Principal Investigator: Carl Mears (Remote Sensing Systems of California Corporation)
Passive microwave measurements of Earth from satellites are useful for retrieving a plethora of surface and atmospheric parameters, but are difficult to work with for users without satellite-specific knowledge. Raw satellite measurements are available in a complicated swath geometry that does not line up well with measurements from other satellites or ancillary data that might be needed. By calculating precisely resampled microwave data on fixed Earth grids, along with a set of ancillary data useful for algorithm development, we lower barriers to working with passive microwave data.
- Resample microwave measurements from multiple satellites onto fixed latitude/longitude and polar grids
- Provide a set of ancillary data including land/water fraction, precipitation, and atmospheric parameters from reanalysis in exactly the same grids with the same footprint sizes
- Provide documentation to show how this collection of data can be used to create machine learning datasets
Microwave radiances measured by conically-scanning satellite radiometers are available from Remote Sensing Systems as Level 1B files with measurements arranged in the native swath format. The measurement geometry is complex and it is not trivial to collocate these data between different satellites or with other types of Earth data, such as model output. To facilitate collocation of these microwave radiance with other sources of Earth data, we have accurately resampled the Level 1B radiances onto circular footprints on Earth-referenced latitude/longitude or polar grids. We also provide various types of ancillary data on the same grid resampled to the same footprint size and shape.
To date, we have resampled Advanced Microwave Scanning Radiometer 2 (AMSR2) radiances onto two types of Earth grids with two footprint sizes. The resampling is performed using the Backus-Gilbert method (to get within 3 km of the target locations) combined with 2D interpolation (to exactly place the resampled footprint).
- Grids Available:
- 0.25-degree Latitude/Longitude Grid with global extent
- 25 km EASE2 polar grid for the Northern Hemisphere
- Footprint Sizes Available:
- 30 km circular footprints (10.65,18.7,23.8,36.5 GHz, V and H polarization)
- 70 km circular footprints (6.9,7.3,10.65,18.7,23.8,36.5 GHz, V and H polarization)
We have resampled/regridded a number of ancillary datasets to the same grid. The method used depends on the spatial resolution of the source dataset. These ancillary datasets are useful for developing retrieval algorithms. Depending on the algorithm sought, a given type of ancillary data could be an input parameter or a desired algorithm output. For some examples, see the Jupyter notebook we have developed.
- Ancillary Datasets Available
- Land/Water Fraction from the Moderate Resolution Imaging Spectroradiometer (MODIS)
- Precipitation from Integrated Multi-satellitE Retrievals for GPM (IMERG)
- Skin Temperature from European Centre for Medium-Range Weather Forecasts Reanalysis v5 (ERA5)
- Total Column Water Vapor from ERA5
- Total Column Cloud Water from ERA5
- Vector wind for ERA5
- Ancillary Datasets calculated using a microwave radiative transfer model; nominal values of Earth incidence angles were used
- Ocean surface emissivity for each frequency/polarization, calculated from ERA5 vector winds and Skin Temperature
- Atmospheric parameters for each frequency
- Upwelling Radiance at the Top of Atmosphere (TOA) (TB_UP)
- Downwelling Radiance at the surface
- Transmissivity from the surface to the TOA
- AMSR2 measurements from 2012 to 2021 resampled onto regular latitude/longitude and polar EASE2 grids
- Collocated (in time and location) ancillary data available on the same grids
- Developed a Jupyter notebook to show simple examples of constructing a machine learning dataset from our data collection
For More Information
Remote Sensing Systems: Earth Gridded Microwave Radiance Data Collection (free account required)