A method to assess significant trends in time series of thematic forest maps

A contextual Mann-Kendall (CMK) test was applied to detect trends in time-series of 9-10 thematic forest maps from two decades at various size of units.

The Mann-Kendall test tests for a monotonic trend in the time-series and it is suitable for small sample sizes. The CMK test incorporates spatial autocorrelation, which is often present in remote sensing images and forests, to the test statistic. (Large scale) multiple testing problem arising from simultaneously testing thousands or more pixels from geographical data was controlled by using false discovery rate (FDR) adjustment of the p-values from the CMK test.

Since 1990’s forest resource maps and small area estimates have been produced by combining national forest inventory (NFI) field plot data, optical satellite images and numerical map data. Thematic maps of forest variables have been produced for eight to ten times depending on the geographical area. More information about the multi-source NFI can be find from the Luke website.

The study showed that significant trends can be detected from thematic forest maps produced employing relatively consistent materials and methodology, covering a long enough time period and with a sufficient number of time points.

An R-package ConMK is available at geoportti containing contextual Mann-Kendall-test and multiple testing correction option, among other features.

Figure: CMK test to ten MS-NFI mean volume thematic maps 1994-2017 covering Kanta-Häme and Päijät-Häme regions, 1200*1200 m2 units. Left: The standardized Z statistic, a positive Z-value indicates a positive trend; Right: p < 0.05 pixels (significant trend). Digital map data: contains data from the National Land Survey of Finland municipal division 1:4.5 M 01/2018.