A new perspective on the spatio-temporal variability of soil moisture: temporal dynamics versus time-invariant contributions

摘要

Knowledge about the spatio-temporal variability of soil moisture is essentialto understand and predict processes in climate science and hydrology. Asignificant body of literature exists on the characterization of the spatialvariability and the rank stability (also called temporal stability) ofabsolute soil moisture. Yet previous studies were generally based onshort-term measurement campaigns and did not distinguish the respectivecontributions of time-varying and time-invariant components to thesequantities. In this study, we investigate this issue using measurements from14 grassland sites of the SwissSMEX soil moisture network (spatial extent ofapprox. 150 × 210 km) over the time period May 2010 to July 2011.We thereby decompose the spatial variance of absolute soil moisture over timein contributions from the spatial variance of the mean soil moisture at allsites (which is time-invariant), and components that vary over time and arerelated to soil moisture dynamics. These include the spatial variance of thetemporal soil moisture anomalies at all sites and the covariance between thesite anomalies to the spatial mean at a given time step and those for thetemporal mean values. The analysis demonstrates that the time-invariant termcontributes 50–160% (on average 94%) of the spatial soil moisture varianceat any point in time, while the covariance term generally contributesnegatively to the spatial variance. On the other hand, the spatial variance ofthe temporal anomalies, which is overall most relevant for climate andhydrological applications because it is related to soil moisture dynamics, isrelatively limited and constitutes at most 2–30% (on average 9%) of thetotal variance. Nonetheless, this term is not negligible compared to thetemporal anomalies of the spatial mean. These results suggest that a largefraction of the spatial variability of soil moisture assessed from short-termcampaign may be time-invariant if other regions present a similar behavior.Moreover, we find that the rank (or temporal) stability concept, whenapplied to absolute soil moisture at the sites, mostly characterizes thetime-invariant patterns. Indeed, sites that best represent the mean soilmoisture dynamics of the network are not the same as those that best reflectmean soil moisture at any point in time. Overall, this study shows thatconclusions derived from the analysis of the spatio-temporal variability ofabsolute soil moisture need not generally apply to temporal soil moistureanomalies, and hence to soil moisture dynamics.