Spatial and temporal validation of the MODIS LAI and FPAR products across a boreal forest wildfire chronosequence

Abstract

The leaf area index (LAI) and fraction of photosynthetically active radiation (FPAR) absorbed by the vegetation are key biophysical measures of canopy foliage area and light harvesting potential. Accurately quantifying these properties is important for characterizing the dynamics of mass and energy exchanges between vegetation and the atmosphere. The overall objective of this research was to validate the spatial and temporal performance of the MODIS LAI/FPAR products for a boreal forest landscape in northern Manitoba, Canada. We examined both the MODIS collection 4 (C4) and updated collection 5 (C5) versions over a multiyear period (2004 to 2006) and spatially across seven different-aged forests originating from wildfire and ranging in age from 1- to 154-years-old. We made optical measurements of LAI and FPAR, which were empirically scaled to high-resolution imagery (ASTER; 15–30m pixel size) to derive detailed reference LAI and FPAR maps. These maps were then aggregated to MODIS resolution (i.e. 1km) for comparison. We characterized the temporal accuracy of the MODIS products using repeat measurements of LAI and FPAR, and through comparisons with continuously operating measurements of canopy light interception.MODIS captured the general phenological trajectory of these aggrading forests, however the MODIS LAI and FPAR products overestimated and underestimated the LAI and FPAR for the youngest and oldest sites, respectively. In addition, MODIS displayed larger seasonal variation in LAI and FPAR compared to field measurements for the needle-leaf evergreen dominated sites. The peak growing season difference between MODIS and spatially aggregated ASTER reference values of LAI and FPAR decreased significantly by 69% and 55%, respectively, for the updated C5 versus the previous C4 LAI/FPAR products. The overall uncertainty (i.e. RMSE) in the MODIS LAI retrievals decreased from 1.6 (C4) to 0.63m2m−2 (C5) and from 0.20 (C4) to 0.07 (C5) for FPAR. The incorporation of understory vegetation into the validation of the MODIS products yielded significantly higher agreement between observed and MODIS values, likely due to the relatively open canopy architecture and abundant understory found within boreal forests.