干旱区生态环境知识资源中心

Southern California possesses a Mediterranean climate having semi-arid to arid characteristics and contains shrubland areas at high risk to wildfire. To assess wildfire danger, fire agencies have been monitoring the moisture of vegetation, called live fuel moisture (LFM), using field-based sampling. Unfortunately, spatial and temporal resolution of live fuel moisture data are significantly limited because sampling is labor intensive. Remote sensing satellite data has been used to monitor vegetation moisture content and health of shrublands. Therefore, a potential approach to overcome the limitations of manual measurements of live fuel moisture is to use vegetation indices (VIs) derived from satellite data. The objective of this study is to understand the link between vegetation indices derived from a Moderate Resolution Imaging Spectroradiometer (MODIS) aboard both Terra and Aqua satellites and in-situ live fuel moisture data. In this study, five vegetation indices were calculated using 6 bands of MODIS data within the visible and infrared spectrum collectively with the focus on the three best performing: enhanced vegetation index (EVI), normalized difference water index (NDWI), and visible atmospherically resistant index (VARI). Six sites with multi-year live fuel moisture data collection type were each represented with one pixel of MODIS data with a 500m by 500m spatial resolution covering the time period of February 2000 through December 2017 acquired aboard Terra and June 2002 through December 2017 acquired aboard Aqua. Simple linear regression was then applied to measure the coefficient of determination (R-2) between the vegetation indices and live fuel moisture data. The results show a great variance of R-2 between the sites as well as a variance of best performing VL The two strongest coefficients of determination, R-2=0.74 and R-2=0.72, were calculated at one site for enhanced vegetation index vs. live fuel moisture over a 15-year time period of data collected on Aqua and a 17-year time period of data collected on Terra respectively. The relationship was also affected by annual climate conditions including precipitation. Our preliminary results indicate that the satellite data reasonably well represents the live fuel moisture with higher temporal resolutions over a large area. Utilizing the remote sensing data in wildfire danger assessment will support fire agencies by saving resources for collecting ground data and providing better dataset in both time and space. This will also be beneficial for land management and planning, stakeholders and local governments.