How do I perform spatiotemporal interpolation of time series based using algorithms similar to isotropic weighting kernel function?

I have cross sectional time-series data in different locations (speed collected on vehicle detectors in different positions) I am going to reconstruct spatiotemporal states between these pointly-collected time-series so that the wave propagation patterns are preserved? I have already tried griddedinterpolant function with all its algorithm but it still does not provide the accuracy I'm looking for. Below is a figure I've obtained with griddedinterpolant. Is there any other spatiotemproal interpolation functions based on other algorithms such as isotropic weighting kernel function?