What is ADAS?ADAS is the ARPS Data Assimilation System developed at the University of Oklahoma. Modifications have been made at the University of Utah to take into consideration the effects of complex terrain. ADAS is a 3-dimensional analysis system; however, many of the applications of ADAS are done in terms of 2-dimensional surface analyses.
What versions of the Utah ADAS system are available in AWIPS?A 10 km surface analysis over the western U.S. is run and transmitted over the NWS Western Region LAN every 15 minutes. It is available roughly 45 min past the valid time on AWIPS workstations. Check the log file of the most recently completed analysis to see the operational status of ADAS. Eventually the ADAS surface analysis over the Western Region domain may be run at Western Region Headquarters in order to improve reliability and avoid comms hassles.
A 1 km resolution surface analysis over northern Utah is run at 15 minute intervals and transmitted to the NWS for display in AWIPS. Output from the 1km analysis is also available within about 45 minutes past the valid time. Check the log file
for the 1km adas operational status
What versions of the Utah ADAS system are available via the Internet?
1km 3-dimensional analyses over Northern Utah are run once per hour, with graphics
typically available by 90 minutes past the valid time
(most of the extra time is required to generate the graphics).
The 10 km Western U.S. surface analysis and the 1km Northern Utah surface analysis graphics are also available on the web by about 45 minutes past the valid time.
What are the differences between the surface analyses and the 3-dimensional analysis?The surface analyses rely upon the RUC2 analysis, used as a background field, combined with MesoWest surface observations.
The 3-dimensional analyses rely upon the RUC2 background fields and incorporate as much local data as possible: MesoWest surface observations, Dugway profiler and SLC rawinsonde, ACARS, KMTX reflectivity and velocity, and satellite visible and infrared imagery. In order to make ADAS as timely as possible, the RUC2 background field is normally 1 hour old, i.e., the 12Z ADAS analysis uses the 11Z RUC2 analysis.
The ADAS surface analyses should not be viewed as competitors to 3-dimensional data assimilation systems that are used to initialize weather prediction models. Rather, the surface analyses are simply an extension of the MesoWest surface observations available in AWIPS in NWS Western Region. The RUC2 background and ADAS terrain fields help to build spatial and temporal consistency that are adjusted by the MesoWest observations. Viewed in this manner, the ADAS surface analyses help to visually integrate MesoWest observations and make areas experiencing significant weather easier to identify. Another perspective is to view the ADAS analyses as a quality control step to the MesoWest observations. The temporal and spatial continuity of the ADAS analyses helps to identify those MesoWest observations that are reflecting hazardous conditions underway as opposed to observations that may reflect poor siting or sensor errors.
Isn't a 10 km analysis too coarse for the complex terrain over the West?We're conducting research to identify the tradeoffs in analysis quality as a function of horizontal resolution. Our preliminary results, based on qualitative evaluation of many analyses over northern Utah, indicate that the 10 km analyses provide useful information, even though many isolated mountain ranges and valleys are not captured by the ADAS 10 km terrain compared to 2 km terrain.
It is natural to expect that numerical weather forecasts should improve when forecast models begin to resolve local terrain features in greater detail than at the present time. The results from many operational and research modeling groups suggest that high resolution models are most useful when the large-scale flow is predictable and add little value when the large-scale flow is poorly forecast. In other words, even though more realistic local weather features are simulated when the models are run at high spatial resolution (e.g., mountain/valley circulations), objective measures of forecast skill indicate greater sensitivity to errors in the positioning and timing of large-scale circulation features.
While it may be equally intuitive to expect that analyses of the current surface weather will be better when the analysis systems resolve the local terrain in greater detail, that is not necessarily the case. Relatively simple analyses, such as the successive correction technique used by ADAS, depend strongly upon the resolution of the background field (in our case the RUC2) and the average distance between surface observations. The average station seperation varies strongly across the West with isolated pockets of very dense observing networks (e.g., the average distance is on the order of 10 km in the Tooele Valley of northwest Utah) separated by vast regions where the average distance is on the order of 50-100 km. For example, over northeastern Arizona where there are no surface observations, the ADAS analysis simply interpolates the RUC2 wind and temperature fields to the higher resolution ADAS terrain background. No new information is gained, except for the fact that the terrain of a higher resolution analysis will help to define ridge tops to be typically colder and windier than nearby valleys; we expect most forecasters know that already.
To provide more documentation on this subject, we are actively evaluating the relative performance of local analyses at 1 km and 10 km resolution. You may want to contribute to this evaluation by comparing the 10 km version available in AWIPS to the 1 km version available via the Internet. (Both analyses are available in AWIPS at the SLC NWSFO). Examples of analyses over northern Utah at both 10 and 1 km resolution are shown for selected situations below.
| 10 km | 1km | |
| Current | 10km | 1km |
| Summer | 10km | 1km |
| Strongly Forced | 10km | 1km |
Key Points Regarding Horizontal Resolution
Aren't successive correction analysis techniques obsolete?
ADAS uses the Bratseth technique of successive corrections, which has been shown to converge to the optimal interpolation solution. Three-dimensional and 4-dimensional variational assimilation approaches are now considered to be more relevant techniques to improve model forecast skill. However, those techniques rely upon dynamical constraints (e.g., weak geostrophic or thermal wind constraints) that are not necessarily appropriate for local circulations in regions of complex terrain. We are working on a stability-dependent constraint on the 3-dimensional wind field that will help to make the winds more consistent with local terrain features. However, no such constraint is available at the present time for the surface analyses.
A major advantage of the Bratseth technique is that it is fast and can run on a single processor workstation. The computationally expensive variational techniques are impractical for running analyses in near-real time.
Are the ADAS analyses available reliably in AWIPS?
We have implemented procedures to insure that the ADAS analyses are available as reliably as possible. However, occasional outages may be expected, most likely late at night and on weekends. We are particularly susceptible to communication problems with the OSO server at NCEP; if the RUC2 is more than 2 hours old, we use the previous ADAS analysis for the background field of the current run. If communication problems persist for more than a few hours, this technique may result in a degradation of ADAS analysis quality.
What surface fields are available on AWIPS?
Who is involved with the use of ADAS in AWIPS? 
