World Meteorological Organization Symposium on Nowcasting - WMO
World Meteorological Organization Symposium on Nowcasting - WMO
World Meteorological Organization Symposium on Nowcasting - WMO
Create successful ePaper yourself
Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.
58<br />
can be solved for positi<strong>on</strong> and amplitude decisi<strong>on</strong> variables using stochastic methods, thus<br />
corresp<strong>on</strong>ding with ensemble data assimilati<strong>on</strong>. We then show that if an Euler-Lagrange<br />
approximati<strong>on</strong> is made, can solve the objective nearly as well in two steps. This approach is<br />
entirely c<strong>on</strong>sistent with c<strong>on</strong>temporary data assimilati<strong>on</strong> practice. In the two-step approach,<br />
the first step is field alignment, where the current model state is aligned with observati<strong>on</strong>s by<br />
adjusting a c<strong>on</strong>tinuous field of local displacements, subject to certain c<strong>on</strong>straints. The sec<strong>on</strong>d<br />
step is amplitude adjustment, where c<strong>on</strong>temporary assimilati<strong>on</strong> approaches are used. We will<br />
then dem<strong>on</strong>strate several choices of c<strong>on</strong>straints <strong>on</strong> the displacement field, first starting with<br />
fluid-like viscous c<strong>on</strong>straints and then proceeding to a multiscale wavelet representati<strong>on</strong> that<br />
allows better balance in the factorizati<strong>on</strong> of error into positi<strong>on</strong> and amplitude subspaces. Our<br />
new data assimilati<strong>on</strong> by field alignment approach does not rely <strong>on</strong> the detecti<strong>on</strong> of storm<br />
features, and can be used with sparse stati<strong>on</strong> observati<strong>on</strong>s, just as easily as with detected<br />
features. The two-step approach can be used with any assimilati<strong>on</strong> method in practice;<br />
3DVAR, 4DVAR, and EnKF (and variants). The new method has been implemented with<br />
multivariate fields, and extensi<strong>on</strong>s to 3D is straightforward. Ancillary benefits to velocimetry<br />
for rainfall modeling and wind-from-satellites have also been realized. A new data<br />
assimilati<strong>on</strong> system, FAVAR, has been developed to co-exist with using the WRF-VAR<br />
system. Dem<strong>on</strong>strati<strong>on</strong>s <strong>on</strong> storms using OSSEs and identical twins will be shown as a runup<br />
to a live Hurricane DA & Forecasting System at the Atmospheric Sciences Group at MIT.<br />
SESSION 4: Applicati<strong>on</strong>s or Op Systems<br />
4.1<br />
INCA - A new operati<strong>on</strong>al nowcasting system for mountainous areas<br />
6.12<br />
Data mining for thunderstorm nowcast system development<br />
John K. Williams, David A. Ahijevych, Matthias Steiner and Susan Dettling<br />
Nati<strong>on</strong>al Center for Atmospheric Research, Research Applicati<strong>on</strong>s Laboratory<br />
This paper describes a data mining statistical analysis approach to developing real-time<br />
thunderstorm nowcasts for aviati<strong>on</strong> users. While fuzzy logic expert systems are often used to<br />
combine observati<strong>on</strong> and NWP model data to form short-range predicti<strong>on</strong>s of thunderstorm<br />
intensity evoluti<strong>on</strong>, evaluating and incorporating new data sources is often a time-intensive<br />
manual process, and it is difficult to know whether available informati<strong>on</strong> is being used<br />
efficiently. A technique called random forests (RFs) provides a means of objectively<br />
identifying the potential c<strong>on</strong>tributi<strong>on</strong> of candidate predictor variables, al<strong>on</strong>g with a method for<br />
creating a nowcast logic using a minimal skillful set of predictors. The RF methodology was<br />
used to evaluate radar, satellite, lightning, and RUC model data and derived features<br />
collected during the summer of 2007 and 2008 over the eastern U.S., including data fields<br />
produced by MIT Lincoln Laboratory’s Corridor Integrated Weather System (CIWS). The<br />
resulting data fusi<strong>on</strong> system produces real-time probabilistic and deterministic nowcasts of<br />
thunderstorm intensity (VIP level). Statistical evaluati<strong>on</strong>s of the RF-based nowcast’s<br />
performance are shown and several case studies are analyzed, dem<strong>on</strong>strating the value of<br />
this approach. This research has been funded by the U.S. Federal Aviati<strong>on</strong> Administrati<strong>on</strong> to<br />
support the development of the C<strong>on</strong>solidated Storm Predicti<strong>on</strong> for Aviati<strong>on</strong> (CoSPA), which is<br />
intended to provide the thunderstorm nowcast capability for the Next Generati<strong>on</strong> Air<br />
Transportati<strong>on</strong> System (NextGen).<br />
SESSION 7: Verificati<strong>on</strong> and Impacts<br />
Thomas Haiden, Alexander Kann<br />
Central Institute for Meteorology and Geodynamics, Vienna, Austria 7.1<br />
Verificati<strong>on</strong> methods for spatial forecasts<br />
The high-resoluti<strong>on</strong> analysis and nowcasting system INCA (Integrated <strong>Nowcasting</strong> through<br />
Comprehensive Analysis) provides 3-D hourly fields of temperature, humidity, and wind, and<br />
2-D fields of cloudiness, precipitati<strong>on</strong> rate, and precipitati<strong>on</strong> type at an update frequency of 15<br />
min. The system operates <strong>on</strong> a horiz<strong>on</strong>tal resoluti<strong>on</strong> of 1 km and a vertical resoluti<strong>on</strong> of 100-<br />
200 m. It combines surface stati<strong>on</strong> data, remote sensing data (radar, satellite), forecast fields<br />
of numerical weather predicti<strong>on</strong> models, and high-resoluti<strong>on</strong> topographic data. In the alpine<br />
area, the system provides meteorological input for operati<strong>on</strong>al high-resoluti<strong>on</strong> flood<br />
forecasting and is used for winter road maintenance. INCA employs a new radar/raingauge<br />
combinati<strong>on</strong> algorithm and includes elevati<strong>on</strong> effects <strong>on</strong> precipitati<strong>on</strong> using an intensitydependent<br />
parameterizati<strong>on</strong>. In temperature analysis and nowcasting the pooling of cold air is<br />
parameterized as a functi<strong>on</strong> of terrain parameters. Verificati<strong>on</strong> results showing the skill of the<br />
nowcast compared to a high-resoluti<strong>on</strong> NWP model (AROME) are presented and the<br />
potential for applicati<strong>on</strong> of INCA to the 2010 Winter Olympics in the Vancouver/Whistler area<br />
is discussed.<br />
Barbara Brown [1] Eric Gilleland [1] David Ahijevych [1] Barbara Casati [2] Beth Ebert [3]<br />
[1] NCAR, USA [2] Ouranos, Canada [3] Bureau of Meteorology, Australia<br />
In recent years, many new methods have been developed to evaluate forecasts that have<br />
coherent spatial structures. Several different categories of approaches have been developed,<br />
including object- or features-based, scale separati<strong>on</strong>, neighborhood, and field deformati<strong>on</strong><br />
methods. Because the majority of nowcasts (e.g., c<strong>on</strong>vective nowcasts based <strong>on</strong> radar<br />
reflectivity) are characterized by identifiable spatial features and structures, the spatial<br />
verificati<strong>on</strong> methods are appropriate for the evaluati<strong>on</strong> of these products and can provide<br />
meaningful informati<strong>on</strong> <strong>on</strong> their quality. Over the last several years, the developers of many of<br />
the spatial verificati<strong>on</strong> methods have been involved in an intercomparis<strong>on</strong> project that was<br />
designed to compare the capabilities of the methods and provide informati<strong>on</strong> about how the<br />
aspects of performance are measured by each approach. The intercomparis<strong>on</strong> was based<br />
<strong>on</strong> the evaluati<strong>on</strong> of the same set of high-resoluti<strong>on</strong> spatial forecasts by each method. In<br />
additi<strong>on</strong>, the methods were applied to a set of artificial geometric cases with simple, known<br />
forecast errors. This presentati<strong>on</strong> will summarize the results of the intercomparis<strong>on</strong> and<br />
83