Published works

Live fuel moisture content is a key factor that determines the flammability of vegetation in ecosystems. Prediction of live fuel moisture content is inherently a very difficult problem since it is modulated by the complex physiological, phenological and ecological processes characteristic of the plant species. Soil moisture is one of the key variables that is known to influence plant water use. Recently, a new live fuel moisture content near-real-time product has been developed for Australia using a radiative transfer model inversion technique on the MODerate Resolution Imaging Spectroradiometer reflectance data. This live fuel moisture content product forms the basis of the Australian Flammability Monitoring System. At the same time, an advanced soil moisture analysis system has been developed by the Bureau of Meteorology recently, called the Joint United Kingdom Land Environment Simulator based Soil Moisture Information (JASMIN). JASMIN can estimate soil moisture at 5 km resolution on a daily timestep for the whole of Australia.

The present study brings together the above two products and explores the live fuel moisture content–soil moisture relationship on a national scale. This study will report the preliminary work carried out in understanding live fuel moisture content–soil moisture relationship and suggests an approach that may be constructive in advancing the ability to predict live fuel moisture content reliably to support fire management. A preliminary analysis is being conducted over 60 selected locations where JASMIN is found to have good skill. These 60 sites together represent a range of land cover types and climate zones typical of the Australian landscape. All the possible soil moisture profiles that can be derived from the four JASMIN soil layers are used for the analysis. Lag-correlation analysis shows that the strength of the relationship between live fuel moisture content and soil moisture varies from site to site and in general, is moderately strong (median lag-correlation of ~0.5). However, the strength of the relationship varies with vegetation type and also with soil profile depth. At all the sites, soil moisture is found to be an (important) leading indicator of live fuel moisture content. The lag also varies with the location and is found to range from days to months. Except for the forested sites, the top two soil layers exhibit a higher correlation with live fuel moisture content compared to the deeper layers. We develop a simple model to predict live fuel moisture content. The model is found to have good skill with an average R2 of 0.64 over the 60 sites. The normalized root mean square error for the model prediction is found to be less than 25% in general.