I assume I'll have to use one of the above formulas with r.mapcalc to
create the Linke layer from the SODA data & a local DEM for delta-z. But
that just makes me wonder if turbidity really belongs in a 3D voxel grid,
not a 2D coverage map? i.e. I can derive a value for Linke at the
ground-surface for each raster cell in the DEM easily enough using the
above formulas, but isn't the important Linke value(s) what the beam
encounters on it's path through the atmosphere high above the ground,
not just the value at ground level at the beam's terminus?
You have to think on the Linke turbidity as an integrated value from top
of the atmosphere to the level you want the solar irradiance. That means
you dont need to worry about 3D voxel grids at all. If you retrieve Linke
values from Soda Service (www.soda-is.com) what I propose is to extract
values for sea level, for example, and then applied the
pressure-correction formula to shift vertically to your elevation grid.
Note that the extracted values will represent the integrated atmospheric
attenuation on the solar beam from top of the atmosphere to the sea level.
You can find useful information about the derivation of the Linke database
in the Soda Service in this link
(http://www.helioclim.net/publications/ises2003_linke.pdf). Particularly,
note that eq. (9) is the pressure-correction I proposed bellow. A somehow
similar approach can be found in
http://www.ing.unitn.it/~grass/conferences/GRASS2002/proceedings/proceedings/pdfs/Hofierka_Jaroslav.pdf
or in the paper Šúri M., Hofierka J., 2004. A New GIS-based Solar
Radiation Model and Its Application for Photovoltaic Assessments.
Transactions in GIS, 8, 2, 175-190
If it helps, I do have quite a bit of PAR light meter data we collected
at sea level throughout the region over full years; and do know the air
is nearly as clear as it gets. What instrumentation could I add to our
met stations to get a better record? Deploying light loggers on mountain
peaks for the summer months may be an option (granite is beautiful stuff
to climb :).
I dont understand what you mean with better record. It depends what you
want/need. If you really need a good characterization of the atmosphere
you would need spectral measurements but the instrumentation is expensive
and require very expertise knowledge. A good alternative is measuring beam
irradiance with a pyrheliometer (mounted in a sun-tracker). Using eqs
(6-7) in http://www.helioclim.net/publications/ises2003_linke.pdf you
could calculated the equivalent Linke turbidity from these direct
measurements. But even the pyrheliometers are somehow expensive and
difficult to maintain. A less accurate approach, but easier to operate and
maintain, could be the use of a shaded pyranometer to measure the diffuse
irradiance and another unshaded to measure the global (broadband) one.
Then, you could derive direct irradiance and calculate TL.
I hope this helps,
Jose
Hamish:
I have worried about r.sun's Linke Turbidity factor values in areas
with
big changes in elevation. Is turbidity value heavily dependent on
altitude?
Jose:
Linke turbidity is certainly height-dependent given the higher optical
path length at lower elevations. A simple approximated pressure
correction
might be applied if the turbidity at a given altitude z' is known:
TL(z) = TL(z') exp( -(z-z') / 8435.2)
Dylan:
It should, as it is based on a measure of optical thickness (air mass):
m = \frac{1}{sin(\alpha) + 0.15(\alpha + 3.885)^{-1.253}} e^{-0.0001184
\timesA}
(yay LyX, ctrl-m, paste selection, view PostScript)
buraq wrote:
I took the turbidity values from soda-is.com. I entered no altitude
value for the latitude and longitude.
I did the same, I attempted to match their model grid (approx 1x1-deg
IIRC)
for an array of lat/lon extractions over my study area, then v.surf.rst to
produce a Linke layer for r.sun. I can't recall off the top of my head if
we entered an altitude or not, I think not.
I work in fjords with >1000m vertical drops (including one of the top
10 tallest waterfalls in the world). With little idea of the elevation
values used for the SODA data, I just guess that it will be very rough
and subject to a somewhat arbitrary sampling error.
I assume I'll have to use one of the above formulas with r.mapcalc to
create the Linke layer from the SODA data & a local DEM for delta-z. But
that just makes me wonder if turbidity really belongs in a 3D voxel grid,
not a 2D coverage map? i.e. I can derive a value for Linke at the
ground-surface for each raster cell in the DEM easily enough using the
above formulas, but isn't the important Linke value(s) what the beam
encounters on it's path through the atmosphere high above the ground,
not just the value at ground level at the beam's terminus?
If it helps, I do have quite a bit of PAR light meter data we collected
at sea level throughout the region over full years; and do know the air
is nearly as clear as it gets. What instrumentation could I add to our
met stations to get a better record? Deploying light loggers on mountain
peaks for the summer months may be an option (granite is beautiful stuff
to climb :).
hope that makes sense,
Hamish
ps- If anyone is interested, at one point I wrote a little C program
that fit a sine curve to the monthly data and gave you a per-day value
from per-month data, avoiding big jumps at the month changes. The same
could in theory be adapted to work with r.mapcalc to get daily maps from
the monthly raster layers, at considerable computational cost. (but you
just need to run it once(ie x365) in a dedicated mapset)
_______________________________________________
grass-user mailing list
grass-user@lists.osgeo.org
http://lists.osgeo.org/mailman/listinfo/grass-user
--
José A. Ruiz-Arias
Solar Radiation and Atmosphere Modelling Group
http://www.ujaen.es/investiga/tep220
Physics Department, University of Jaén
Campus Lagunillas, Building A3 066
23071 Jaén Spain
Tlf: +34 953 212 474
Fax: +34 953 212 838
