Exploring the seasonal dynamics of Australian temperate grasslands through phenocam imagery, remote sensing and field data

University of Technology Sydney. Faculty of Science.Management of temperate grasslands in south-eastern Australia is critical to support biodiversity conservation and agriculture under altered rainfall and warming conditions of future climates. Remote sensing is a common tool for monitoring vegetation but the dynamics of temperate grasslands present some unique challenges to conventional remote sensing methods. Land surface phenology—changes in large-scale vegetation dynamics—can improve the characterisation of temperate grasslands but the bulk of research in this field occurs in deciduous systems that show predictable vegetation changes. This research aims to explore drivers of grassland phenology and quantify the vegetation response by using field measurements, spectral instruments, time-lapse ‘phenocams’ and satellite data. A series of controlled experiments explored the fundamental expression of grassland characteristics. Spectral (NDVI₇₀₅ vegetation index) and biophysical (fractional cover) response of grassland mesocosms was investigated through manipulation of species, density and disturbance. C₃ and C₄ species showed distinctive phenology profiles, density treatments demonstrated a logical increase in NDVI₇₀₅ from low- to high-density, and recovery of grasslands from disturbance was quantified. Standing litter, a common feature of Australian grasslands, strongly suppressed reflectance-based vegetation indices. To investigate grassland response to changes in rainfall quantity and timing, phenocams collected sub-daily imagery from rainfall exclusion plots. Five treatments were assessed: ambient rainfall, increased rainfall, decreased rainfall, summer drought and extreme events. The Green Chromatic Coordinate index (gcc) showed dynamic response to rainfall in all treatments. Increasing quantities of rainfall resulted in significantly higher productivity throughout the year. Grassland productivity increased during cooler months from extreme events, but was equivalent to ambient rainfall during summer months. Summer drought unexpectedly drove higher gcc during non-drought periods, which was attributed to exotic forb invasions following disturbance. Field research was conducted on native and exotic C₃- and C₄-dominated grasslands in the South Eastern Highlands bioregion. Floristic surveys showed high variation in species richness and condition throughout the year, with highest detection of native species during summer months. Sites comprising multiple dominant species with overlapping phenophases showed a complex relationship with land surface phenology. Comparison of satellite NDVI (MODIS Terra, Landsat ETM+/OLI), phenocam and field variables showed that satellites and phenocams were equivalent at estimating green cover but the higher temporal capacity of phenocams allowed more precise definition of greening/browning trends. Dynamic knowledge of field conditions is essential for validating remote sensing phenology studies. This research develops a greater understanding of non-conventional phenology and provides practical tools to improve management of temperate grasslands