Identifying and quantifying sources of variability in temporal and spatial soil moisture observations

Abstract

[1] Soil moisture is an important component of the hydrological cycle. It is a control in the partitioning of energy and water related to evapotranspiration and runoff and thereby influences the hydrological response of an area. Characterizing the temporal and spatial distribution of soil moisture has important hydrologic applications, yet soil moisture varies in response to many processes acting over a variety of scales; the relative importance of different temporal and spatial controls on soil moisture is still poorly understood. In this paper we analyze both temporal and spatial soil moisture data empirically for two catchments in Australia and a further three in New Zealand. Hydrological conditions at these field sites covered a wide range over a 2 year period. The ground-based soil moisture data set is unique in its temporal and, in particular, its spatial coverage. Analyses attempt to isolate and quantify different deterministic sources of variability, measurement error, and a remaining unexplained component of variability. Because of limited data (especially relating to soils) we take a pragmatic approach of removing patterns that we can define in time and space (namely, seasonality and terrain) and then analyzing the unexplained variation. We then look for consistent patterns in this unexplained variability and argue that these are related to meteorological conditions, especially precipitation events, in the temporal case, and a combination of soils and vegetation in the spatial case. We were able to explain most of the observed variance in time and space, and the temporal variance was typically 5 times larger than spatial variance. Seasonality is the dominant source of temporal variability at our sites, although this conclusion obviously depends on climate and does not hold where soil water storage is limited. Most importantly, in controlling the distribution of soil moisture in space, the spatial distribution of soils and vegetation seems to be of similar importance to that of topography, a fact often ignored in hydrological modeling, or else surrogate soils patterns are used, but these are often not well correlated to the actual patterns [Grayson and Blöschl, 2000]. Better methods for defining the spatial properties of soils and vegetation as they affect soil moisture patterns are a key challenge.

Estimating spatial mean root-zone soil moisture from point-scale observations

“…This result is both surprising and contradicting: while the pattern of the wetness coefficient seems very similar to detailed observed soil moisture patterns at Tarrawarra, there is hardly any correlation between the individual NMM sites and the wetness coefficient. This is in line with the findings by Wilson et al (2004), who showed that even in catchments with significant topographic variability, the topographic component might not be the largest contributor to the overall spatial variance.…”

Spatial cross‐correlation patterns of European low, mean and high flows

“…This is in agreement with the common notion that soil moisture has a huge amount of spatial variability in the field [e.g. Famiglietti et al, 2008;Wilson et al, 2004;Daly and Porporato, 2005;Rodríguez-Iturbe et al, 2006;Western et al, 2002] and the observation that satellite based, continental scale, estimates of soil-moisture are highly correlated to the equivalent precipitation and evapotranspiration patterns [Liu et al, 2009].…”

Using lagged dependence to identify (de)coupled surface and subsurface soil moisture values

“…In assessing lagged dependence, event-scale patterns were of interest rather than large-scale trends within the time series (Wilson et al, 2004). This required seasonal patterns to be addressed prior to applying the DLNM.…”