Markus Neteler
cb3063d989
r.sun: mode 1 fix for time offset due to timezone, see https://trac.osgeo.org/grass/ticket/2941 for explanations (merge of https://trac.osgeo.org/grass/changeset/68254)
|
9 år sedan | |
|---|---|---|
| .. | ||
| Makefile | 12 år sedan | |
| README | 15 år sedan | |
| TODO | 11 år sedan | |
| local_proto.h | 15 år sedan | |
| main.c | 9 år sedan | |
| r.sun.html | 9 år sedan | |
| rsunglobals.h | 13 år sedan | |
| rsunlib.c | 13 år sedan | |
| sunradstruct.h | 13 år sedan | |
README
From Thomas.Huld jrc.it Thu Jul 26 11:01:44 2007
To: Markus Neteler
CC: Jaro Hofierka , "marcel.suri jrc.it"
, Tomas Cebecauer
Date: Thu, 26 Jul 2007 11:01:44 +0200
Subject: Re: r.horizon source code
Lines: 1533
I'm sending the r.sun sources to you here, even though the description is not
yet up to date. I think it works OK now, though I haven't tried all possible
combinations of options. For instance, I haven't ever used the incidence
angle or insolation time outputs. There's no reason why they should have
stopped working but there are always changes that break something unexpected.
The main changes are:
- possibility to read in horizon rasters to use with the shadow calculation.
- Calculation of the shadowing effect should now also work in lat/lon
projection.
- There was a bug in r.sun whereby the shadowing effect was calculated wrong
when the left-right and up-down directions in the projection are different
from east-west and north-south, as is the case in the projection we use
(Lambert Azimuthal) when moving away from the point of origin. This has been
fixed now though only checked with Lambert Azimuthal.
- I'm not sure if this went in already in a previous version, but the
algorithm now also takes into account the curvature of the earth when
calculating shadows.
- new output possibility: glob_rad, which is the sum of the three radiation
components
- new input parameter: civiltime. When this parameter is given, the
single time calculation will calculate the irradiance at the *same* time for
the entire raster instead of using the local solar time. The value of
civiltime is the timezone, relative to GMT. (+1 for central Europe)
- new input parameter numpartitions No change in the calculation
methods, but the program will read the input rasters and do the calculations
in a number of chunks, instead of reading in everything at the start. Output
is still only at the very end. This is only to save memory. It will not work
if you try to calculate shadows without the horizon information, and the
program will tell you so.
- I reorganized the source code quite a lot. In particular, I spun off several
functions into a separate source file called rsunlib.c. I also organized many
of the global variables into struct's for an easier overview. I did
this because we work here with several derivatives of r.sun:
* r.sunyear which calculates the optimum inclination angle for maximum
insolation
* r.pv which calculates the PV output taking into account also temperature
effects
* r.sun.2axis which is r.sun for a moving plane, whose normal always points
to the sun (two-axis tracking solar energy systems).
Splitting the source code made it easier to reuse bits for these r.sun
derivatives. I haven't updated the Makefile to account for these changes yet.
Look at the little script "compdebug" in the source directory.
There may be other changes that I've forgotten about but nothing dramatic.
Have fun!
Thomas
--
--------------------------------------------------
Thomas Huld
Joint Research Centre of the European Commission
T.P. 450
I-21020 Ispra, Italy
phone: +39 0332785273
e-mail: Thomas.Huld@jrc.it
--------------------------------------------------
To: Markus Neteler
CC: Jaro Hofierka , "marcel.suri jrc.it"
, Tomas Cebecauer
Date: Thu, 26 Jul 2007 11:01:44 +0200
Subject: Re: r.horizon source code
Lines: 1533
I'm sending the r.sun sources to you here, even though the description is not
yet up to date. I think it works OK now, though I haven't tried all possible
combinations of options. For instance, I haven't ever used the incidence
angle or insolation time outputs. There's no reason why they should have
stopped working but there are always changes that break something unexpected.
The main changes are:
- possibility to read in horizon rasters to use with the shadow calculation.
- Calculation of the shadowing effect should now also work in lat/lon
projection.
- There was a bug in r.sun whereby the shadowing effect was calculated wrong
when the left-right and up-down directions in the projection are different
from east-west and north-south, as is the case in the projection we use
(Lambert Azimuthal) when moving away from the point of origin. This has been
fixed now though only checked with Lambert Azimuthal.
- I'm not sure if this went in already in a previous version, but the
algorithm now also takes into account the curvature of the earth when
calculating shadows.
- new output possibility: glob_rad, which is the sum of the three radiation
components
- new input parameter: civiltime. When this parameter is given, the
single time calculation will calculate the irradiance at the *same* time for
the entire raster instead of using the local solar time. The value of
civiltime is the timezone, relative to GMT. (+1 for central Europe)
- new input parameter numpartitions No change in the calculation
methods, but the program will read the input rasters and do the calculations
in a number of chunks, instead of reading in everything at the start. Output
is still only at the very end. This is only to save memory. It will not work
if you try to calculate shadows without the horizon information, and the
program will tell you so.
- I reorganized the source code quite a lot. In particular, I spun off several
functions into a separate source file called rsunlib.c. I also organized many
of the global variables into struct's for an easier overview. I did
this because we work here with several derivatives of r.sun:
* r.sunyear which calculates the optimum inclination angle for maximum
insolation
* r.pv which calculates the PV output taking into account also temperature
effects
* r.sun.2axis which is r.sun for a moving plane, whose normal always points
to the sun (two-axis tracking solar energy systems).
Splitting the source code made it easier to reuse bits for these r.sun
derivatives. I haven't updated the Makefile to account for these changes yet.
Look at the little script "compdebug" in the source directory.
There may be other changes that I've forgotten about but nothing dramatic.
Have fun!
Thomas
--
--------------------------------------------------
Thomas Huld
Joint Research Centre of the European Commission
T.P. 450
I-21020 Ispra, Italy
phone: +39 0332785273
e-mail: Thomas.Huld@jrc.it
--------------------------------------------------