A11E-01 INVITED 

Fusing measurements statistically: combining aerosol data from MISR and MODIS

* Cressie, N (ncressie@stat.osu.edu), The Ohio State University, Department of Statistics 1958 Neil Avenue Room 404, Columbus, OH 43210-1247, United States Braverman, A (Amy.J.Braverman@jpl.nasa.gov), Jet Propulsion Laboratory, California Institute of Technology Mail Stop 306-463 4800 Oak Grove Drive, Pasadena, CA 91109-8099, United States Nguyen, H (hnguyen@stat.ucla.edu), University of California, Los Angeles, Department of Statistics 8125 Math Sciences Bldg Box 951554, Los Angeles, CA 90095-1554, United States

We are interested in producing an aerosol data set that provides 1) the best possible representation of aerosol properties, given information from the MISR and MODIS instruments, and 2) quantitative measures of uncertainty associated with that representation. Uncertainties are due to instrument measurement errors, aggregations over space and time arising from different sampling characteristics and footprints, and incomplete data. Our approach is to consider this as a statistical estimation problem. That is, using all information available, find the best statistical estimate of the quantity of interest, say aerosol optical depth (AOD), as a function of location and time. We do this in two steps. First, we use geostatistical smoothing (GS) to estimate the true values of AOD using each instrument's data individually, on a reference grid of locations and times. GS, also known as kriging, is a spatial analog of simple linear regression that accounts for and exploits spatial autocorrelation to produce optimal estimates, or predictions, of unobserved values. Estimates from GS are routinely accompanied by the kriging variance, a formal measure of estimation uncertainty. In the second step, which we call Bayesian Data Fusion (BDF), we form linear combinations of instruments' smoothed estimates at each point of the reference grid. The coefficients for these linear combinations are derived from a statistical model for the relationship between the smoothed data and the true but unobserved values of AOD. BDF not only combines the individual instruments' information in a statistically optimal fashion, but also propagates their uncertainties through the fusion step to produce the desired data set. Both GS and BDF have been used successfully for many years in social- and physical-science applications; their combination in this context offers a coherent way to make inferences with NASA data in the presence of uncertainty.

A11E-02 INVITED 

merging the strengths of different remote sensing techniques for global aerosol data-sets

* Kinne, S (stefan.kinne@zmaw.de), Max-Planck-Institute for Meteorology, Bundesstrasse 53, Hamburg, 20146, Germany

Adequate global distributions of aerosol properties are essential for quality estimate of the aerosol impact on climate. With new aerosol dedicated space sensors during the last decade, regional, seasonal and vertical distributions of aerosol, many different (complementary and/or redundant) aerosol data-sets are offered. Even for the most important aerosol property (the aerosol optical depth AOD) differences among sensor retrievals are commonly on the order of the retrieved value. This is explained in part by retrieval differences, as they are tied to specific sensor capabilities and in part by necessary a-priori assumptions (e.g. aerosol absorption, background reflection). In other words, satellite retrievals have different regional and seasonal strengths. Or alternatively, weaknesses often do not allow for globally (and temporarily) complete coverage. Thus, the overall goal is to combine strengths of individual retrievals to yield a product superior to any single retrieval in accuracy and/or spatial and temporal coverage. The overall concept is to select a particular space-retrieval superiority based on statistical (e.g. bias, error) comparisons to quality references (e.g. statistics from sun-/sky-photometry) and to combine individual regional and seasonal choices into data composites. Space sensors participants for tropospheric aerosol data-sets considered in this study are multi-annual monthly statistics offered by MODIS (collection 5), MISR, TOMS, POLDER and several AVHRR retrievals, while AERONET data represent the quality reference. In addition, to the retrieval composites also improved composites are offered, in which quality references are directly forced (or merged) into the satellite sensor composites for improved accuracy (since even the ‘best' performing satellite retrieval is usually still biased with respect to the quality reference). New global aerosol property composite products tied to measurements are presented and discussed (also in the context of individual sensor products).

A11E-03 INVITED 

Application of Multi-Sensor Fusion to the Aerosol Forecasting Problem

* Reid, J S (jeffrey.reid@nrlmry.navy.mil), Naval Research Laboratory, 7 Grace Hopper St., Stop 2, Monterey, CA 93943, United States Zhang, J (jianglong.zhang@nrlmry.navy.mil), Dept. of Atmospheric Science, University of North Dakota, Grand Forks, ND 58202, Hsu, C (christina.hsu@nasa.gov), NASA Goddard SFC, Mail Code 613, Greenbelt, MD 20771, Christopher, S A (sundar@nsstc.uah.edu), Dept. of Atmospheric Science, Univ. of Alabama, Huntsville, AL 35805, Hyer, E J (edward.hyer@nrlmry.navy.mil), Naval Research Laboratory, 7 Grace Hopper St., Stop 2, Monterey, CA 93943, United States Kuciaskas, A P (arunas.kuciaskas@nrlmry.navy.mil), Naval Research Laboratory, 7 Grace Hopper St., Stop 2, Monterey, CA 93943, United States Westphal, D L (westphal@nrlmry.navy.mil), Naval Research Laboratory, 7 Grace Hopper St., Stop 2, Monterey, CA 93943, United States Kahn, R A (ralph.kahn@nasa,gov), NASA Goddard SFC, Mail Code 613, Greenbelt, MD 20771, Hansen, J A (james.hansen@nrlmry.navy.mil), Naval Research Laboratory, 7 Grace Hopper St., Stop 2, Monterey, CA 93943, United States

The current Earth observing satellite constellation is at a pinnacle, and the near-future schedule suggests that a host of environmental monitoring capabilities will likely be lost over the next five to ten years. The MODIS carrying satellites are nearing the end of their design life, and the NPOESS satellites have been de-scoped and delayed. Many A-Train products, currently taken for granted, will simply not exist to support continuing operational aerosol missions. This situation places the aerosol forecasting community in a particularly difficult position. For air quality systems, the utilization of multiple satellite products and sensors in a data assimilation framework will become a necessity. In this talk we give a brief overview of current capabilities and challenges based on existing sensor products. This includes not only those pertaining to aerosol particle retrievals, but related environmental products that have great potential in improving simulations of 4-dimensional aerosol fields. However, we also show that situations can arise where multi-product fusion and assimilation can actually be detrimental if not interpreted carefully. Given future sensor efficacy and mission plans, possible solutions will then be given on how to maintain our current capabilities into the future. Potential exists not only for fusion of level 2 or 3 products, but also for incorporating numerous level 1b products in single retrievals. Similarly, model results should influence the retrieval process, whether through external meteorological and ensemble input, iterative product data assimilation, or ultimately, radiance assimilation. We see hope for consistency through the Deep Blue algorithm or through combined polar orbiter/geostationary products such as the proposed multi-angle spectropolarimetric imager (MSPI) with GOES-R.

A11E-04 INVITED 

DUE GlobAEROSOL - A multi-instrument satellite aerosol product

* Thomas, G E (gthomas@atm.ox.ac.uk), University of Oxford, Atmospheric, Oceanic and Planetary Physics Clarendon Laboratory Parks Road, Oxford, OX1 3PU, United Kingdom Siddans, R (R.Siddans@rl.ac.uk), Rutherford Appleton Laboratory, Chilton, Didcot, OX11 0QX, United Kingdom Poulsen, C A (C.A.Poulsen@rl.ac.uk), Rutherford Appleton Laboratory, Chilton, Didcot, OX11 0QX, United Kingdom Grainger, R G (r.grainger@physics.ox.ac.uk), University of Oxford, Atmospheric, Oceanic and Planetary Physics Clarendon Laboratory Parks Road, Oxford, OX1 3PU, United Kingdom Navarro, Ø P (operez@gmv.com), GMV Aerospace and Defence, Isaac Newton, 11 P.T.M. Tres Cantos, Marid, E-28760, Spain Kerridge, B J (B.J.Kerridge@rl.ac.uk), Rutherford Appleton Laboratory, Chilton, Didcot, OX11 0QX, United Kingdom

GlobAEROSOL is an ESA Data User Element project to produce a ten year global aerosol product from a range of European satellite radiometers. The GlobAEROSOL products will include data from the ATSR-2 (on board ERS- 2), SEVIRI (on board the Meteosat Second Generation satellites) and AATSR and MERIS (on board Envisat) instruments in both single-sensor products as well as a single merged product. Data will be available on a 10x10 km grid in an orbit by orbit form, as well as in monthly composites, and will cover 1995 -- 2007 date range. This presentation will give an outline of the instruments and algorithms behind the GlobAEROSOL product, with particular emphasis on the approach taken in merging data from different instruments. Example data products will also be presented as well as validation results. http://www.globaerosol.info

A11E-05 

Intercomparison of Atmospheric Aerosol Climatologies Obtained From Passive Satellite Instruments

* Leptoukh, G (Gregory.Leptoukh@nasa.gov), NASA, NASA GSFC, Greenbelt, MD 208771, United States Zubko, V), RSIS, NASA GSFC, Greenbelt, MD 20771, United States Gopalan, A (Arun.Gopalan@nasa.gov), RSIS, NASA GSFC, Greenbelt, MD 20771, United States Remer, L (Lorraine.Remer@nasa.gov), NASA, NASA GSFC, Greenbelt, MD 208771, United States

Atmospheric aerosols show a complex relationship with the climate system. The research of the global effects of aerosol on the climate system is currently in a quantitative phase due to a increasing network of in situ, ground and space-based sensors with some of them in use for long time. In this paper we focus on the spatial and temporal variability of aerosol properties as retrieved by the passive satellite instruments like Terra-MODIS, Terra-MISR, Aqua-MODIS and Parasol-POLDER. Previous studies (Mishchenko et al., 2007) have found that differences between the aerosol climatologies developed from the various passive satellite sensors on both the global and local scale and over both short and long term periods exceeds the corresponding individual uncertainty estimates. For this study we have worked with the latest versions of the Daily Global 1 deg. X 1 deg. Level-3 Aerosol Product from the MODIS, MISR and POLDER instruments. We have derived a global and regional monthly mean- climatology for a few key aerosol parameters (e.g Aerosol Optical Depth, Angstrom Exponent, Fine Mode Fraction). We then looked at the quantitative differences in the long term global and regional statistics of these aerosol parameters. The impact of quality and pixel-weighting on these statistics has been examined and it was found that there can be significant differences in the resulting global aerosol climatologies. Differences with the previous version of MODIS data have been also examined. We also address various statistical issues in working with these aerosol parameters within the framework of the GES DISC Interactive Online Visualization and Analysis System (GIOVANNI). http://giovanni.gsfc.nasa.gov

A11E-06 

Comparing MISR and MODIS Data to Output from the IMPACT Aerosol Transport Model Using the Aerosol Measurement and Processing System

* Braverman, A (Amy.Braverman@jpl.nasa.gov), Jet Propulsion Laboratory, California Institute of Technology, 4800 Oak Grove Drive, Pasadena, CA 91109, United States Penner, J (penner@umich.edu), University of Michigan, 2455 Hayward, Ann Arbor, MI 48109, United States Xu, L (lixum@umich.edu), University of Michigan, 2455 Hayward, Ann Arbor, MI 48109, United States Chuang, C (chuang1@llnl.gov), Lawrence-Livermore National Laboratory, 7000 East Ave., Livermore, CA 94550, United States Wilson, B (Brian.Wilson@jpl.nasa.gov), Jet Propulsion Laboratory, California Institute of Technology, 4800 Oak Grove Drive, Pasadena, CA 91109, United States Manipon, G (Geraldjohn.Manipon@jpl.nasa.gov), Raytheon Corporation, 299 N. Euclid Ave., Pasadena, CA 91101, United States Xing, Z (Zhangfan.Xing@jpl.nasa.gov), Jet Propulsion Laboratory, California Institute of Technology, 4800 Oak Grove Drive, Pasadena, CA 91109, United States Paradise, S (Susan.Paradise@jpl.nasa.gov), Jet Propulsion Laboratory, California Institute of Technology, 4800 Oak Grove Drive, Pasadena, CA 91109, United States

Large-scale comparisons of model output and observational data can be notoriously difficult. Typically, observational data like those from EOS instruments MISR (the Multi-angle Imaging SpectroRadiometer) and MODIS (the Moderate Resolution Imaging Spectrometer) are massive, distributed at different physical locations, and heterogenous: their observation grids are not coincident either with each other or with model grids. The usual strategy for comparing such disparate data sets is to aggregate the high-resolution observations to a coarse common spatial and temporal resolution that matches that of the model. For example, one may compare IMPACT (the Integrated Massively Parallel Atmospheric Chemical Transport model) output, which are daily predictions of aerosol optical depth (aod) by aerosol particle type, on a one-degree spatial grid, to averages of MISR and MODIS optical depths for the same grid cells on the same days. However, this makes no use of the distributions of optical depth provided by both MISR and MODIS in those grid cells. To compare IMPACT predictions to MISR and MODIS, we conduct hypothesis tests by grid cell and day, testing whether IMPACT predicted values could arise from statistical populations represented by the MISR and MODIS data distributions. The Aerosol Measurement and Processing System (AMAPS), a distributed scientific computing environment for aerosol science, makes it possible for us to carry out these tests on a large scale. In this talk, we report the results for a global, full-year (2001) analysis.

A11E-07 

Using Neural Networks for Instrument Cross-Calibration

* Lary, D (David.Lary@umbc.edu), UMBC/GEST NASA/GSFC, University of Maryland Baltimore County 5523 Research Park Drive, Suite 320, Baltimore, MD 21228, United States

Neural networks are non-linear non-parametric learning algorithms that are universal approximators. They have proved very useful to us in a variety of applications (Lary et al., 2004, 2007a,b), from the acceleration of expensive code elements to learning the cross-calibration between large earth observing datasets including atmospheric composition, aerosol optical depth, and vegetation indices. We have been using a variety of networks including feed-forward multi-layer perceptron networks trained with the Levenberg-Marquardt algorithm, and neuro-fuzzy networks. The success of the neural networks largely depends on two factors. First, having a training dataset that adequately spans the parameter space. Second, including the variables that explain the variance in the dataset. If these two criteria are met then the neural networks give excellent results as they are universal approximators. We present several examples of neural network cross-calibration.

Author(s) (2007), Title, Eos Trans. AGU, 88(52), Fall Meet. Suppl., Abstract #####-##.