2007 JCSDA Seminars

With the adjoint of a data assimilation system (forecast model plus analysis scheme), the impact of any or all assimilated observations on measures of forecast or analysis skill can be estimated accurately and efficiently. The approach is especially well suited for assessing the impact of hyper-spectral satellite instruments on numerical weather forecasts because it easily allows aggregation of results in terms of individual data types, channels or locations, all computed simultaneously based on a single pass of the adjoint system.

The NASA Global Modeling and Assimilation Office (GMAO) has developed the adjoint of its GEOS-5 atmospheric data assimilation system, consisting of the GEOS-5 finite volume atmospheric model and Gridpoint Statistical Interpolation (GSI) analysis scheme developed at the National Centers for Environmental Prediction (NCEP). In this study, the impacts of various observing systems, including the Atmospheric Infrared Sounder (AIRS), are examined during July 2005 and January 2006. It is found that both conventional and satellite observations contribute significantly to the reduction of forecast errors, with asymmetries in the magnitudes of their impacts depending on the season and hemisphere. Map views of these impacts reveal possible deficiencies in the usage of some observation types. The adjoint- based impact calculations are compared with results from standard observing system experiments (OSEs). The two approaches are shown to provide unique, but complementary, information. The adjoint method also reveals explicit redundancies and dependencies between observing system impacts as observations are added or removed. Understanding these dependencies poses a major challenge for optimizing the use of the current observational network and defining requirements for future observing systems. Preliminary results showing the impact of observations in a newly developed 4DVAR version of GEOS are also presented.

The field phase of the Tropical Cloud Systems and Processes (TCSP) experiment took place between in July 2005, based out of San Jose, Costa Rica. Although the focus area for TCSP was planned to be the Eastern Pacific, the unusually early genesis of tropical disturbances in the Eastern Caribbean prompted missions dedicated to the formation and evolution of Hurricane Dennis. The high- altitude (20-km) NASA ER-2 flew 12 missions, including three dates dedicated to Hurricane Dennis. The flights on July 5-6 captured Dennis as it was transitioning from a tropical storm to a hurricane, and July 9 as it was entering a period of rapid intensification.

The ER-2 deployed four key microwave sensors, the Advanced Microwave Precipitation Radiometer (AMPR), the High Altitude MMIC Sounding Radiometer (HAMSR) and the ER-2 Doppler Radar (EDOP) and Cloud Radar System (CRS). The AMPR is the aircraft “simulator” of the TRMM sensor; from 20-km altitude its 85 GHz imagery has an on- Earth resolution of 700-m (2.8-km at 10 GHz) at nadir, nearly 20 times finer than typical TRMM or SSMI satellite imagery. EDOP is a 10 GHz Doppler radar capable of resolving the fine scale vertical cloud structure. Since microwave observations respond to the presence of water vapor, liquid and ice hydrometeors, they are useful for evaluating the capabilities and deficiencies of mesoscale prediction models in representing the spatial evolution of the underlying hurricane structure.

In this presentation, the microphysical outputs from a simulation of Hurricane Dennis using the Coupled Ocean Atmosphere Mesoscale Prediction System (COAMPS^®) were used to forward model observed TRMM and SSMI satellite data, and the AMPR, HAMSR and EDOP data. This allows model-vs-observation diagnostics to be performed in observational space (similar to AMSU radiance-level monitoring that is routinely done for clear sky conditions), and provides an assessment of the capabilities of COAMPS or other cloud resolving models to replicate cloud and rain affected satellite radiances. Statistical intercomparisons show model overpredictions of the reflectivities at upper levels, and an overprediction of the coldest 85 GHz brightness temperatures reflecting excessive graupel. Although the results are limited to a single case, the methodology is potentially applicable to routine model runs for monitoring modifications such as microphysical parameterization schemes.

Much of the recent progress in skill of operational numerical weather prediction systems can be attributed to new satellite sensors and to advances in methodologies for assimilating satellite data in general. However, much remains to be done both in terms of maximizing the impact of some of the advanced instruments launched in recent years and in terms of preparing to assimilate data not yet used in operations as well as data from sensors yet to be launched. The Joint Center for Satellite Data Assimilation plays a key role in both these areas since it is tasked with preparing new satellite data and related research for use in operational prediction system. This talk will provide an overview of the context of the Joint Center and some of the ways in which it strives to achieve its goals. Challenges and opportunities facing the Joint Center over the next few years will also be discussed. This includes the potential role of the Center in helping to define the Global Observing system of the future.

The Advanced Microwave Sounding Unit (AMSU) can be used to retrieve improved typhoon parameters because cloud interference effects are minor compared to the infrared and AMSU has higher resolution than the Microwave Sounding Unit (MSU). The three dimensional rotational winds component can be obtained by solving the nonlinear balance equations using the retrieved temperatures from AMSU under the following assumptions: hydrostatic balance, the height of 50hPa over the top of typhoon is same as environment, and gradient balance. The divergent wind component can be evaluated from the omega equation. The diabatic term in the omega equation was estimated from the rainfall rate obtained from AMSU observations. The frictionally-induced convergence in the boundary layer was represented by a parameterization. In this presentation we formulate a procedure to analyze the structure of temperature and winds in a typhoon through AMSU observation. Analysis cases are presented. Some typical features of a typhoon can be captured by AMSU data. And one simulation case using MM5 with retrieved winds also has been accomplished. The simulation results show the potential of AMSU data in numerical weather forecasts.

To be supplied.

To Be Supplied

The Eumetsat Satellite Application Facility for Numerical Weather Prediction (the NWPSAF) forms part of the Eumetsat Distributed Ground Segment. The mission of the NWPSAF is to improve and support the interface between satellite data and products and European activities in global and regional NWP. The NWPSAF partnership involves the Met Office (coordinators), ECMWF, KNMI and Meteo France. An important focus of the NWPSAF is the development of software modules for use in NWP Data Assimilation (DA) systems. Deliverables to date, since the development phase of the project started in 1998, have included AAPP, RTTOV, a range of 1DVar schemes, the Quickscat Data Processor and the SSMIS preprocessor. The NWPSAF also has an active visiting scientist programme.

The QuikSCAT satellite was launched on June 19, 1999 into a sun-synchronous, circular, 803 km orbit with a local equator crossing time at the ascending node of 6:30am. QuikSCAT carries a conically-scanning, dual pencil beam Ku-band scatterometer that acquires global backscatter measurements at 47 degrees (H-pol) and 55 degrees (V-pol) incidence angles. These measurements yield high quality 25 km and 12.5 km spatial resolution surface wind vector retrievals over 90% of the world's oceans in a single day.

NOAA's National Environmental Satellite, Data, and Information Service (NESDIS) in cooperation with NASA/JPL has been providing near real-time QuikSCAT ocean surface wind vector products at 25 km and 12.5 km resolutions to the operational community since shortly after launch. Significant improvements in operational weather forecasting and warnings have been realized through utilization of these near real-time products. This real-world experience has also revealed some of the limitations of QuikSCAT, which is a research mission, with respect to the operational forecasting and warning environment.

To address some of these limitations the scatterometer project at JPL implemented several changes in the QuikSCAT processing algorithm, and since May 2006 these improvements have been implemented in a parallel test mode at NOAA / NESDIS / STAR. The NRT QuikSCAT processing improvements were validated by examining 6 months of vector wind data from 2003 processed with both the old and the new algorithms. Validation was conducted by the Ocean Surface Winds Team in STAR, with evaluation from the operational forecaster perspective being conducted by colleagues at the Ocean Prediction Center (OPC) and the Tropical Prediction Center (TPC). Results of these analyses are presented here, and show that the retrievals from the new processing performs better than those from the old processing, especially at the swath edges. Also, the rain impact flag, which results in less data being flagged as potentially contaminated by rain, does not result in a degradation of the overall wind vector retrieval. Project website and data links here.

The accuracy of analyses produced by modern data assimilation systems depends strongly on the precision of forecast error covariances specified as input. Typically, these errors are synoptically dependent, anisotropic, and and inhomogeneous. This talk will begin with a review of techniques used to date to represent flow-dependent errors in variational data assimilation systems. Current NCAR efforts in this direction are based on the WRF model, and are two-fold. Firstly, the application of 4D-Var implicitly introduces flow-dependent covariances via the use of a linearised forecast model (and its adjoint). Secondly, the use of ensemble-based forecast error covariances in 3/4D-Var via additional control variables in a hybrid approach is seen as a way to practically combine the best of both variational and ensemble approaches to data assimilation for operational NWP. Preliminary results from WRF applications for both 4D-Var and the hybrid will be presented.