Skip to main content

Principal Investigator (PI): Michael Bosilovich, NASA's Goddard Space Flight Center

Reanalyses have proven a useful tool in studying weather and climate. Often the evaluation of reanalyses and research investigations stop at standard output variables. However, there is a wealth of information from the data assimilation statistics that goes unused owing to the size and complexity of the data files and formats. Often, users have difficulty finding out which and how many observations are even present in their region of interest. Yet, the presence and quality of observations directly affect the reanalyses data.

Here, we propose to evaluate, validate, and deliver the assimilated observations processed in the Modern Era-Retrospective Analysis for Research and Applications-2 (MERRA-2) reanalyses along with key statistics derived from the data assimilation, including forecast departure, root mean square (RMS) of the forecast departure, number of data counts and, for the satellite radiances, bias correction. The observations will include all of the varied conventional observations assimilated in MERRA-2, including temperature and wind from radiosondes and commercial aircraft, surface station pressure, and ship and buoy temperature. There are also numerous satellite-retrieved observations in the system such as atmospheric motion vector winds and ocean surface winds. Satellite radiances make up the bulk of assimilated data. These are also provided, by channel and instrument, including Microwave Limb Sounder (MLS), ARMS, High-resolution Infrared Radiation Sounder (HIRS), Microwave Sounding Units (MSU), Special Sensor Microwave Imager (SSM/I), and Advanced Microwave Sounding Unit-A (AMSU) to name a few.

These observations and their assimilation statistics will be binned to a regular grid that is similar to MERRA-2, at 6-hourly intervals following the data assimilation cycles. Incorporating the hyperspectral instruments, Atmospheric Infrared Sounder (AIRS) and Cross-track Infrared Sounder (CrIS) presents numerous challenges that we will consider in this project. In addition, MERRA-2 is the first reanalysis to incorporate assimilated and interactive aerosols, so that we will also include the assimilated aerosol optical depth observations, e.g., Moderate Resolution Imaging Spectroradiometer (MODIS) and Multi-angle Imaging SpectroRadiometer (MISR).

The proposed data set will serve numerous user communities and purposes including:

  • Research using MERRA-2, identifying the presence and quantity of observations.
  • Model development for the MERRA-2 model.
  • General model development (global gridded observations, including a multitude of satellite radiances, that can be used for model evaluations).
  • Review of the quality and consistency of the assimilated observations.
  • Support for data fusion activities; since the multitude of data are provided in a common framework, data fusion methodologies could be easily tested, prior to implementing more accurate methods with the native data formats.
  • Provide a reference point for forecast departures from observations for MERRA-2, to demonstrate improvements in subsequent reanalyses.

At this time, reanalysis projects do not routinely provide the assimilated observations as diagnostic output data. By synthesizing the multitude of multiplatform observations and the assimilation statistics into a common format, we can provide detailed insight into the inner working of the reanalysis and easily accessible information on the quantity and quality of the assimilated observing system at any point in space and time. Such a data set will augment the dynamically consistent global MERRA-2 data. This project will build the ground work to turn the processing of the gridded observations into a routine product from future Goddard Earth Observing System Model, Version 5 (GEOS-5) Data Assimilation System products.