[QGIS-Developer] [Qgis-user] Wrong result with Saga Raster calculator

[Pedro Venâncio]

Hi Eugenio,

I've opened a ticket in QGIS bug tracker, to try to identify the source of
the problem:

https://issues.qgis.org/issues/16646

Please see the recent findings and give your feedback there.

Thanks!

Best regards,

Pedro Venâncio





2017-06-01 8:43 GMT+01:00 Eugenio Trumpy <frippe12573 at hotmail.com>:

> Hi Pedro,
>
>
> I did the same for 5 pixels in the test raster I shared with you yesterday.
>
> The different seems to be a little bit greater.
>
> See the attached sheet.
>
>
> I don't know if the version of saga, 2.3.1 the mine vs 2.3.2 the yours
> could generate this error,
>
> or if the problem is in qgis process model.
>
>
> E.
>
>
>
>
>
>
> ------------------------------
> *Da:* Pedro Venâncio <pedrongvenancio at gmail.com>
> *Inviato:* mercoledì 31 maggio 2017 22:16
> *A:* Eugenio Trumpy
> *Cc:* qgis-user; qgis-developer
> *Oggetto:* Re: [Qgis-user] Wrong result with Saga Raster calculator
>
> Hi Eugenio,
>
> I've created a sample processing model (attached), that works ok, except a
> small error in the result (decimals).
>
> For instance:
>
> *Raster* *ID* *Pixel value*
> Z_top_res a 1202,03510276
> t_top_res b 53,6123428345
> topo c1 -182,584899902
> SAGA Result 1 c 15,1783676147
> SAGA Result 2 d 31,9437332153 *Diff*
> Expected Result 1
> 15,1783679504 0,000000335741
> Expected Result 2
> 31,9740872222 0,0303540069
> *Formula*
> 1 ifelse(c1>0,-(0.0065*c1-16),(0.0045*c1+16))
> 2 ifelse(a<500,0,((b-c)/a)*1000)
> Using SAGA GUI there is no error.
>
> Could this be because of some roundings in the input / output? (I'm
> sending for the developer's list also).
>
> Thanks!
>
> Best regards,
> Pedro Venâncio
>
>
>
> 2017-05-31 10:56 GMT+01:00 Eugenio Trumpy <frippe12573 at hotmail.com>:
>
>>
>> Hi Pedro,
>>
>> thank you very much to have tested the computation with my test data!
>> You are right, I checked again the computation and actually it works.
>>
>> However, I was using Saga raster calculation in a processing modeler, and
>> one of the layer I share with you
>> is the resulting from a previous computation.
>>
>> In other word the raster layer c=temp_recal.tif comes from a previous
>> computation (raster calculator):
>> ifelse(a>0,-(0.0065*a-16),(0.0045*a+16))
>> where a is the layer attached, and hence c of the previous  formula is
>> the resulting of this first computation.
>>
>> E.
>>
>> ------------------------------
>> *Da:* Pedro Venâncio <pedrongvenancio at gmail.com>
>> *Inviato:* mercoledì 31 maggio 2017 11:37
>> *A:* frippe12573; qgis-user
>> *Oggetto:* Re: [Qgis-user] Wrong result with Saga Raster calculator
>>
>> Hi Eugenio,
>>
>> The result with SAGA Raster Calculator seems ok here, on QGIS 2.18.9,
>> SAGA 2.3.2, both from OSGeo4W64 (result attached).
>>
>> Z_top_res.tif have values from 724.403 to 1423.553, and so, the first
>> condition is never true. Then, it always make ((b-c)/a)*1000, and the
>> result seems ok to me.
>>
>>
>> Pedro
>>
>>
>>
>> 2017-05-31 10:19 GMT+01:00 frippe12573 <frippe12573 at hotmail.com>:
>>
>>> Hi Paolo,
>>>
>>> the raster have the same size and origin.
>>> The pixels are precisely overlapped.
>>>
>>> I tried also r.mapcalc, but I got a worst result.
>>> Here the output of the log:
>>> https://pastebin.com/jNyyzHgf
>>>
>>> E.
>>>
>>>
>>>
>>> --
>>> View this message in context: http://osgeo-org.1560.x6.nabbl
>>> e.com/Wrong-result-with-Saga-Raster-calculator-tp5322429p5322442.html
>>> Sent from the QGIS - User mailing list archive at Nabble.com.
>>> _______________________________________________
>>> Qgis-user mailing list
>>> Qgis-user at lists.osgeo.org
>>> List info: https://lists.osgeo.org/mailman/listinfo/qgis-user
>>> Unsubscribe: https://lists.osgeo.org/mailman/listinfo/qgis-user
>>>
>>
>>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.osgeo.org/pipermail/qgis-developer/attachments/20170602/7c387d6e/attachment.html>