Each year the National Ecological Observatory Network’s airborne observation platform (AOP) collects high-resolution hyperspectral imagery, discrete and waveform lidar, and high-resolution digital photography at a range of terrestrial and aquatic sites throughout the United States. These remote sensing data, together with other in situ data made freely available to the public, enable researchers to characterize vegetation structure, canopy biochemistry and ecological processes at local, regional and continental scales.
In an effort to minimize variability in spectral reflectance retrievals among data sets collected over multiple years, AOP flight campaigns are conducted during peak vegetation greenness, at solar elevation angles greater than 40 degrees and in atmospheric conditions where cloud cover represents less than 10% of sky coverage. For each NEON site, 16-year time series of 8-day composite enhanced vegetation indices (EVI) from MODIS were processed for each flight box, terrestrial sampling area and major vegetation type. Peak greenness windows for each class were determined by identifying end-of-spring (EOS) and start-of-fall (SOF) phenophase transition dates in the EVI time series. Phenocam-based EOS/SOF phenophase transition dates were then generated to validate EVI peak greenness windows in areas characterized by similar vegetation types. VIIRS EVI2 and Landsat 8 EVI phenological time series were also produced to test cross-sensor consistency with data sources that will likely extend beyond the lifetime of the MODIS sensors.
Finally, daily MODIS cloud cover percentages were derived for each flight box from a 16-year time series of MODIS surface reflectance data, from which mean monthly cloud-free fractions were calculated. Monte Carlo simulation was then used to determine the probabilities that flights of specific durations would result in cloud-free data acquisitions during the months of the year coinciding with each site’s peak greenness window.
Results/Conclusions
Inter-annual variability in the timing of EOS and SOF based on MODIS EVI is biome-specific, and within a given flight box there may be multiple asynchronous peak greenness periods for different vegetation types, which can make flight planning a challenge. Deciduous broadleaf forests and evergreen needle forests were relatively consistent year-to-year since green-up and senescence are typically linked to broad-scale seasonal changes in temperature and precipitation; grasslands and savannas were highly variable, with green-up principally driven by individual rainfall events. Agricultural sites were also variable due to cropping practices that track annual climate trends. Tundra showed the greatest inter-annual consistency, though this may have been due to the small sample size or length of time series employed.