I have monthly observation from 13 station for my case study basin. I need to make gridded product out of those point data for zonal statistics through ArcGIS. Therefore, I am looking for an appropriate method of interpolation. I can work in QGIS, ArcGIS and GraDS, please suggest.
Thanks all for the response and suggestions. Could you please send R script for Kriging?
Krigging and IDW are good methods when your basin is relatively flat, but for temperature I will suggest to find out how temperature changes with elevation. It should be almost linear depends on the elevation range. Then use the reference station, DEM and temperature lapse rate for interpolation.
Hello Mr Jeeban Panthi,
I would suggest to use Regression-krigging method, based on regression method (ej. elevation, coastline distance, etc) and krigging for residuals:
http://www.sciencedirect.com/science/article/pii/S0016706103002787
http://spatial-analyst.net/wiki/index.php?title=Regression-kriging_guide
In case of Temperature interpolation main consideration are your basin characteristics! whether you do interpolation using spatial analyst or R doesn't matter.
Depending on your basin characteristics i.e) orographic or plains and whether tropical or temporate you have to discover your site specific algorithm.
For temperature interpolation good to use Geomedi-Grid Densify, K+ kriging method for interpolation.
Dear Jeeban,
You can use kriging method in ArcGIS to interpolate your data, but there is a problem: 13 stations could be not enough to obatin a great variogram to perform kriging method. A suggestion that I would like to share with you is to obtain thermal bands from Landsat-8 or 5, which you will be able to calculate the estimation of surface temperature. So, with this information, you can perform a correlation analysis with the 13 stations. If the correlation results are good, you can use the thermal bands to represent the temperature in your study basin, If not, you can perform deterministic interpolation methods like IDW, TIN with the 13 stations.
Best regards!!
I agree with Diego. If you've got just 13 stations kriging will not be better than IDW. But if you go for the kriging, don't forget to calculate a variogram, because if you just use the built-in linear function, the result may be worse than the IDW.
You have a deterministic data, so the best way according to me is the kriging.
Kriging
http://www.ncgia.ucsb.edu/conf/SANTA_FE_CD-ROM/sf_papers/collins_fred/collins.html
Dear Jeeban, You can use both GIS and GS+ software for selecting the best spatial-interpolation mothed. In GIS, you should use from GEOSTATISTICAL extension. using it you can compare them together. Indeed you will calculate RMSE, MAE and etc. and decide about the best method.
Hello Jeeban, although there are many works that propose different methods, due to the nature of this variable, you may try regression kriging. the advantage of this method is that it combines a regression model (for example allows you to see the influence of the sea distance or altitude) and kriging interpolation for residuals. In this article (T. Hengl)
https://dx.doi.org/10.1016%2Fj.geoderma.2003.08.018
And use QGis, R andGSIF package for RK
- Badges
- Science topic