Surface soil moisture: getting the best estimates from models and satellites
New research shows that satellite and model-based methods to estimate near-surface soil moisture both have their relative strengths and weaknesses – but not always in the ways science has assumed.
Soil moisture plays a key role in the hydrological cycle and in governing water and energy exchanges between the land, vegetation and the atmosphere. Understanding how soil moisture varies in time and space is key for a number of scientific applications, including: drought monitoring, weather forecasting, national water accounting, bushfire danger assessment, and flood forecasting.
Soil moisture is currently estimated through a variety of methods, including ground-based measurements, satellite remotely sensed retrievals and modelled predictions. Over large areas, ground-based methods are not feasible, however. The OzEWEX Soil Water Estimation and Evaluation Project (SWEEP) brought method developers together with researchers measuring soil moisture in the field.
Led by Chiara Holgate, currently PhD student at Australian National University, the group of Australian and Dutch researchers developed a framework to determine how well different soil moisture satellite and model estimates reflect measured in-situ measurements in the top 5 to 10 cm of soil. In this study 11 data sets were evaluated to determine how they compared across Australia and its varied climate zones.
“With the large number of soil moisture data sets currently available across a range of satellite, model and in-situ platforms, we hope this study provides useful input into the selection of data sets for the many application areas utilising the information,” says Chiara Holgate.
The paper, published in Remote Sensing of Environment, found that some products correlated more strongly with in-situ measurements than others. However all products were more effective in detecting inter-annual and seasonal soil moisture dynamics rather than individual events.
Most satellite products all showed a similar level of association with in-situ measurements. At most sites, the two estimates derived from the Soil Moisture Ocean Salinity (SMOS) instrument better matched the in-situ measurements than those from the AMSR2 and ASCAT instruments.
Similarly, most models, adequately represented dynamics in soil moisture of the very top soil, with WaterDyn slightly outperforming CABLE, AWRA-L and other models. However the soil moisture models used in bushfire risk warning systems (KBDI, MSDI) agreed poorly with surface soil moisture and appear to be more representative of deeper soil moisture.
Clustering methods were used to determine which products were most similar to each other. This is useful for example when deciding whether to average between methods or to assimilate observations.
Data: The SWEEP data used in the study are publicly available from the OzEWEX project data page.
Article: ‘Comparison of remotely sensed and modelled soil moisture data sets across Australia’ by Holgate, C., De Jeu, R., van Dijk, A., Liu, Y., Renzullo, L., Vinodkumar, Dharssi, I., Parinussa, R., Van Der Schalie, R., Gevaert, A., Walker, J., McJannet, D., Cleverly, J., Haverd, V., Trudinger, C. and Briggs, P., Remote Sensing of Environment, doi: 10.1016/j.rse.2016.09.015
