Spanning scales: The airborne spatial and temporal sampling design of the National Ecological Observatory Network. Issue 9 (28th July 2022)

• Record Type: Journal Article
• Authors: Musinsky, John; Goulden, Tristan; Wirth, Gregory; Leisso, Nathan; Krause, Keith; Haynes, Mitch; Chapman, Cameron
• Is Part Of: Methods in ecology and evolution. Volume 13:Issue 9(2022)
• Page Start: 1866
• Page End: 1884
• Publication Date: 2022-07-28
• Subjects: airborne remote sensing -- cloud prediction -- ecological sampling -- phenology; Ecology -- Periodicals; Evolution -- Periodicals; 577
• DOI: 10.1111/2041-210X.13942
• Languages: English
• ISSNs: 2041-210X

Abstract

Abstract: 1. Each year, the National Ecological Observatory Network's (NEON) Airborne Observation Platform (AOP) collects high‐resolution hyperspectral imagery, discrete and waveform lidar, and digital photography at a subset of 81 terrestrial and aquatic research sites throughout the United States. These open remote sensing data, together with NEON in situ sensor measurements and field observations, enable researchers to characterize ecological processes at multiple spatial and temporal scales. 2. Here we describe the sampling design for the AOP that aims to meet the diverse research needs of the ecological science community within the operational constraints affecting airborne data collection. Our spatial sampling protocol captures NEON instrumented systems, field plots and environmental gradients around each site while considering the context of airspace restrictions and remote sensing instrument capabilities. We use time series of moderate resolution imaging spectroradiometer (MODIS) satellite and PhenoCam near‐surface observations to define temporal sampling windows based on vegetation peak foliar greenness. We developed a probabilistic model based on MODIS reflectance imagery and Monte Carlo simulation to estimate sampling durations for cloud‐free data collection at each site. 3. Agreement in the estimated phenophase transition dates between MODIS Enhanced Vegetation Index and PhenoCam Green Chromatic Coordinate varied by vegetation class. Results from both sensors show that some vegetation classes have relatively consistent interannual peak greenness start‐ and end‐dates, while others experience high year‐to‐year variability in green‐up and senescence. In addition to phenological variability among sites, certain vegetation forms demonstrate distinct, asynchronous responses to climate, resulting in non‐overlapping peak greenness periods within a single site. Results from flight campaigns showed that the cloud‐likelihood model underestimated actual cloud conditions by 13%–26%, depending on the probability used. 4. Where interannual or intra‐site phenology is highly variable or clouds are a persistent problem, it becomes challenging to schedule domain deployments so that all sites are flown in cloud‐free conditions while their vegetation communities are in peak greenness. Despite limitations, application of cloud and peak greenness models to airborne sampling results in significant improvements to AOP data quality. Although most applicable to airborne sampling with hyperspectral and lidar instruments in piloted aircraft, these methods may be a valuable resource to deployment of Unmanned Aerial Vehicles for ecological research.