Sea surface temperatures are another parameter that is incorporated into the models. The sea surface temperature is just that -- a temperature taken at the surface of a body of water. Like most aspects of the atmosphere, sea surface temperatures are linked to the features described above. Holton (1992) stated that "the global distribution of evaporation clearly depends on the sea surface temperature." Therefore, the distribution of SSTs can be seen to have an indirect effect on the distribution of moisture through the atmosphere.
The concept of surface roughness acts as a type of "friction term." Models using this factor must define a roughness length, which is a measure of the surface roughness over which a fluid must flow. The main effect of surface roughness implemented into models will be evident in the winds. Wind speeds will tend to slow down with rougher terrain. Some models determine roughness lengths over oceans from the surface wind stress, while over land, roughness lengths are prescribed from data that includes twelve vegetation types. It should be noted that surface roughness is not a function of orography (Model status, 2000).
Perhaps one of the most important parameters used by forecast models is that of topography. Temperature, precipitation, and wind fields are just a few of the forecast variables that can be affected by topographic features. For example, if there is an incorrect representation of topography in the model, then orographic lifting may be inaccurately forecasted by the models. In addition, leeside cyclogenesis also has a potential to be incorrectly forecasted. In both cases, erroneous precipitation amounts may result.
From the examples given above, one can begin to see the influence just a few terrestrial quantities have on the computer model forecast. Clearly, an accurate representation of items such as these needs to be placed into the models in order to produce the most reliable forecast possible. The following sections will take a closer look at some of these features by looking at the individual models.
Return to the Top
a. Topography
In general, two different ways of resolving terrain were used by the models examined. Both NGM and MRF/AVN models incorporate orography using atlas corrected United States Navy 10 minute resolution terrain data. After employing a general filtering process, the terrain is given a mean value of the area surrounding each grid point (Hoke et. al, 1989) as determined by the U.S. Navy data . However, the Sierra and Cascade Mountain ranges are not included as part of the orography in both models and both smooth the eastern extent of the Rocky Mountains into the Great Plains. Figure 1 and Figure 2 show examples of how the ETA model is able to resolve terrain in selected regions of the United States. As you can see by Figure 3 the AVN terrain representation is of much lower quality than that of ETA. Also, take note of how the oceans are represented in the AVN model. Although this is not a source of large error, it occurs in all spectral models.
According to Rogers et. al (1995), the Eta orography is organized in a series of discrete steps, computed using a modification of the silhouette orography method. Using this method, the silhouette elevation is replaced by the average elevation when concave land features are found. Both the "Early Eta" and the "Meso-Eta" models incorporate 10 minute topographic data in each of their sixteen sub boxes (which are devised from the original 32 or 29 kilometer grids) from the United States Geological Survey (Staudenmaier, 1997) (Figure 4).
The RUC-2 model uses a
"slope envelope" topography. This is a process where the terrain
variance is computed using a plane fit to the topography resulting in more
accurate terrain values (40 km RUC Info, 1998) (Figure
5).
b. Surface Characteristics
Junker et. al (1989) stated that surface and ground features, consisting of snow and ice cover, soil moisture, and subsoil temperature, strongly influence the behavior of the NGM model in the lower troposphere. Shortwave, radiative heating of the land surface by the sun is computed as a function of cloud clover, surface albedo, and solar zenith angle (Hoke et. al, 1989). The land surface temperature is forecasted using the surface energy budget, which encompasses many radiative processes. This surface energy budget is not necessary over the water covered portions of the planet as sea surface temperatures (SST) are assumed to be constant throughout the duration of the forecast period. These SST's are updated daily, primarily through the use of buoys, satellite information, and ship reports. Snow and ice cover, whose fields are updated weekly, are other factors influencing the model. They increase the surface albedo and reduce the heat exchange between the surface and the atmosphere.
In addition to soil moisture, which is determined by climatology, soil temperature and snow cover are included in the soil parameterization scheme for the Eta model. The surface temperature and moisture are updated every four adjustment time steps. However, over large bodies of water, these quantities are held constant (Black, 1994). Unlike the NGM model, where snow depth remains constant throughout the valid forecast time, physics included in the Meso-Eta will either increase or decrease the amount of snow cover over a given region. The Meso-Eta also includes several different soil types to aid in the determination of soil temperatures (Staudenmaier, 1997).
Soil moisture updates in the MRF and AVN models are computed from the model forecast. This is because there are no routine measurements of soil moisture (Wu et al, 1997). By using the model to forecast soil moisture, anomalies in forecasted precipitation may cause errors in soil moisture amounts. To counteract this, soil moisture is nudged to climatological values with a relaxation period of sixty days.
Soil moisture information spanning a period of months to years is incorporated into the RUC model. A soil model containing five soil levels resulted in a much improved specification of the surface. Daily lake surface temperature and snow-ice cover information are used. Additionally, daily SST's are used in the 40 km RUC.
Return to the Top
Several advantages of the Eta model can be noted over the NGM. Because of higher resolution in the step topography, the Eta model is more accurate in predicting precipitation and forecasting the winds in the lowest layers of the atmosphere (Staudenmaier, 1997). For similar reasons, this model does a superior job in forecasting precipitation on the lee-side of the Cascades and Sierra Mountains. However, the model is not without its disadvantages. Because the "Early Eta" is conservative with returning moisture to the Gulf Coast region, the resulting atmosphere in this region is too stable (Model Biases, 1997).
Like the NGM, terrain smoothing problems in the MRF and AVN result in the over prediction of precipitation in the Great Plains east of the Rockies and over the southern Appalachians. This trend was also noted east of the Cascades and across the Sierra mountain regions (Junker et. al, 1992). Precipitation amounts along the United States West Coast are underestimated due to details of the terrain being neglected by the model. Because of the incorporation of climatology, soil moisture values are skewed toward normal values to alleviate poorly forecasted precipitation by the model (Wu et. al, 1997).
Another problem seen with all operational models is the inability to forecast downslope wind events. This comes up because of a lack of sufficient resolution in model grids. Even the current ETA, which now runs on a 12km grid, cannot adequately represent the topography of the Rocky Mountains to forecast these windstorms.
Because of a lack of accurate water
temperatures, the 60 km RUC will tend to be too dry in return flow situations
from a body of water. The 40 km RUC's improved physics affords consistency
in data void regions. Both versions of the model use the topography
field to determine surface and dew point temperatures close to where surface
stations are actually located. In the 40 km model, differences can
occur between actual precipitation and soil moisture data from that predicted
by the model. This can result from inaccurate input into the soil
model or shortcomings in the model itself.
Albedo, surface moisture,
sea surface temperatures, surface roughness, and topography can all affect
the model predictions. This is important because terrain can cause
and/or modify weather phenomena. Some examples of affected weather
processes are: torrential rain and flash floods, cold-air damming,
and clear-air turbulence (USWRP, 1999). Because
of these effects, orography is essential for an accurate model.
For more information and some neat pictures, visit
http://www.meted.ucar.edu/nwp/pcu1/ic4/frameset.html
.
Black, T.L., 1994: The new NMC mesoscale eta
model: Description and forecast examples.
Wea. Forecasting, 9, 265-278.
Brooks, C.E.P, J.E. McDonald, and M.G. Wurtele,
1959: Glossary of Meteorology, R.E.
Huschke, Ed., Amer. Meteor. Soc.,
638 pp.
Fetter, C. W., 1994: Applied Hydrogeology (Third Edition), Prentice Hall, 183-186.
Forecast Systems Laboratory, 1998: Information
on 40 km RUC/MAPS, FSL,
http://maps.fsl.noaa.gov/ruc2.tpb.html.
Hoke, J. E., N. A. Phillips, G. J. DiMego, J.
J. Tuccillo, and J. G. Sela, 1989: Regional analysis
and forecast system of the National
Meteorological Center. Wea. Forecasting, 4, 323-334.
Holton, J. R., 1992: An Introduction to Dynamic Meteorology (Third Edition), Academic Press, 361-363.
Junker, N. W., J. E. Hoke, and R. H. Grumm,
1989: Performance of NMC's regional models.
Wea. Forecasting, 4, 368-390.
------,J. E. Hoke, B. E. Sullivan, K. F. Brill,
and F. J. Hughes, 1992: Seasonal and
geographic variations in quantitative
precipitation prediction by NMC's nested-grid
model and medium-range forecast model.
Wea. Forecasting, 7, 410-429.
National Centers for Environmental Prediction,
1995: Model status as of January 24, 2000, NCEP,
http://www.sgi.wwb.noaa.gov:8080/tpb97/T170/html/T170.html.
Storm Prediction Center, 1999: RUC2 Model Composite Map Page, http://www.spc.noaa.gov/compmap/index.shtml.
Texas A&M University, 1999: NGM Analyses and Forecasts, http://www.met.tamu.edu/weather/mp/ngmtable.html.
Texas A&M University, 1999: Site of the MRF and ECMWF, http://www.met.tamu.edu/newmodels/newmodels.html.
Texas A&M University, 1999: Aviation Model Analyses and Forecasts, http://www.met.tamu.edu/weather/mp/avntable.html.
Texas A&M University, 1999: ETA Model Analyses and Forecasts, http://www.met.tamu.edu/weather/mp/etatable.html.
Perkey, P. J., 1986: Formulation of mesoscale
numerical models. Mesoscale Meteorology
and Forecasting, P.S. Ray,
Ed., Amer. Meteor. Soc., 573-596.
Rogers, E., D. G. Deaven, and G. J. Dimego, 1995: The regional analysis system for the operational "Early" eta model: Original 80 km configuration and recent changes. Wea. Forecasting, 10, 810-825.
------, M. Baldwin, T. Black, K. Brill, F. Chen, G. DiMego, J. Gerrity,
G. Manikin, F. Mesinger,
K. Mitchell, D. Parrish, and Q. Zhao,
1998: Changes to NCEP's operational "Early" eta
analysis/forecast system,
NCEP, http://www.nws.noaa.gov/om/447body.htm.
Staudenmaier, M., 1997: The soil model in the eta. Western Region Technical Attachment 97-07, NWS, http://nimbo.wrh.noaa.gov/wrhq/97TAs/TA9707/ta97-07.html.
United States National Weather Service, 2000: Sea Surface Temperature Analysis, http://www.nws.noaa.gov/er/box/satellite/seatemps.htm.
University Center For Atmospheric Research, 2001: COMET NWP Module , http://meted.ucar.edu/nwp/pcu1/ic2/frameset.htm
United States Weather Research Program, 1999: Orographic influences, uswrp.mmm.ucar.edu/uswrp/PDT/one/d.html
Wu, W., M. Iredell, S. Saha, and P. Caplan, 1997:
Changes to the 1997 NCEP operational MRF model
analysis/forecast system,
NCEP, http://sgi62.wwb.noaa.gov:8080/tpb97/tpb975_wp.html.
Updates:
February 1, 1999 by Amanda Fox, Cody LIndsey, Stephanie Naumann.
February 25, 2002 by Heather Moore, Matt Heard, Jason Sippel.
Back to the Top
Return to the Model Forecasting page
