I have two test shape files (more than 50 points in each file) with slightly varied latitudes and longitudes, but same area (Bay of Bengal). Each file contains Latitude, Longitude and Data_Field as common headers.
I want to take average of the Data_field from these two files (may be nearest points of shape file1 and shape file 2) (an illustrate picture attached).
Eg.
Shape file 1 Lat1 Long1 Data1
Shape file 2 Lat1 Long1 Data1
Shape file 3 Lat1 Long1 Data1
Shape file 4 LatNIL LongNIL DataNIL
--------------------------
Shape file x Lat1 Lat1 Data1
--------------------------
I have read methods regarding Point Statistics and Cell Statistics in ArcGIS, but not succeed.
It doesn't matter about slight reduction of resolution, because I have high resolution data.
Two test shape files are attached here along with an illustration image of expected output.
Kindly help me to do this in ArcGIS.
Thanks a lot.
you can watch this video ......https://www.youtube.com/watch?v=M9-zPreo5sU
Thanks for your suggestion Vasu. But I did a mistake in the example of the question (now corrected). I want to take means from two or more shape files (little varied locations in each shape file).
The "right" answer depends on where your data came from and what your desired outcome is, though I will explain one method to derive values for spatially weighted means of values from your points.
Providing that your two point sets are of an equivalent sampling effort your first step, as Vasudevan Sankar suggested, is to merge the points into one unified shapefile or feature class. This method gives generally equal weight to the two data sets.
Then you may create an interpolated raster surface which estimates the data values between the point locations. To get a generally linear (i.e., first order polynomial) trend between the nearest datapoints, which I assume is the goal, the "Spline" interpolation tool should be used with a high weight (i.e., weight = 10). This will provide essentially a spatially weighted averagation of your combined data set. "Natural Neighbor" or other methods may be more appropriate if you are interested in points outside of your nearest two. Two use a defined radius for the interpolation the IDW (Inverse distance weighting) tool would work.
You can then use the "Extract Values to Points" to collect the resultant interpolated values to the precise points you want your predictions to occur at (i.e, at intervals of 30 minutes of a degree). The placement of these points may be better determined based on your original sampling methodology.
See the attached image, which shows the tension spine result between your test points. The grid overlay is arranged at 30' intervals as described.
Also, keep in mind that if your two sets of data are representative of two unique sampling conditions (e.g, collected on different dates) it may be more appropriate to create two interpolated surfaces, one for each sampling event, then average the two results (using raster calculator or the image classification window's mean mosaic operator). This will add complexity to the resultant surface, but may more meaningfully describe the combination of the data you are interested in.
I hope that this method works well for your research.
Regards,
Matthew Bolton
Thank you very much Matthew Bolton,
Good method to extract the mean data from multi sources. Actually I have the satellite based temperature data for each month, but I need to derive a seasonal picture. Some of those monthly data sets having cloud cover. So this method will be useful to derive seasonal mean.
Thank you.
To get a seasonal mean you'll have to create a surface for each month and take the average of your 12 rasters. May I also suggest that you may get better - "more correct" surfaces if you simply modify your input satellite data by only filling in (and thereby changing) the values within cloudy areas. This presumes that you already have your clouds classified. If not, maximum likelihood classification or other classification method may be used.
If you set your cloud areas to "null" an equation like:
Con(IsNull(Raster), FocalStatistics(Raster, NbrCircle(10, "CELL"), "MEAN"), Raster)
in raster calculator (or something very similar) should work to interpolated a focal mean of the surrounding data in clouds. See http://support.esri.com/technical-article/000004792 for more information.
This method is preferred because you are using the direct (unchanged) satellite data for non-cloud cells. In this way, the majority of your cells (i.e, non-cloud cells) will be as accurate as your original data.
If you use an interpolation method (as described in my earlier answer) you are inherently simplifying the monthly observations. The "right" method all depends on the details of the data and the analysis required.
Regards,
Matthew
Thank you very much for your descriptive answers. It will be useful to me and many of the researchers.
Thank you.
- Badges
- Science topic
- Similar topics
- Biological Science
- Biota
- Eukaryota
- Fungi