Quantitative Evaluation of Uncertainty in Water Vapor Atmospheric Motion Vectors and Implications for Data Assimilation and OSSEs

Measurements of the three dimensional distribution of horizontal winds have been highlighted as a priority for future space-based observing systems. While space-borne lidar is commonly viewed as the gold standard for clear-air measurement of winds, there are inherent limitations on spatial and temporal sampling. Atmospheric motion vectors (AMVs) derived from sequences of images (e.g., of clouds, water vapor, and/or other trace gases) constitute a complementary source of wind information that can be obtained over larger areas and with higher frequency in time. Prior to design of a new AMV-focused mission, it is important to assess the information contained in the measurements.

Observing system simulation experiments (OSSEs) have been used to quantify the anticipated impact of a new set of measurements on weather prediction. Forecast OSSEs are useful, in that they measure the effectiveness of a set of measurements in the context of the current global observing system. However, effective use of forecast OSSEs depends on careful specification of measurement uncertainty. In the case of AMVs, there is uncertainty in the observed radiances, the retrieval methodology used to estimate water vapor from space, and the procedure used to track water vapor features.

We have designed a set of OSSEs to:

1. Evaluate spatial and temporal sampling for potential future AMV missions
2. Quantify uncertainty in AMV retrievals from various measurement platforms
3. Assess the impact of AMVs in a forecast OSSE context

We find that AMV uncertainties are state-dependent, and are a strong function of the underlying water vapor and wind fields. This state-dependence can be quantified, and used to more accurately represent measurement errors in a forecast OSSE.