A changepoint approach to automated estimation of soil moisture drydown parameters from time series data.

Advances and the growing deployment of in-situ soil sensing has the potential to deliver new insights into soil system dynamics. However, it also calls for the development of efficient data analysis methods that can extract interpretable information from continuous data. This study utilises an automated, changepoint-based method for analysing soil moisture time series data. The method is used to autonomously detect wetting events and dynamically estimate parameters describing the drydown characteristics of the soil moisture following the event. Information can then be extracted from the output of the changepoint analysis. This provides an indication of how soils are responding to wetting events, and here we explore if this information corresponds with soil characteristics. In an illustration using soil moisture data from nine different field sites in the United States, different drydown characteristics were observed from the distributions of the estimated parameters. We find that these features can be associated to the climatic regimes and the soil texture of the sites. The potential for identifying changes in soil properties and processes based on shifts in drydown characteristics over time is discussed.