Article Preview
Top1. Introduction
Most of drought alert systems for forest fires prevention rely on vegetation water content (VWC) monitoring since this parameter plays a key role in the ignition and burning rate of woody or shrublands areas (Yi & Huili, 2010). In the Mediterranean area of southern France, the mean burnt area is about 20 000 ha per year (DGPM, 2012). Up to now, VWC monitoring is performed through a costly field survey network, requiring a periodic field work of a large number of agents followed immediately by a 24-hour laboratory data processing (Dauriac, 2004).
VWC is known to be sensitive to remote sensing infrared signal, especially shortwave infrared (SWIR) (Hunt & Rock, 1989; Ceccato et al., 2001; Chen et al., 2005; Ustin et al., 2008; Wang et al., 2008; Yilmaz et al., 2008). Yilmaz et al. (2008) recently succeed in the use of the Normalized Difference Infrared Index (NDII) from Landsat, ASTER and AWiFS images to remotely sense Equivalent Water Thickness (EWT) of corn leaves, which was found to be highly correlated to corn VWC.
With its high temporal resolution, the use of Terra-MODerate Resolution Imaging Spectroradiometer (Terra-MODIS) data has also been already explored for crop water content monitoring (Chen et al., 2005; Trombetti et al., 2008), for semi-arid ecosystems monitoring (Fensholt & Sandholt, 2003; Bajgiran et al., 2009) or for forest water content monitoring (Dasgupta, 2007; Liu et al., 2009; Qu & Ambrose, 2009). Walker et al. (2012) recently demonstrated the feasibility of a combined use of Terra-MODIS images (with high temporal resolution) and Landsat images (high spatial resolution) in order to monitor dryland forest phenology. In this study, forest dryland phenology was only measured through NDVI (Normalized Difference Vegetation Index) data and VWC was not considered. In addition, (Duveiller et al., 2011) pointed out the fact that modelling the instrument point spread as well as filtering MODIS data according to target and pixel overlaps, increase the accuracy in parameter estimations of crop phenomenology. However, the approach they proposed requires a precise spatial delineation of the objects to monitor, within MODIS pixels. This is hardly feasible in the case of Mediterranean shrubland VWC that corresponds to a continuous mixing of species. Most of these studies were developing using existing or newly defined near and mid-infrared bands ratio indices, among which can be cited: the Normalized Difference Water Index (NDWI) (Zarco-Tejada et al., 2003), the Simple Ratio Water Index (SRWI) (Zarco-Tejada et al., 2003), the Shortwave Infrared Water Stress Index (NDII) (Fensholt & Sandholt, 2003), the Vegetation Temperature Condition Index (VTCI) (Wan et al., 2004) and the Normalized Difference Infrared Index (NDII) (Hardisky et al., 1983).