- Solar System Exploration Data Services Office (690.1)
- Astrochemistry Laboratory (691)
- Planetary Systems Laboratory (693)
- Planetary Magnetospheres Laboratory (695)
- Planetary Geology, Geophysics, and

Geochemistry Laboratory (698) - Planetary Environments Laboratory (699)

- High Energy Astrophysics Science Archive Research Center Office (660.1)
- Office of Scientific Computing (660.2)
- Instrument Development Group (660.3)
- Astroparticle Physics (661)
- X-ray Astrophysics (662)
- Gravitational Astrophysics (663)
- Observational Cosmology (665)
- ExoPlanets and Stellar Astrophysics (667)

- Global Modeling and Assimilation Office (610.1)
- Global Change Data Center (610.2)
- Field Support Office (610.W)
- Goddard Institute for Space Studies (611)

- ASSISTANT RESEARCH SCIENTIST

- 301.614.6322

- Org Code: 612

- NASA/GSFC
- Mail Code: 612
- Greenbelt , MD 20771

02/01/2013 -
02/01/2016
Expectation Maximization Analysis to Improve the Consistency and Accuracy of GPM Combined Retrievals ,
NASA

A particular challenge in deriving combined precipitation retrievals is the heterogeneous information content across the GPM radar swath. Specifically, in a portion of the radar-swath around the nadir dual-frequency radar observations will be available, while in a roughly equal portion of the swath off-nadir only single frequency radar observations will be available. Given the incomplete knowledge regarding the space-time variability of particle size distributions, GPM dual-frequency radar retrievals will not necessarily be consistent with GPM single-frequency radar retrievals. These potential radar retrieval inconsistencies may translate into GPM combined retrieval inconsistencies and inaccuracies. Although ad-hoc adjustments of outer swath combined retrievals based on inner swath combined retrievals can significantly reduce the potential systematic differences, only a rigorous analysis would warrant such a reduction to the level that both inner swath and outer swath combined precipitation retrievals can be used in climate and microphysical studies. The alternative of ignoring the outer swath retrievals in the favor of inner swath retrievals would significantly increase sampling uncertainties and is not acceptable. In this project, we develop a statistical methodology to improve the consistency and the accuracy of GPM combined precipitation estimates across the radar scan. The methodology is directly applicable to the official GPM combined algorithm implemented by the PI in the past under the PMM auspices.

; PI

; PI

06/24/2011
Perturbed 3-D Monte Carlo method for inversion of high frequency microwave radiometer observations ,
NASA

In various situations, one dimensional radiative transfer models are not adequate in correctly quantifying the relations between geophysical variables and satellite observations. For example, existing studies suggest that the simulation of 85-GHz (and higher) brightness temperatures in convective regions using one dimensional radiative transfer models may be subject to large errors. Three dimensional models are necessary in such situations. Therefore, three dimensional (3D) radiative transfer (RT) models and effective methodologies to invert them (i.e. determine the underlying geophysical variables from their outputs) are necessary. Computationally feasible 3D RT models can be derived based on the Monte Carlo approach.

In this project, an efficient methodology to determine the sensitivity of 3D Monte Carlo RT models to the underlying geophysical variables using a perturbation method is developed. Because most of the computational effort in Monte Carlo approaches resides in generating points in the solution-definition domain where simple functions need to be evaluated, it is possible to evaluate perturbed functions with little additional cost and thus economically derive the sensitivity of 3D Monte Carlo models to various geophysical variables of interest.

; PI

In this project, an efficient methodology to determine the sensitivity of 3D Monte Carlo RT models to the underlying geophysical variables using a perturbation method is developed. Because most of the computational effort in Monte Carlo approaches resides in generating points in the solution-definition domain where simple functions need to be evaluated, it is possible to evaluate perturbed functions with little additional cost and thus economically derive the sensitivity of 3D Monte Carlo models to various geophysical variables of interest.

; PI