Hi Iain, Some remarks about the cloud cover. If you look at the last graph at this page, you will see the solar flux for this period of year at the location where the RWIS station is: http://ptaff.ca/soleil/?l1pays=Canada&l1etat=Colombie+Britannique&l1ville=Abbotsford&year=2008&month=03&day=15&l1cityname=Christina+Lake%2C+Colombie+Britannique%2C+Canada&l1ltd=49&l1ltm=14&l1lts=33&l1ltx=N&l1lgd=118&l1lgm=3&l1lgs=00&l1lgx=W&l1tz=-8.0&l1dst=US&l2ltx=N&l2lgx=E&l2tz=0&lang=en_CA&go=Voir+le+graphe%21
Like you see, the contribution of the sun is very important at this period of year. This probably explain the sensitivity to cloud cover value. A road forecast produced for January would probably not be as much affected by a small change in cloud cover. About the positive bias in the model, it could be true but we need more than one specific case to illustrate this bias. To have looked at a lot of road forecast vs. obs, bias can be from both side depending of the location of the RWIS station. As example, we had a lot of trouble at Sault-Ste-Marie because this city is between 2 Great Lakes. Maybe other people in the list have an opinion about this bias? I suggest you produce a lot of road forecast for this specific place and that you verify with the observations (you have everything in vix to create graphics with observation and road forecast). Then you will be able to see if this bias occur all year long for this specific place. Miguel Iain Russell wrote: > Hi everyone... > > Thanks to Miguel who helped me to get Vix running on my Linux box, I can now > post some png image files of the tests that I've been doing for this > particular site in question. > > First image, "original.png", shows pavement temperatures forecast to peak at > ~17C in the afternoon, and we know that the forecast temperature (t) and > dewpoint (td) values are too high by several degrees between the hours 18-03 > GMT (afternoon hours local time of the site). > > Second image, "t_td_corrected.png" shows the effect of fixing the t and td > forecast values to the observed values during the same time period (18-03 > GMT). Note that the peak pavement temperature forecast is reduced to ~14C as > a result of correcting the t and td values. > > Third image, "t_td_cc_corrected.png" shows the additional effect of fixing > the cloud cover (cc) forecast values to the observed values as well as fixing > the t and td values (the adjustment of cc values from forecast to observed > was only of the order of 1 or 2 octas). Note the striking reduction in peak > pavement temperature forecast, it goes down to ~6C as a result of also > correcting the cc values. > > Fourth image, "cc_corrected.png" shows the singular effect of fixing the cc > forecast values to the observed values (no adjustment of t and td forecast > values). Peak pavement temperature is then reduced to ~9C. > > Some conclusions from this case; > > 1. Cloud cover values have a large effect on pavement temperature - in this > case when we corrected them we only increased them by 1 or 2 octas, yet the > effect on the pavement temperature was about 8C (i.e. 17 - 9) > > 2. T and Td have a much smaller effect - in this case we reduced them by 5 > or 6C at some of the afternoon hours, yet the reduction in pavement forecast > temperature was only about 3C (i.e. 17 - 14) > > 3. Even if we supply perfect t, td, cc data to the model during the afternoon > hours for this site, the pavement temperature forecast still peaks at 6C. > This seems too high for a road surface that is snow covered, where we have > cloudy conditions, and snow falling on the road surface throughout the > afternoon.This suggests a positive pavement temperature bias error in the > model...perhaps the albedo of the surface is too low ? > > Regards > Iain > > ________________________________________ > From: Iain Russell > Sent: Thursday, March 20, 2008 3:36 PM > To: 'Miguel Tremblay' > Cc: '[email protected]' > Subject: RE: [Metro-developers] High pavement forecast temperature with > melting snow road condition > > Thanks Miguel, > > Have a nice weekend. > > When I've regenerated the forecast using the corrected temperature and > dewpoint values I'll post to the list. > > I too had thought about modifying the solar flux directly - this may be a > plausible solution, like I said I'll give it some thought. > > Regarding the melting snow condition that we found in our case - I see why > the model produced this based on the definition. What's puzzling me though is > how the road temperatures got so high - they went from zero Celsius to about > 17 Celsius in 4 hours, and the forecast was for snow to continue falling. Yes > the air temperature forecast was 3-4 degC above freezing, but I don't think > that can be the source of this heat. I am suspecting that the radiation > fluxes are the problem, perhaps due to too much energy absorbed at the road > surface (e.g. if the albedo is set too low in the model).....for now we are > assuming that we need to pay attention to the cloud cover field to limit the > amount of incoming direct solar radiation. Does the rate of increase of > pavement temperature (17 C in 4 hours on a road which is at freezing point > with snow falling) seem high to you ? > > Regards > Iain > > -----Original Message----- > From: Miguel Tremblay [mailto:[EMAIL PROTECTED] > Sent: March 20, 2008 3:27 PM > To: Iain Russell > Cc: '[email protected]' > Subject: Re: [Metro-developers] High pavement forecast temperature with > melting snow road condition > > Hi Iain, > > Iain Russell wrote: > >> Thanks Miguel. >> >> 1. I won't be able to re-generate the roadcast until over the weekend; sorry >> to be a burden, but do you have time to change the air temperature and >> dewpoint temperatures so that they match the observed and then re-run the >> model ? >> >> >> > I don't have time either, I am going for the weekend. > >> 2. Yes I understand about site exposure, and yes we could adjust the cloud >> cover values to control the amount of solar flux. Would prefer to have some >> other parameter to control this other than cloud cover though..I'll think >> about that... >> >> >> > Starting from next version (3.2.0, release is schedule for this spring), > you will be able to set directly the solar flux in W/m² in the > atmospheric forecast file and use it directly instead of the cloud > cover. The infrared flux would be created using the cloud cover and you > could modify directly the solar flux. Does this sound like a possible > solution? > >> 3. So the definition of the melting snow condition is when snow is falling >> on a road that is already above freezing ? >> >> >> > Yes. > > Have an happy Easter, > > Miguel > > > ------------------------------------------------------------------------ _______________________________________________ METRo-developers mailing list [email protected] https://mail.gna.org/listinfo/metro-developers
