I had SPI (Standard precipitation index) values for 1000 stations (30 years monthly data for each station) . I needed to perform the interpolation method ( kriging). What confuses me is that each station/point has 30 years of monthly data ranging from +2 to -2, Can anyone suggest to me how to interpolate these values, thanks in advance.

well generally for each point in time you would need run the kriging process separately (each monthly data set for all points counts as one), so that would be 30*12 = 360 times …
So a lot of runs … if you really wanted to do that best would be to run that as a batch process or script it in python for the processing toolbox…
Another good tool to look at for such things is the R Program https://www.r-project.org/ and possibly using Raster instead of vector data to create stacks of precipitation time series…

However, if I where you, I might step back first and determine what results you would like to get or which comparisons you would want to make…
I don’t know what you are after - but for example would it make more sense to look at the differences of one month over the 30 years time sequence to detect changes in precipitation?
This would make sense if you where to find dry versus wet years e.g. for the growing season of crops or the like ?
Let’s say if July was interesting for you then you could run the interpolations for July of each year and that way get a time sequences you can look at to dervive your conclusions from…

Cheers
Karsten

Karsten Vennemann
Principal

Terra GIS LTD
2119 Boyer Ave E
Seattle, WA 98112
USA www.terragis.net

I had SPI (Standard precipitation index) values for 1000 stations (30 years monthly data for each station) . I needed to perform the interpolation method ( kriging). What confuses me is that each station/point has 30 years of monthly data ranging from +2 to -2, Can anyone suggest to me how to interpolate these values, thanks in advance.

Thanks for your reply, please accept my apologies for the late reply.

like you said, if i focus on the month of july ( for example) can i take
the average SPI values for all the 30 years to denote/represent a station?

Best
Dani

On Thu, Sep 3, 2020 at 9:26 PM karsten <karsten@terragis.net> wrote:

Hi Dani.

well generally for each point in time you would need run the kriging
process separately (each monthly data set for all points counts as one), so
that would be 30*12 = 360 times ...
So a lot of runs ... if you really wanted to do that best would be to run
that as a batch process or script it in python for the processing toolbox...
Another good tool to look at for such things is the R Program https://www.r-project.org/ and possibly using Raster instead of vector
data to create stacks of precipitation time series...

However, if I where you, I might step back first and determine what
results you would like to get or which comparisons you would want to
make....
I don't know what you are after - but for example would it make more sense
to look at the differences of one month over the 30 years time sequence to
detect changes in precipitation?
This would make sense if you where to find dry versus wet years e.g. for
the growing season of crops or the like ?
Let's say if July was interesting for you then you could run the
interpolations for July of each year and that way get a time sequences you
can look at to dervive your conclusions from....

Cheers
Karsten

Karsten Vennemann
Principal

Terra GIS LTD
2119 Boyer Ave E
Seattle, WA 98112
USA
www.terragis.net

Phone ++1 206 905 1711
Fax ++1 925 905 1711

------------------------------
*From:* Qgis-us-user [mailto:qgis-us-user-bounces@lists.osgeo.org] *On
Behalf Of *Dani Varghese
*Sent:* Thursday, September 03, 2020 03:14
*To:* qgis-us-user@lists.osgeo.org
*Subject:* [Qgis-us-user] Standard precipitation index interpolation

Dear All

I had SPI (Standard precipitation index) values for 1000 stations (30
years monthly data for each station) . I needed to perform the
interpolation method ( kriging). What confuses me is that each
station/point has 30 years of monthly data ranging from +2 to -2, Can
anyone suggest to me how to interpolate these values, thanks in advance.

like you said, if i focus on the month of july ( for example) can i take the average SPI values for all the 30 years to denote/represent a station?
I assume you meant using the July value for each of the 30 years i.e. averaging 30 values for each individual point.
You could do that but it would average out many differences over the timeline of 30 years - which likely (with assumed climate change) would show smaller differences among location points as a result. That still might work - depending what you are trying to do …
But you could also average always a decade of data for each point to get 3 SPI averaged values and determine if that gives you any different results than the first approach.
Also you need to consider that the SPI already is a summarized value which shows deviations itself for a long term precipitation series… So overall you might be better off to calculate the standard precipitation index yourself from a given time series of precipitation values. That way you can calculate for your preferred time period and get only one value , instead of averaging out already averaged statistical values…

You can also take a look at these RASTER data source /www.chc.ucsb.edu/data/chirps
I used those for areas in Africa to determine dry vs. wet years (calculated in R)

well generally for each point in time you would need run the kriging process separately (each monthly data set for all points counts as one), so that would be 30*12 = 360 times …
So a lot of runs … if you really wanted to do that best would be to run that as a batch process or script it in python for the processing toolbox…
Another good tool to look at for such things is the R Program https://www.r-project.org/ and possibly using Raster instead of vector data to create stacks of precipitation time series…

However, if I where you, I might step back first and determine what results you would like to get or which comparisons you would want to make…
I don’t know what you are after - but for example would it make more sense to look at the differences of one month over the 30 years time sequence to detect changes in precipitation?
This would make sense if you where to find dry versus wet years e.g. for the growing season of crops or the like ?
Let’s say if July was interesting for you then you could run the interpolations for July of each year and that way get a time sequences you can look at to dervive your conclusions from…

Cheers
Karsten

Karsten Vennemann
Principal

Terra GIS LTD
2119 Boyer Ave E
Seattle, WA 98112
USA www.terragis.net

I had SPI (Standard precipitation index) values for 1000 stations (30 years monthly data for each station) . I needed to perform the interpolation method ( kriging). What confuses me is that each station/point has 30 years of monthly data ranging from +2 to -2, Can anyone suggest to me how to interpolate these values, thanks in advance.