If raster contains a single NaN Linear() works but Cubic() returns all NaNs

[Alex Gardner]

I'm seeing issues when using Cubic interpoaltion of data that contains NaN values. Here's what I see:

using Interpolations
A = rand(100,100);
x = 1:100;
y = 1:100;
interp = Cubic();

itp = Interpolations.extrapolate(Interpolations.scale(Interpolations.interpolate(A, BSpline(interp)), x, y), NaN);

itp has no NaN values

julia> sum(isnan.(itp.itp))
0

add a NaN value

 A[1,1] = NaN
itp = Interpolations.extrapolate(Interpolations.scale(Interpolations.interpolate(A, BSpline(interp)), x, y), NaN);

now all itp values are NaNs

julia> sum(isnan.(itp.itp))
10000

change interp to Linear

interp = Linear();
 itp = Interpolations.extrapolate(Interpolations.scale(Interpolations.interpolate(A, BSpline(interp)), x, y), NaN);

now itp contains only a single NaN

julia> sum(isnan.(itp.itp))
1

[Mark Kittisopikul]

What behavior would you expect if there are NaN values?

[Alex Gardner]

I would expect something similar to the behavior of when using Linear()... , if any point included within the footprint of the interpolation kernel (5x5 for Cubic) is a NaN then I would expect interpolations to return a NaN. I did not expect the existence of a single NaN in a 100x100 matrix to create NaNs everywhere