Hi,
I am hoping to produce rainfall and temperature maps for a current project in QGIS.
I have csv data downloaded which have been imported into QGIS.
My aim is to be able to produce rainfall distribution maps. I have attemtped ordinary kriging with some degree of success but the final product is not good.
Any help or advice would be appreciated.
Hello Sean, It is possible to generate such kriging results. Even, with the use of ArcGIS.
If you know how to use ArcGIS, try to use it and see how the output will look like.
But if the result is the same, it means YOUR POINTS (coordinates) are not well distributed.
You can now use IDW interpolation method.
Best of luck.
Hi, i agree with @Salami
Kriging required a great number of sample, at least 30 in arcgis, to run and produce relatively good results, more samples need to be well distribute spatially to considere the spatial extension of the phenomenum.
An alternative to kriging is IDW interpolation method.
Both method can be run on QGIS
Sincerly
Hi,
Thanks for your reply.
The data downloaded is all island rainfall data consisting of over 4000 data points in 1x1km square grids to it is evenly distributed across the mapping area. (.csv file).
Unfortunately I only have access to QGIS.
The IDW method does not appear to work either I fear i may be incorrectly inputting parameters somewhere along the line.
Many thanks
Sean, the primary issue you are facing is that precipitation data is rarely uniformly distributed even across a small area. Work I did years ago across an area somewhat larger than your 1x1km area with good distribution, for conventional stratiform precipitation... about as evenly distributed as possible... demonstrated wide variations. Simple krieging such as TINs don't do justice to the (apparent) random distribution of rainfall. I haven't tried it, but you might want to look at Voronoi interpolation to create your surface. While Voronoi polygons can be 3-vertex shapes, a larger vertex selection or freeing the constraint and allowing the system to select the required degrees of freedom have been known to produce better results. QGIS does have Voronoi plugins available.
This is less an error in your GIS process than in the nature of the dataset. And location of your gauges might be evenly distributed but your data didn't cooperate.
I always use Surfer (Golden Software) for surface modeling....the results are far superior to anything ArcGIS or QGIS can produce. 4000 pnts should give a good result - especially if you elect to eliminate areas with no data before running the krig. You could try Surfer on a trial basis and see if it gives the results you are wanting. It is a very inexpensive software, relative to other mapping tools, and the support is out of this world!!
- Badges
- Science topic
- Similar topics
- Geoscience
- Cartography
- Mapping