Dear Karl
Thank you very much for your reply.
My problem is not systematic, it is only in some areas across the study
site. I have tried to improve the fit between the satellite images and
the digitased hard copy maps by many different ways (georeferencing,
orthorectifications) but I haven't succeeded. At this point I think it
is a problem with the maps and not the images.
S
Karl Hennermann wrote:
> Dear Shaenandhoa,
> an obvious coordinate shift is a data error, and you should correct it
> before doing any analysis.
>
> A systematic shift of 100-200m in most cases is due to differences in
> coordinate systems or coordinate transformations; or due to the way
> specific software handles coordinates.
>
> However, to investigate the cause of your coordinate shift, one would
> need to see the data and your previous processing steps.
>
> Hope this helps
>
> Karl Hennermann
> IT Officer - Spatial Data Research
> School of Environment and Development
> The University of Manchester
> Tel: +44(0)161 275 3655
>
>
>
> Quoting Shaenandhoa García Rangel <[log in to unmask]>:
>
>> Dear members of the list,
>>
>> I am writing to seek advice regarding minimum mapping units (MMU) and
>> the geographic fit of different data sources. Currently, I am working
>> on a habitat availability map that I develop for my study site
>> )Sierra de Portuguesa - Venezuela Andes), based on the needs of the
>> Andean bear (/Tremarctos ornatus/). I chose a MMU of 1hectare for
>> the output and the Kappa analysis yield 69% accuracy for the overall
>> map and 82% for Andean bear habitat. Using this map I have estimated
>> area for habitat available, an evaluated the effects of different
>> edge depths and fragmentation scenarios.
>>
>> I would now like to run an analysis to evaluate factors involved on
>> habitat prevalence such as rivers, roads, distance to villages etc.
>> The issue I have here is that the shift between the satellite images
>> and the topographic maps that provide these data is about 100-200m
>> and thus the MMU would at most has to be 2 hectares. With this MMU,
>> the kappa for the overall map is about 54% and the accuracy for the
>> habitat available is reduced as well regarding commission error.
>> Thus, I was wondering if I could run these two analysis with
>> different MMU, in order to take the full advantage of a higher
>> accuracy in the estimation of habitat extent, it would be a shame to
>> loose accuracy given possible errors introduced by hard copy maps.
>>
>> I would really appreciate any advice on this subject,
>>
>> S
>>
>> Shaenandhoa García Rangel
>> PhD Candidate
>> Wildlife Research Group
>> University of Cambridge
>>
>
>
|