These methods have been used previously to detect the presence of atmospheric aerosols over ocean. In contrast, the accuracy of the relative detection over land is affected by several uncertain factors. Given the reasons above, it is important to develop a retrieval algorithm system over land for detecting the distribution of different types of aerosol and studying its radiative optical properties at infrared wavelength. Based on high spectral resolution infrared sensor observation and radiative transfer model computation, we have developed an unique retrieval algorithm suitable for the ultra-spectral data processing of the Atmospheric Infrared Sounder (AIRS) over land, and it can be applied to other similar current and future sensor counterparts as well. We also have developed a principal component analysis for the ultra-spectral data to filter unwanted random noise and to enhance aerosol/dust retrieval quality.Guided by radiative transfer modeling of the effects of dust (aerosol) on satellite thermal infrared radiance by many different imaging radiometers, we present the aerosol-effected satellite radiative signal changes in TOA. The simulation of TOA radiance for Infrared Atmospheric Sounding Interferometer (IASI) is performed by using the RTTOV fast radiative transfer model. The model computation is carried out with setting representative geographical atmospheric models and typical default aerosol climatological models under clear sky condition. The radiative differences (in units of equivalent black body brightness temperature differences, BTDs) between simulated radiances without consideration of the impact of aerosol (Aerosol-free) and with various aerosol models (Aerosol-modified) are calculated for the whole IASI spectrum between15.5and 3.6μm. The comparisons of BTDs are performed through11aerosol models in5classified atmospheric models. The results show that the Desert aerosol model has the most significant impact on IASI spectral simulated radiances than the other aerosol models (Continental, Urban, Maritime types and so on) in Mid-latitude Summer, contributing to the mineral aerosol components contained. The value of BTDs could reach up to1Kelvin at peak points. The atmospheric window spectral region between900and1100cm-1(9.1～11.1μm) is concentrated after the investigation for the largest values of aerosol-affected radiance differences. The IASI highest window peak-points channels (such as9.4and10.2μ.m) are obtained finally, which are the most sensitive ones to the simulated IASI radiance.Radiances observed by spectrum of AIRS instrument, and the simulated radiances, transmission by a fast radiative transfer model (RTTOV) are applied for the dust detection over land area of China. The study was made for a case study on6March2009when a heavy dust event occurred over northern of China. In infrared spectral range, several AIRS Channel-pairs are chosen to obtain the channels measured residual (Channels brightness temperature differences). After the comparison among these Channel-pairs, a proper pair which is around9.3and10.9μm, is used for the dust detection. The surface, cloud, atmospheric dust and surface dust are detected and masked. The channel-pairs could be utilized for the infrared aerosol optical depth retrieval over China from AIRS. We also examined the infrared radiative forcing from the dust aerosol and clouds, the results showed that the ice cloud have maximum impact on AIRS observed radiance. The dust effects radiance at Channel809(9.3μm) and Channel593(10.9μm) with a significant large V-shaped absorption feature in atmospheric window spectral region (800～1200cm-1).Principal Component Analysis (PCA) is a powerful tool for filtering random noise and data compression of hyperspectral satellite spectra. PCA produces filtered radiance spectra that have a large majority of the random noise removed. Our approach for dependent set PCA of radiance spectra from AIRS is presented without using "noise normalization" and with "noise normalization". PCA was performed separately for each AIRS Level1B granule. The reconstruction errors were examined by comparing to spectral noise equivalent temperature difference (NEDT) at250K computed from AIRS on-board blackbody data. The results have showed that, the range of spectra with highest noise is between12.5and15.3μm, the channels with highest noise are14.3μm and15.3μm. The proper range of NPC (Number of PCs) utilized for PCA in AIRS granule radiance is30～50. The PCA used "noise normalization" has the important effect of down-weighting those channels at the long wavelength region (700～1200cm-1) with high noise. The PC filter is very efficient to represent sensor noise opposed to a wide range of real atmospheric and surface variability. Observed cases from AIRS granules in March2009over land area of China were concentrated for studying, by utilizing AIRS radiance products, the Moderate Resolution Imaging Spectroradiomater (MODIS) aerosol products, and the Cloud-Aerosol Lidar and Infrared Pathfinder Satellite Observations (CALIPSO) aerosol layer products. The simulated AIRS observed radiances were computed by RTTOV. The infrared (10.3μm) aerosol optical depth (AOD) was retrieved. The method developed relies on the construction of a look-up-tables of aerosol loading for two typical aerosol climatological model (Desert and Asian dust), and look-up-table was built for a selection of clear sky atmospheric situation. The comparison was conducted between AIRS retrieved AODs and visible (0.55μm) AOD from MODIS during the period of AIRS granule observation. Through the analysis of results, it is showed the method based on AIRS Channel-Pairs BTDs, which channels are most sensitive ones to AIRS TOA radiance in infrared atmospheric window, could be used availably for the detection of aerosol and cloud spatial distribution in AIRS observed granules. The retrieved results for the cases show a good agreement between MODIS AODs and AIRS AODs distribution, especially for the high AOD value. The average ratio between visible and infrared AOD is0.27, which stands for the ratio between visible and infrared extinction coefficient. The aerosol model Desert can be applied to the AOD retrieval study under the regular atmospheric aerosol background without dust event, and the Asian dust is very useful for the dust detection and AOD retrieval during the period when the dust storm event happens.