Inter-Comparison of Land Surface Model Soil Moisture Data with Traditional Soil Dryness Indices

Australia has a long history of frequent forest fires, owing to its hot and dry climate. The McArthur Forest Fire Danger Index (FFDI; McArthur, 1967) was introduced in 1958 for operational fire warnings over Australia and is still used operationally, albeit with continuous development. The formulation of FFDI is based on air temperature, wind speed, relative humidity, and a component representing fuel availability called the Drought Factor (DF). The DF is defined on the assumption that the fuel moisture content (FMC) is affected by both long term and short term drying effects. The short term drying effects are based on the time since recent rain and past 20 days rainfall amount. The long term drying effects are based on either the Keetch-Byram Drought Index (KBDI; Keetch and Byram, 1968) or Mount's Soil Dryness Index (MSDI; Mount, 1972). KBDI and MSDI are estimates of the cumulative soil moisture deficit (SMD) and represent the degree of drought in the landscape. Studies show that the occurrence of large destructive fires corresponds to very large SMD values. SMD therefore is a key variable in the FFDI calculations with accurate estimates of soil moisture crucial for effective wildfire management, rating and warning. The KBDI is widely used in the Australian states of Victoria, New South Wales and Queensland while MSDI is used in the states of Tasmania, South Australia and Western Australia (Finkele et al., 2006). KBDI and MSDI are simple water balance models that do not take into account the majority of physical factors which affect soil moisture dynamics such as soil type, vegetation type, terrain or aspect. They over-simplify the evapotranspiration and runoff processes, which are critical in calculating accurate soil moisture states, leading to large errors. Recent progresses in the remote sensing of soil moisture, data assimilation techniques and physically based land surface models has led to the development of new soil moisture products. Two examples of such datasets are the soil moisture analyses produced from the Bureau of Meteorology's operational Numerical Weather Prediction (NWP) system and remotely sensed soil wetness measurements from the Advanced Scatterometer (ASCAT; Wagner et al., 2013) instrument. This study undertakes an evaluation of the latter two datasets along with KBDI, MSDI and another simple water balance model called the Antecedent Precipitation Index (API; Crow et al., 2005). In-situ observations of soil moisture from the OzNet hydrological monitoring network (Smith et al., 2012) and Australian national cosmic ray soil moisture monitoring facility (CosmOz; Hawdon et al., 2014) are used to validate the modelled and remotely sensed soil moisture datasets. The verification shows that the NWP soil moisture analyses have greater skill and smaller biases than the KBDI, MSDI and API analyses. This is despite the NWP system having a coarse horizontal resolution and not using observed precipitation. The average temporal correlations between observed CosmOz and modelled soil moisture are 0.81, 0.63, 0.76 and 0.73 for NWP, KBDI, MSDI and API. Verification also shows that the remotely sensed Advanced Scatterometer soil wetness product is of good quality. This study suggests that analyses of soil moisture can be greatly improved by using physically based land surface models, remote sensing measurements and data assimilation.