[R-sig-Geo] rsaga, radiation, rsaga.pisr

[Melita Percec Tadic]

Dear Alex,

thank you for the answers and suggestions. I checked what you suggested 
(the saga forum also) and the only thing left is how to avoid the latitude 
definition in case when in.latitude.grid is specified.


At 19:48 28.11.2011, Alexander Brenning wrote:
>Dear Melita,
>
>I will try to answer your questions, but you will likely get better 
>feedback from the SAGA pros if you post them to the SAGA user forum on 
>sourceforge, see
>http://sourceforge.net/projects/saga-gis/support
>
>>Dear saga and rsaga users and developers (Olaf, Alex, Victor..)<br><br>
>>I ran a small piece of  code, testing the rsaga.pisr function for
>>calculation of potential solar radiation.<br><br>
>><ul>
>><li>Question is: if I calculate the rsaga.pisr for April 1th till April
>>4th, day by day, sum those and compare it with the rsaga.pisr calculated
>>for the range of days April 1-4th, there is an unexpected difference
>>ranging from (4.821417, 6.996113)kWh/m2. Can you help me to fix or
>>explain this?
>
>Solar radiation was likely only calculated until April 3rd. I haven't 
>tried this out with rsaga.pisr, but some time ago in an earlier version I 
>found this in rsaga.solar.radiation, i.e. a different SAGA modules that 
>takes similar arguments:
>
>>In SAGA 2.0.2, solar radiation sums calculated for a range of days, say 
>>days=c(a,b) actually calculate radiation only for days a,...,b-1 (in 
>>steps of day.step - I used day.step=1 in this example). The setting a=b 
>>however gives the same result as b=a+1, and indeed b=a+2 gives twice the 
>>radiation sums and potential sunshine duration that a=b and b=a+1 both give.
>
>This might explain your problem, but if you want to be sure you'd better 
>examine this in more detail by looking at different time spans, especially 
>one versus two days (does radiation increase by a factor of 2?)

Thanks for the confirmation. That is what I also assumed.
But it seams it does not include the first but the last day of the range of 
days in the calculation. I checked the calculation for the April 1st and 
April 2nd individually. Insolation for the range of days 1-2 April equals 
to pisr for the April 2nd. The difference of those grids are 0 everywhere.

And actually it writes in the R session:
initialising gradient...
day 92(91-92), local time 05:00
day 92(91-92), local time 08:00
day 92(91-92), local time 11:00
day 92(91-92), local time 14:00
day 92(91-92), local time 17:00
day 92(91-92), local time 20:00
day 92(91-92), local time 23:00

and that is April 2nd.

It is not so perfect if I compare several (4) days. Most of the cells have 
0 difference but some diff. are larger. But not so large that I couldn't 
work with that.


>><li>Second question is a technical one: when the R script calls
>>rsaga.pisr I can not have the SAGA gui opened since it crashes down. Is
>>that expected?
>
>no it isn't, and I haven't experienced this problem with SAGA 2.0.7 on 
>Windows.

I also run SAGA 2.0.7 on Windows.
It seems the problem was with the available working memory. I ran the 
example on a large grid. When I made a smaller selection, it didn't crash 
any more.

Thanks again.


>><li>Considering the latitude=user defined, why do we need it since the
>>latitude grid is defined?
>
>not all the arguments are mandatory. you would specify either latitude or 
>in.latitude.grid

I tried the command with in.latitude.grid specified and
1. without  latitude in the command. I got an error:
Error in latitude >= -90 : 'latitude' is missing

2. vith latitude=NULL
but than also latitude grid was neglected

3. I have no idea what else to try. I will put latitude=45 instead.

So this seems the only unsolved question. Not bad at all! Thanks.


>><li>And finally, there is a comment concerning the units in the
>>rsaga.pisr when you chose kJ/m2. If I'm not mistaken, the resulting grid
>>is actually in MJ/m2.
>
>sorry I cannot confirm this, but it should be possible to determine that 
>by comparing average hourly PISR with the solar constant times 1 hour, 
>which will be higher but of the same order of magnitude as PISR if in the 
>same units. If that doesn't help, please follow up in the SAGA GIS forums 
>to find out if that's an error and issue a bug report if necessary.

I can explain this.
If pisr is calculated with the option: unit=c("kJ/m2") it is exactly 3.6 
times larger compared to one calculated with the option: unit=c("kWh/m2")
but:
kW h = [h=3600s] = 3600 kWs = [W = J/s] = 3600 kJ = 3.6 MJ

So calculation is OK, only for the calculated grid the units should be: 
"units=MJ/m2".

Smaller example is on http://radar.dhz.hr/~melita/

Kind regards,
Melita.


>I hope this helps
>   Alex
>
>
>></ul>The R script and input grids are provided in
>><a href="http://radar.dhz.hr/~melita" eudora="autourl">
>>http://radar.dhz.hr/~melita<br><br>
>><br>
>></a>Thank you for the help and a nice tools that we can all use and
>>benefit from it.<br><br>
>>Regards,<br><br>
>>Melita Percec Tadic<br><br>
>
>
>
>Date: Fri, 25 Nov 2011 16:52:57 +0100
>From: Melita Percec Tadic <melita at cirus.dhz.hr>
>To: r-sig-geo at r-project.org
>Subject: [R-sig-Geo] rsaga, radiation, rsaga.pisr
>Message-ID: <6.2.1.2.2.20111125163122.081c4ac0 at cirus.dhz.hr>
>Content-Type: text/plain; charset="us-ascii"
>
>An HTML attachment was scrubbed...
>URL: 
><https://stat.ethz.ch/pipermail/r-sig-geo/attachments/20111125/2faacf51/attachment-0001.html>
>
>
>--
>Alexander Brenning
>brenning at uwaterloo.ca - T +1-519-888-4567 ext 35783
>Department of Geography and Environmental Management
>University of Waterloo
>200 University Ave. W - Waterloo, ON - Canada N2L 3G1
>http://www.environment.uwaterloo.ca/geography/faculty/brenning/
>
>__________ NOD32 6667 (20111128) Information __________
>
>This message was checked by NOD32 antivirus system.
>http://www.eset.com
>