Menu Close

Atmospheric Motion Vector Bias and Uncertainty

Figure: Vertical profiles of co-located LEO AMVs and RAY (red) and MIE (blue) winds. The top row shows the Arctic (north of 60° N), (a) mean AMV HLOSV (solid lines), Aeolus HLOSV (long dashed lines; m s−1), and mean AMV wind speed (short dashed lines; m s−1), (b) MCDs (solid), SDCDs (short dashed), and AMV HLOSV error, as represented by SDCD–Aeolus L2B uncertainty (long dashed; m s−1), and (c) co-location counts. Panels (d–f) are as in panels (a–c) but for the Antarctic (south of 60° S). Colored open circles indicate levels where MCDs are statistically significant at the 95 % level (p value < 0.05), using the paired Student’s t test. Vertical zero lines are displayed in the center panels in black. Levels with observation counts > 25 are plotted.
Figure: Vertical profiles of co-located LEO AMVs and RAY (red) and MIE (blue) winds. The top row shows the Arctic (north of 60° N), (a) mean AMV HLOSV (solid lines), Aeolus HLOSV (long dashed lines; m s−1), and mean AMV wind speed (short dashed lines; m s−1), (b) MCDs (solid), SDCDs (short dashed), and AMV HLOSV error, as represented by SDCD–Aeolus L2B uncertainty (long dashed; m s−1), and (c) co-location counts. Panels (d–f) are as in panels (a–c) but for the Antarctic (south of 60° S). Colored open circles indicate levels where MCDs are statistically significant at the 95 % level (p value < 0.05), using the paired Student’s t test. Vertical zero lines are displayed in the center panels in black. Levels with observation counts > 25 are plotted.
Figure: Vertical profiles of co-located LEO AMVs and RAY (red) and MIE (blue) winds. The top row shows the Arctic (north of 60° N), (a) mean AMV HLOSV (solid lines), Aeolus HLOSV (long dashed lines; m s−1), and mean AMV wind speed (short dashed lines; m s−1), (b) MCDs (solid), SDCDs (short dashed), and AMV HLOSV error, as represented by SDCD–Aeolus L2B uncertainty (long dashed; m s−1), and (c) co-location counts. Panels (d–f) are as in panels (a–c) but for the Antarctic (south of 60° S). Colored open circles indicate levels where MCDs are statistically significant at the 95 % level (p value < 0.05), using the paired Student’s t test. Vertical zero lines are displayed in the center panels in black. Levels with observation counts > 25 are plotted.

ESSIC/CISESS Scientists Katherine Lukens (a former CISESS grad student), Kayo Ide, Hui Liu, and Ross Hoffman have a new article in the journal Atmospheric Measurement Techniques about their work with the NOAA/NESDIS Office of Projects, Planning, and Acquisition (OPPA) Technology Maturation Program (TMP). The need for highly accurate atmospheric wind observations is a high priority in the science community, particularly for numerical weather prediction (NWP). To address this need, this study leverages Aeolus wind lidar level-2B data provided by the European Space Agency (ESA) as a potential comparison standard to better characterize atmospheric motion vector (AMV) bias and uncertainty. 

The researchers compare AMV products from geostationary (GEO) and low Earth orbiting (LEO) satellites with reprocessed Aeolus horizontal line-of-sight (HLOS) global winds from two observing modes, namely Rayleigh-clear (RAY; derived from the molecular scattering signal) and Mie-cloudy (MIE; derived from the particle scattering signal), observed in August–September 2019. As shown in other comparison studies, the level of agreement between AMV and Aeolus wind velocities (HLOSVs) varies with the AMV type, geographic region, and height of the co-located winds, as well as with the Aeolus observing mode. In terms of global statistics, quality controlled (QC’d) AMVs and QC’d Aeolus HLOSVs are highly correlated for both observing modes. Aeolus MIE winds are shown to have great potential value as a comparison standard to characterize AMVs, as MIE co-locations generally exhibit smaller biases and uncertainties compared to RAY co-locations. Aeolus RAY winds contribute a substantial fraction of the total standard deviation of co-location differences (SDCDs) in the presence of clouds where co-location/representativeness errors are also large. Stratified comparisons with Aeolus HLOSVs are consistent with known AMV bias and uncertainty in the tropics, NH extratropics, the Arctic, and at mid- to upper-levels in clear and cloudy scenes. AMVs in the SH/Antarctic generally exhibit larger-than-expected SDCDs, most probably due to larger AMV height assignment errors and co-location/representativeness errors in the presence of high wind speeds and strong vertical wind shear, particularly for RAY comparisons.

Lukens is a Post Doctoral Associate at the University of Maryland/ESSIC/CISESS. Her research background in large-scale atmospheric dynamics and winter storm track impacts has prepared her for her current work that focuses on the intercomparison and verification of new and existing wind datasets (e.g., Aeolus, near-space balloon networks, AMVs, aircraft, rawinsondes) and their impact on global NWP. Additionally, she is a co-architect and manager of a novel wind archive available for public use: https://bin.ssec.wisc.edu/wind-datasets (check the website often for updates). Please contact her with any questions at katherine.lukens@noaa.gov.

Ide is an Associate Professor in the Department of Atmospheric & Oceanic Science, Center for Scientific Computation & Mathematical Modeling, and Institute for Physical Science & Technology, along with Earth System Science Interdisciplinary Center. With background in engineering, theoretical fluid dynamics and applied mathematics, she studies earth system sciences from interdisciplinary perspectives. Her primary research interests are scientific prediction for earth systems and mixing & transport in geophysical fluids. For more information, please visit her website.

Liu is an associate research scientist at ESSIC/CISESS. His research focuses on impact optimization of satellite observations (Aeolus space-based lidar winds, AMVs, and radio occultations etc.) on regional and global weather forecasts.

To access the article, click here: “Exploiting Aeolus Level-2B winds to better characterize atmospheric motion vector bias and uncertainty”.

Posted in Highlights

Related Posts