Panel Kriging
Panel Data Kriging · Also known as: longitudinal kriging, repeated-measures kriging, spatio-temporal panel kriging, panel geostatistical interpolation
Panel Kriging is a geostatistical interpolation method that combines kriging's spatial prediction framework with a panel (longitudinal) data structure. It estimates unknown values at unobserved locations and times by borrowing strength from repeated spatial observations across multiple time periods, accounting for both spatial dependence and temporal autocorrelation simultaneously.
Read the full method
Sign in with a free account to read this section.
Method map
The neighbourhood of related methods — select a node to explore.
When to use it
Use Panel Kriging when you have repeated spatial measurements at the same or overlapping sites across multiple time periods and you need to interpolate a continuous surface at unobserved locations or times. It is especially valuable when the single-period sample size is too small to estimate the variogram reliably but pooling across time provides enough pairs. Ideal applications include environmental monitoring networks, ecological surveys with seasonal data, and economic or demographic spatial panels. Avoid it when time periods are too few to identify temporal autocorrelation, when observations are spatially irregular to the point of near-zero overlap across periods, or when the spatial process is strongly non-stationary in ways the covariance model cannot capture.
Strengths & limitations
- Pools information across time periods, yielding more stable variogram estimates than a single cross-section.
- Produces optimal linear unbiased predictions (BLUP) of the spatial surface under the assumed covariance model.
- Quantifies prediction uncertainty at every interpolation target through the kriging variance.
- Accommodates both spatial and temporal autocorrelation in a single coherent framework.
- More efficient than running independent cross-sectional kriging at each time period.
- Requires an adequate number of time periods and spatial sites to estimate the spatio-temporal variogram reliably.
- Covariance model misspecification (e.g., assuming separability when the process is non-separable) can produce biased predictions.
- Computational cost grows with the product of the number of sites and time periods; large panels may require approximations.
- Stationarity assumptions (constant mean and covariance structure across space and time) are often violated in practice.
Frequently asked
How is Panel Kriging different from running kriging separately at each time period?
Running kriging independently at each period discards the temporal correlation and requires estimating the variogram from a single cross-section, which is often unreliable with small networks. Panel Kriging pools all periods jointly, producing more stable variogram estimates and allowing borrowing of strength across time for individual surface predictions.
What is a separable spatio-temporal covariance model?
A separable model factorises the joint spatio-temporal covariance into the product of a purely spatial covariance and a purely temporal covariance. It is computationally convenient but assumes that spatial and temporal dependences do not interact. Product-sum or fully non-separable models relax this assumption at the cost of more parameters.
Do I need observations at the same locations in every time period?
No — Panel Kriging can handle unbalanced panels where some sites are missing in certain periods, as long as there are enough observed pairs to estimate the spatio-temporal variogram. Missing observations at some sites or times simply reduce the information contributed by those panels rather than invalidating the analysis.
How should I validate a Panel Kriging model?
Use leave-one-out cross-validation or hold out a block of space-time observations. Compare predicted values to withheld observations using the mean squared error and the mean standardised squared error (which should be close to 1 if the kriging variances are calibrated correctly).
When is space-time kriging preferable to Panel Kriging?
The terms are often used interchangeably. When authors distinguish them, Panel Kriging emphasises the panel econometrics framing — repeated observations at fixed sites — whereas space-time kriging may refer to fully continuous space-time processes. In practice the underlying prediction equations are the same; the distinction is mainly in how the data structure and covariance model are motivated.
Sources
- Cressie, N. A. C. (1993). Statistics for Spatial Data (revised ed.). Wiley. ISBN: 978-0471002550
- Cressie, N., & Wikle, C. K. (2011). Statistics for Spatio-Temporal Data. Wiley. ISBN: 978-0471692744
How to cite this page
ScholarGate. (2026, June 3). Panel Data Kriging. ScholarGate. https://scholargate.app/en/spatial-analysis/panel-kriging
Which method?
Set this method beside its closest kin and read them side by side — the library lays the books on the table; the choice is yours.
- Geographically Weighted RegressionSpatial analysis↔ compare
- Ordinary KrigingSpatial analysis↔ compare
- Panel Spatial RegressionSpatial analysis↔ compare
- Space-Time KrigingSpatial analysis↔ compare
- Spatial AutocorrelationSpatial analysis↔ compare