Satellites, particularly low-Earth-orbiting (LEO) asynchronous platforms, have become the primary vehicle for monitoring global environmental behavior from space because of their capability of global coverage. Satellite data, combined with in situ observations, are increasingly important for global climate and environmental assessment and prediction. The sampling properties of satellites have a great impact on satellite-based data products. Sampling errors due to poor sampling design or improper use of satellite data can cause rapid deterioration in accuracy. Evaluation of sampling errors thus constitutes an integrated part of the uncertainty assessment of environmental assimilation, estimation, and prediction.
The overall goal of this research is to improve our understanding of sampling errors for global observing systems and to enhance our ability to assess the impact of sampling errors on the estimation of global environmental parameters. Ultimately the proposed research will lead to effective design of sampling schemes and quality assurance of satellite-based data products.
To achieve the overall goal, this research proposes to develop general statistical methodology and procedures for exploration and evaluation of sampling errors in various derived parameters from various observing systems. The observing systems considered include single polar and nonpolar platforms, intercalibrated multiple polar and nonpolar satellites, and combinations of in situ point-gauge networks with polar/nonpolar satellites. The derived parameters of interest include the coefficients in the advanced global representations using the spherical harmonics and the empirical orthogonal functions (EOFs). Special attention is paid to small-scale/short-wavelength parameters that are important to the estimation of regional and interseasonal to interannual changes and trends. For efficient representation of regional activities, the research also proposes to develop statistical theory and algorithms of spherical wavelets and investigate sampling errors in the spherical wavelet representations.
The proposed research is exploratory in nature, conducted both analytically, via theoretical analysis, and numerically, via computer simulations and numerical experiments using physical-statistical models and assimilated data. Successful completion of the research is expected to improve our understanding of sampling errors for various important global observing systems and our ability to assess the impact of sampling errors on the estimation of important global environmental parameters. The methodology and procedures developed in the research are expected to be helpful in guiding the data-collection planners to evaluate different sampling designs prior to a data-collecting mission and make optimal choices according to their understanding of the environmental fields to be observed. The research results are also expected to be useful in the uncertainty assessment and quality assurance of satellite-based data products that are now widely used for environmental studies.
This research project in environmental statistics is jointly supported by the MPS Office of Multidisciplinary Activities (OMA) and the DMS Statistics Program.
