In: Wagner W., Szekely, B. (eds.): ISPRS TC VII Symposium - 100 Years ISPRS, Vienna, Austria, July 5-7, 2010, IAPRS, Vol. XXXVIII, Part 7B 
429 
over almost all continents. Values of obtained parameters 
suggest that we should trust AW more in the Eastern US, an area 
well covered by AERONET sites where SR is expected to work 
well, while in Africa, an area poorly covered by AERONET, we 
should trust C005 more (Radosavljevic et al, 2010b). 
4. CONCLUSION 
The reported results provide strong evidence that accurate 
statistical AOD retrieval is possible by problem partitioning 
through competition of local spatio-temporal models 
(Radosavljevic et al, 2008) and developing an ensemble 
statistical model with components specialized for small and for 
large AOD prediction (Radosavljevic et al, 2010a). It was also 
found in our study how to analyze uncertainly of statistical 
AOD retrievals (Ristovski et al, 2009) and how to use this tool 
for optimized placement of ground-based observation sensors 
(Das et al, 2009). A data mining analysis can also reveal the 
conditions when deterministic AOD retrievals can be 
significantly enhanced (Vucetic et al, 2008). Our work in 
progress on structural regression by continuous conditional 
random fields suggests that further major improvements of 
retrieval accuracy can be achieved by unifying multi-source 
retrievals (Radosavljevic et al, 2010b). 
References 
Das, D., Obradovic, Z., Vucetic, S. 2009. Active Selection of 
Sensor Sites in Remote Sensing Applications, Proc. IEEE Int 7 
Conf. Data Mining, December, Miami, FL. 
Das, D., Radosavljevic, V., Vucetic, S., Obradovic, Z. 2008. 
Reducing Need for Collocated Ground and Satellite based 
Observations in Statistical Aerosol Optical Depth Estimation, 
IEEE Int 7 Geoscience and Remote Sensing Symposium, 
Boston, MA. 
Han, B., Vucetic, S., Braverman, A., Obradovic, Z. 2006a. A 
Statistical Complement to Deterministic Algorithms for 
Retrieving Aerosol Optical Thickness from Radiance Data, 
Engineering Applications of Artificial Intelligence, vol. 19, no. 
7, pp. 787-795. 
Han, B., Obradovic, Z, Li, Z., Vucetic, S., 2006b. Data Mining 
Support for Improvement of MODIS Aerosol Retrievals, Proc. 
IEEE Int’l Geoscience and Remote Sensing Symp., Denver, CO. 
Han, B., Vucetic, S., Braverman, A., Obradovic, Z. 2005a. 
Integration of Deterministic and Statistical Algorithms for 
Aerosol Retrieval, Proc. Int’l Conf. Novel Applications of 
Neural Networks in Engineering, Lillie, France, pp. 85-92. 
Han, B., Vucetic, S., Braverman, A., Obradovic, Z. 2005b. 
Construction of an Accurate Geospatial Predictor by Fusion of 
Global and Local Models, Proc. IEEE 8 th Int 7 Conf. 
Information Fusion, B.l 1.2 pp. 1-8, Philadelphia, PA. 
Jeong, M.-J., Li, Z., Chu, D. A., Tsay, S.-C. 2005. Quality and 
Compatibility Analyses of Global Aerosol Products Derived 
from the Advanced Very High Resolution Radiometer and 
Moderate Resolution Imaging Spectroradiometer, J. Geophys. 
Res., 110, D10S09. 
Levy, R. C., Remer, L. A., Martins, L.V., Kaufman, Y.J., Plana- 
Fattori A., Redemann, J., Wenny, B. 2005. Evaluation of the 
MODIS Aerosol Retrievals over Ocean and Land during 
CLAMS, J. Atmos. Sci., vol. 62(4), pp. 974-992. 
Obradovic, Z, Han, B., Xu, Q., Li, Y., Braverman, A., Li, Z. 
Vucetic, S. 2006. Data Mining Support for Aerosol Retrieval 
and Analysis - Project Summary, NASA Data Mining 
Workshop, Pasadena, CA. 
Radosavljevic, V., Vucetic, S., Obradovic, Z. 2010a. A Data 
Mining Technique for Aerosol Retrieval Across Multiple 
Accuracy Measures, IEEE Geoscience and Remote Sensing 
Letters, vol. 7, no.2, pp. 411-415. 
Radosavljevic, V., Obradovic, Z., Vucetic, S. 2010b. 
Continuous Conditional Random Fields for Regression in 
Remote Sensing, Proc. 19 th European Conf. on Artificial 
Intelligence, Lisbon, Portugal. 
Radosavljevic, V., Vucetic, S., Obradovic, Z. 2009. Reduction 
of Ground-Based Sensor Sites for Spatio-Temporal Analysis of 
Aerosols, Proc. 3 rd International Workshop on Knowledge 
Discovery from Sensor Data at the 15 th ACM SIGKDD Conf. 
Knowledge Discovery and Data Mining, Paris, France. 
Radosavljevic, V., Vucetic, S., Obradovic, Z. 2008. Spatio- 
Temporal Partitioning for Improving Aerosol Prediction 
Accuracy, Proc. Eight SIAM Int 7 Conf. Data Mining, Atlanta, 
GA. 
Ramanathan, V., Crutzen, P.J., Kiehl, J.T., Rosenfeld, D. 2002. 
Atmosphere - Aerosols, climate, and the hydrological cycle, 
Science, vol. 294, pp. 2119-2124. 
Remer, L. A., Tanre, D. , Kaufman, Y. J., 2006. Algorithm for 
Remote Sensing of Tropospheric Aerosol from MODIS for 
Collection 005, http://modis-atmos.gsfc.nasa.gov/atbd02.pdf. 
Ristovski, K., Vucetic, S., Obradovic, Z. 2009. Evaluation of a 
Neural Networks based Approach for Aerosol Optical Depth 
Retrieval and Uncertainty Estimation, Proc. Int’l Conf. on 
Space Technology, Thessaloniki, Greece. 
Vucetic, S., Han, B., Mi, W., Li. Z., Obradovic, Z. 2008. A 
Data Mining Approach for the Validation of Aerosol Retrievals, 
IEEE Geoscience and Remote Sensing Letters, vol. 5, no. 1, pp. 
113-117. 
Xu, Q., Han, B., Li, Y., Braverman, A., Obradovic, Z., Vucetic, 
S. 2005. Improving Aerosol Retrieval Performance by 
Integrating AERONET, MISR, and MODIS Data Products, 
Proc. IEEE 8 th Int’l Conf. Information Fusion, B.l 1.3 pp. 1-8, 
Philadelphia, PA. 
Zhuang, W., Radosavljevic, Han, B., Obradovic, Z., Vucetic, S. 
2008. Aerosol Optical Depth Prediction from Satellite 
Observations by Multiple Instance Regression, Proc. Eight 
SIAM Int’l Conf. Data Mining,, Atlanta, GA. 
Acknowledgement 
This work was supported in part by the National Science 
Foundation under Grant IIS-0612149.