[GRASS-user] Use of coefbh and coefdh in r.sun

Tue Jun 26 17:27:24 EDT 2007

On Tuesday 26 June 2007 01:18, José Antonio Ruiz Arias wrote:
> > -----Mensaje original-----
> > De: Dylan Beaudette [mailto:dylan.beaudette at gmail.com]
> > Enviado el: lunes, 25 de junio de 2007 22:37
> > Para: José Antonio Ruiz Arias
> > CC: grassuser at grass.itc.it; grass-dev at grass.itc.it
> > Asunto: Re: [GRASS-user] Use of coefbh and coefdh in r.sun
> >
> > On Monday 25 June 2007 12:27, José Antonio Ruiz Arias wrote:
> > > Hi Dylan,
> >
> > Hi, thanks for the comments!
> >
> > > I will expose what I modestly think about.
> > >
> > > 1. does (1-coefbh)=coefdh? No.
> > > I think you are confusing the clear-sky indices for the direct and
> > > diffuse components with the direct and diffuse fractions. These
> >
> > latter
> >
> > > are both referred to the total global radiation coming into the
> >
> > ground
> >
> > > so that, considering the reflected component negligible, the addition
> > > of the direct and diffuse fractions are always the unity.
> >
> > ok- this makes sense.
> >
> > > In turn, the clear-sky index
> > > for the direct horizontal component (coefbh) is the fraction of
> >
> > direct
> >
> > > component attenuated by the clouds, that is, the ratio of direct beam
> > > horizontal radiation under overcast-skies to the direct beam
> > > horizontal radiation under clear-skies. The same is applied to the
> >
> > diffuse component.
> >
> > It sounds like in order to use 'real-sky' estimates I would need more
> > than a
> > pyranometer- as I would need to measure diffuse and beam components on
> > both clear AND cloudy days... ?

Thanks for the reply,

> Depending on the type of radiation you need (global, direct or diffuse), if
> you want a 'real-sky' estimation with r.sun you will have to measure that
> radiation for one (or a few sites - better) in your study area. That will
> be your "ground-truth" to tune up the r.sun results and you can - hopefully
> - expect to be representative for the whole study area (raster).

Indeed. This makes sense now. I do not think that I will be able to estimate 
real-sky values for a couple reasons- mainly the short historical record, and 
only a single weather station with pyranometer.

> > > 2. does coefdh correspond to the notion of the 'diffuse fraction'
> > > which is commonly calculated via something called the 'clearness
> > > index' ? I think I have answered above.
> >
> > Ok, it sounds like my proposed method was in error.
> >
> >
> > In this case, Kt (see named attachment) and the derived Kd (see
> > attachment) - are describing the diffuse fraction of total radiance?
> > Values which cannot be used for the coefdh and coefbh inputs to r.sun.
>
> Yes. Maybe the definition help you to see the subtle difference among these
> indices:
>
> Global_rad=Direct_rad + Diffuse_rad  (If reflected rad is negligible)
> Kt=Global_rad/Extraterrestrial_rad
> Kb=Direct_rad/Global_rad and Kd=Diffuse_rad/Global_rad => Kb+Kd=1

Ok, this makes sense and I understand this now. 

> regardless the sky conditions. However, coefdh and coefbh are defined as:
>
> coefbh=horizontal direct radiation under any-sky (to be
> measured)/horizontal direct radiation under clear-skies (r.sun)
> coefdh=horizontal diffuse radiation under any-sky (to be
> measured)/horizontal diffuse radiation under clear-skies (r.sun)
>
> which distinguishes among sky conditions. Thus, you multiply your clear-sky
> estimation raster map (r.sun) for the corresponding coefficient coefbh or
> coefdh calculated on your ground station (measured/r.sun).

Got it.

> > Would it be possible to disaggregate my pyranometer data in this way-
> > giving my the real irradiance on the ground, compute the clear sky
> > radiance at the same location with r.sun, and then compute the ratio of
> > the two in order to get coefdh / coefbh ?
>
> Yes, you could use one of the regression equations to get Kd and Kb from
> the global radiation provided by your pyranometer, then compute the
> clear-sky raster maps with r.sun for your study area and compute the ratio
> of the two. However, take always in account that these equations provide
> you a statistical approach to your problem having a inherent statistical
> error that depends on the regression chosen. In the bibliography you could
> look for one in your site area.

Indeed, this was my initial reason for not doing so. It appears as if I am 
back where I started estimating clear sky conditions from an estimated Linke 
value.

> > > My personal opinion is that the best approach consists on using a
> >
> > more
> >
> > > or less approximated turbidity coefficient of Linke (a climatological
> > > value could be the easiest choice) to get the clear-sky estimation
> > > (r.sun) and then use the ground global radiation measurement
> > > (supposing you have it,
> > > obviously) to correct the clear-sky estimation provided by r.sun.
> >
> > This
> >
> > > approach can be also applied to both, the diffuse and direct
> > > components of the radiation. The problem with the components is the
> > > seldom availability of measurements.
> >
> > This was my initial goal, and seemed to work fairly well. I was able to
> > compute T_L values accurately during the summer based on my pyranometer
> > data, but there were not enough clear days in the 11yr history to
> > compute realistic T_L values in winter months. Using a 15% diffuse
> > fraction estimate for the entire year seemed to yield data which
> > followed the 11yr mean values quite well.... However, I was hoping to
> > avoid a static diffuse fraction- and make better use of the pyranometer
> > data for the computation of 'real-sky'
> > conditions. This may not be possible...?
>
> I think, rather than use a static 15% of diffuse fraction you should use a
> regression equation that will give you slightly better estimations (remind
> that if you have the global radiation, you have the clearness index, and
> then, the diffuse fraction and the direct fraction through a regression
> equation). The problem of estimating the sky turbidity is that you need
> spectral measurements of the radiation and usually we do not have it.
> However, there exist broadband methods to calculate it from direct beam
> measurements. But this approach unavoidably needs of direct beam
> measurements and, in the case of the TL, also needs the humidity (usually
> precipitable water).

I had originally tried this- but my 11 years of data did not have enough clear 
sky winter days, and therefore the computed T_L values were between 7 - 11 ; 
perhaps a little unrealistic. 

Thanks again,

dylan

> > > I hope to have cleared I little the question José A.
> >
> > Yes, thanks for the tips
> >
> > Cheers,
> >
> > Dylan
> >
> > > > -----Mensaje original-----
> > > > De: grassuser-bounces at grass.itc.it [mailto:grassuser-
> > > > bounces at grass.itc.it] En nombre de Dylan Beaudette Enviado el:
> > > > lunes, 25 de junio de 2007 19:25
> > > > Para: grassuser at grass.itc.it
> > > > CC: grass-dev at grass.itc.it
> > > > Asunto: [GRASS-user] Use of coefbh and coefdh in r.sun
> > > >
> > > > Hi,
> > > >
> > > > After consulting with an atmospheric scientist, I have decided that
> > > > it might be simpler to estimate 'real-sky' radiance as opposed to
> > > > trying to estimate the Linke turbidity for the estimation of
> > > > 'clear-sky' conditions.
> > > >
> > > > It looks like r.sun can use input maps: coefbh and coefdh to
> >
> > compute
> >
> > > > real-sky radiance values. I would like to be sure that I am
> > > > interpreting these inputs
> > > > correctly:
> > > >
> > > > 1. does (1- coefbh) = coefdh ?
> > > > 2. does coefdh correspond to the notion of the 'diffuse fraction'
> > > > which is commonly calculated via something called the 'clearness
> > > > index' ?
> > > >
> > > >
> > > > If this is the case, then it is not all that difficult to compute
> > > > coefdh from pyranometer data and the output from r.sun (mode 1)
> > > >
> > > > I will update the man page with my approach, if these assumptions
> > > > are correct.
> > > >
> > > > cheers,
> > > >
> > > > PS:
> > > > a relevent references is:
> > > >
> > > > Jacovides, C.; Tymvios, F.; Assimakopoulos, V. & Kaltsounides, N.
> > > > Comparative
> > > > study of various correlations in estimating hourly diffuse fraction
> > > > of global solar radiation. Renewable Energy, 2006, 31, 2492 - 2504
> > > >
> > > >
> > > > --
> > > > Dylan Beaudette
> > > > Soils and Biogeochemistry Graduate Group University of California
> >
> > at
> >
> > > > Davis
> > > > 530.754.7341
> > > >
> > > > _______________________________________________
> > > > grassuser mailing list
> > > > grassuser at grass.itc.it
> > > > http://grass.itc.it/mailman/listinfo/grassuser
> > > >
> > > > __________ Informacisn de NOD32 2352 (20070625) __________
> > > >
> > > > Este mensaje ha sido analizado con  NOD32 antivirus system
> > > > http://www.nod32.com
> >
> > --
> > Dylan Beaudette
> > Soils and Biogeochemistry Graduate Group University of California at
> > Davis
> > 530.754.7341
> >
> >
> > __________ Información de NOD32 2354 (20070626) __________
> >
> > Este mensaje ha sido analizado con  NOD32 antivirus system
> > http://www.nod32.com

-- 
Dylan Beaudette
Soils and Biogeochemistry Graduate Group
University of California at Davis
530.754.7341