The insolation data did provide an opportunity to evaluate the skill of the individual
weather models in predicting incoming short wave radiation. The data proved to be quite
valuable as it identified several issues and problems related to modeled insolation data.
Short wave radiation data were available from the MM5, WRF, Eta, and GFS, but there
were significant differences in the way they were calculated and provided. For example,
the GFS short wave radiation field is a 3-hour time-averaged value, while the Eta
provides an instantaneous value. Listed below are descriptions of the issues that arose
with the short wave radiation fields used in the MDSS.
Eta Model: The short wave radiation data from Eta model was an instantaneous value,
but it had a time lag problem. On 17 March 2004, an NCEP bulletin identified that the
Eta model insolation data were offset by one hour. For example, the data values given for
06:00 hrs were really valid at 05:00 hrs. In addition, NCEP corrected a problem with long
wave radiation where low cloud and fog were not accounted for properly. These
problems were observed during the MDSS field program. NCEP corrected these
problems on or about 17 March 20041. Although this was too late to help the MDSS this
year, these changes should result in better forecast skill in the future for those using the
Eta model. Hourly output of the Eta insolation data are not available reducing the utility
of these data.
GFS Model: The short wave radiation data from GFS model was a time averaged value
over a 3 hr period. Because of the time averaging characteristic, it was not practical to use
it as an input to a road temperature model.
WRF Model: The radiation data from WRF model provided instantaneous values, but
early in the field program, the WRF provided the net radiation, which was not separated
into long and short wave fields. Later on, an underestimation of the insolation was
discovered on clear days and it was determined that WRF was retaining too many
hydrometeor species, which attenuated short wave radiation. On 17 February, the WRF
model was corrected and the insolation data more closely matched the Ames pyranometer
data on clear and cloudy days.
MM5 Model: The MM5 provided an instantaneous radiation values and matched the field
data more closely than any of the other models, at least until the WRF was corrected.
After that time, the two models values were very close on clear days.
The RWFS was configured to use the MM5 and WRF insolation data for the part of the
forecast period where those data were available (0-15 hrs) and the Eta insolation data for
the 15-48 hour forecast period. The real-time insolation data from Ames were not used in
the RWFS; therefore, the weight blending values for insolation had to be fixed. In the
future, it may be desirable to utilize real time insolation data to forward correct model
data, but additional research is required to assess this approach.
1
Note that the NCEP fixes were not due to MDSS project feedback, but were being addressed
independently.
101
Previous Page
