combining data

[Yi-JyunLin]

Is it possible to combine data sets with different spatial resolutions? For example, I have a temperature data is 0.5° * 0.5° resolution, and have a sea level data is 2° * 2° resolution. I wonder is it possible to combine them and how?

[DanPatterson_Retired]

two routes Resample—Help | ArcGIS for Desktop 

                  Aggregate—Help | ArcGIS for Desktop 

read the fine print.

And by 'combine' do you mean Combine—Help | ArcGIS for Desktop as the real combine, or do you mean merge or some overlay method/

[Yi-JyunLin]

Hello Dan, thank you.

By "combine", I mean merge or overlay method (because I will also need to export the merged data to run statistical analyses). Does this make sense?

BTW, is the "combine" tool used to interpolate values as well?

[CodyBenkelman]

Yi-Jyun

The best method is to use the Mosaic Dataset.  It gives you a great deal of flexibility for configuration, including resampling, reprojecting, and focusing on one dataset at a time if required.  Please see What is a mosaic dataset?—Help | ArcGIS for Desktop  

In most cases I would recommend against resampling your data to write a new dataset to disk, since any resampling will always degrade the quality of your data.

What version of software are you using?

Cody B.

[Yi-JyunLin]

Thank you, Cody. 

I am reading mosaic dataset now. I use ArcGIS 10.3.

[DanPatterson_Retired]

Combine is used to produce a layer of the unique combinations of overlaying rasters...inherent to this, is the fact that your data are categorical... ie classed... nor raw... It is akin to producing the unique combinations that are needed for nonparametric statistical tests like the Chi Square test.  

So if you had 2 rasters with 2 classes, it would produce a summary

Raster 1   Raster 2

1              1

1              2

2              1

2              2

Raster 1 could be any variable as could raster 2, but there would be wanting to see whether these 2 variables have any spatial co-association.  Similarly, they could represent the same variable for 2 different time periods.

If your use of the term 'combine' is to produce a 'picture' or 'image in the remote sensing sense, then combine is not what you want

[Yi-JyunLin]

Thank you, Dan. I think I understand "combine" better.

I, then, think this is not the tool I can use. I have several data sets needed to be put together: temperature data, rainfall data, drought and flood data. These variables are all measured by numerical values and are all grid data sets (with different resolution). So, I first try to visualize these data sets on maps and then use these variables to run statistical models. Next, I need to merge these data sets with a survey data, which I have information about where the survey was conduct (such as city or county), and I wonder is it possible to merge this survey data and the above mentioned climate data sets? If so, how? (Can I using the lat/long of these location where the survey was conducted and the lat/long of the climate data sets?)

[CodyBenkelman]

Hi Yi-Jyun

I need to apologize for giving you incorrect advice.  

I did not read your question carefully, and I focused on "different spatial resolutions" but I should not have advised you to combine different types of data (temperature and sea level elevation values) in a Mosaic Dataset.  

If you had MANY datasets, you could use a Mosaic Dataset to visualize and manage the data before the analysis steps, so my advice was not entirely wrong -- but working inside the Mosaic Dataset to perform the combinations may not be efficient.

I am not an expert in using ArcGIS for this type of spatial analysis, so I probably cannot advise further - but there is one question I think we should clarify.  Dan mentioned categorical data, but does your data represent class values, or are the data values (temperature, sea level, rainfall...) continuous?  

Cody B.

[Yi-JyunLin]

Hi Cody, thank you. Data values (temperature, sea level, rainfall...) are continuous.

[DanPatterson_Retired]

There is no need to combine... in any sense of the word... your data to do any kind of analysis.

You are going to need to articulate exactly what you need.

Is the need to 'combine'/'group' whatever, the data purely for convenience? it is a prerequesite for some type of analysis? If that is the case, then please elaborate.