Getting kriging interpolated grid with None values where the samples are far from the semivariogram range

[carlosfelg]

How can I get an interpolated grid with None (or no data) values where all the data neighbours (samples) are far from the variogram range value? I mean, I would like to use the variogram range value as constraining when there is no input data inside the range radio.

[MuellerSeb]

What do you mean by "variogram range value"? I really don't understand what you mean.
To try a shot in the dark, one thing of interest could be the estimated mean for ordinary krging. Do you mean that?

[carlosfelg]

Dear Sebastian, Thanks a lot for your reply. I am considering that the variogram (used for ordinary kriging) has 4 parameters: Model (spheric, exponential, gaussian,...), Nugget Effect, Contribution (nugget effect + contribution= sill) and Range (the maximum distance for spatial data correlation). I have been used the pykrige ordinary kriging and I have found that the interpolation is made in all of the grid locations, even when THERE IS NO INPUT SAMPLES CLOSE (I mean with distance lower than the variogram range value) to some grid locations. I was expecting that the pykrige ordinary kriging will return a None value (no data, meaning no interpolation) for the grid locations far from the samples. I hope I have been clear. i look forward to hearing from you. Best regards Em 26-01-2020 18:13, Sebastian Müller escreveu:

What do you mean by "variogram range value"? I really don't

understand what you mean.

To try a shot in the dark, one thing of

interest could be the estimated mean for ordinary krging. Do you mean that?

-- You are receiving this because you authored the

thread.

Reply to this email directly, view it on GitHub [1], or

unsubscribe [2].

…

-- Carlos A Felgueiras Computer Science PhD SPRING GIS Software Manager Image Processing Division - DPI Brazilian National Institute for Space Research - INPE Fone: (12)3208-6469 or (12)981449544 Links: ------ [1] #134?email_source=notifications&email_token=ALDKQOCDLBORBB5MSN4FLFTQ7X4HNA5CNFSM4KGG4ALKYY3PNVWWK3TUL52HS4DFVREXG43VMVBW63LNMVXHJKTDN5WW2ZLOORPWSZGOEJ56BGQ#issuecomment-578543770 [2] https://github.com/notifications/unsubscribe-auth/ALDKQOES3XQPMKZJAZD4KALQ7X4HNANCNFSM4KGG4ALA

[MuellerSeb]

In OrdinaryKriging.execute

PyKrige/pykrige/ok.py

Line 578 in a31dbbb

def execute(self, style, xpoints, ypoints, mask=None, backend='vectorized',

you can set n_closest_points to only take a certain amount of nearby values into account. Maybe this is helps?

[carlosfelg]

What distance value is considered, inside the ordinary kriging function, to define the closest points? The n_closest_points parameter is the minimum or the maximum number of closest points?

[carlosfelg]

Dear Sebastian, I have tried to use the n_closest_points parameter but it seems that it does not work properly. What distance value is considered, inside the ordinary kriging function, to define the closest points? The n_closest_points parameter is the minimum or the maximum number of closest points? Regards - Carlos A Felgueiras Computer Science PhD SPRING GIS Software Manager Image Processing Division - DPI Brazilian National Institute for Space Research - INPE Fone: (12)3208-6469 or (12)981449544 Em 28-01-2020 07:54, Sebastian Müller escreveu:

In OrdinaryKriging.execute [1]

https://github.com/GeoStat-Framework/PyKrige/blob/a31dbbb9ee22d89da2378f28dc11c0d882ae564d/pykrige/ok.py#L578 [1]

you can set n_closest_points to only take a certain amount of

nearby values into account. Maybe this is helps?

-- You are

receiving this because you authored the thread.

Reply to this email

directly, view it on GitHub [2], or unsubscribe [3]. - Links: ------ [1] https://github.com/GeoStat-Framework/PyKrige/blob/a31dbbb9ee22d89da2378f28dc11c0d882ae564d/pykrige/ok.py#L578 [2] #134?email_source=notifications&email_token=ALDKQOH7CEEFTYKFOXOLSU3RAAFH5A5CNFSM4KGG4ALKYY3PNVWWK3TUL52HS4DFVREXG43VMVBW63LNMVXHJKTDN5WW2ZLOORPWSZGOEKC3UTA#issuecomment-579189324 [3] https://github.com/notifications/unsubscribe-auth/ALDKQODS6FCDQXF3DWZFI3DRAAFH5ANCNFSM4KGG4ALA

[MuellerSeb]

n_closest_points is exact unless there are less data points in total. The n clostest points are determined by scipys KDTree

...but it seems that it does not work properly.

What exactly do you mean?

[carlosfelg]

My doubt is: What is the distance value for a point to be considered a closer point? Em 28-01-2020 11:02, Sebastian Müller escreveu:

n_closest_points is exact unless there are less data points in total. The n clostest points are determined by scipys KDTree [1]

> ...but

it seems that it does not work properly.

What exactly do you mean? -- You are receiving this because you modified the open/close

state.

Reply to this email directly, view it on GitHub [2], or

unsubscribe [3].

…

-- Carlos A Felgueiras Computer Science PhD SPRING GIS Software Manager Image Processing Division - DPI Brazilian National Institute for Space Research - INPE Fone: (12)3208-6469 or (12)981449544 Links: ------ [1] https://docs.scipy.org/doc/scipy/reference/generated/scipy.spatial.KDTree.html#scipy.spatial.KDTree [2] #134?email_source=notifications&email_token=ALDKQOEYLNNBAPAJXYVS43TRAA3HZA5CNFSM4KGG4ALKYY3PNVWWK3TUL52HS4DFVREXG43VMVBW63LNMVXHJKTDN5WW2ZLOORPWSZGOEKDM3GI#issuecomment-579259801 [3] https://github.com/notifications/unsubscribe-auth/ALDKQOGK7FYMTWQPL34YH4TRAA3HZANCNFSM4KGG4ALA

[MuellerSeb]

The number of the closest points is fixed and not determined by a search radius. If you have a dense data set, the n closest points would be rather near.

This needs to be improved in the future.

[MuellerSeb]

#143 addresses the missing search radius.

Closing for now due to inactivity. Feel free to re-open.