The development of the Australian Phenology Product was funded by the AusCover Facility of the Australian Terrestrial Ecosystem Research Network (TERN) and supported by ARC-DP1115479 grant entitled "Integrating remote sensing, landscape flux measurements, and phenology to understand the impacts of climate change on Australian landscapes" (Huete, CI). Calculations were preformed on the University of Technology, Sydney eResearch high performance computing facility.
Land Surface Phenology (LSP) provides valuable information to scientists and practitioners interested in vegetation and ecosystem response to climate variability such as drought and land use change. Specific applications of LSP information include for example the quantification of crop yields, wildfire fuel accumulation, ecosystem resilience and health and land surface modelling.
R software, Savitzsky-Golay filtering and double-logistics curve modelling is used to generate the entire set of seasonal metrics from MOD13C1 EVI data cubes.
Seasonal/phenologic parameters - Validation of seasonal/phenologic parameters is accomplished across a wide range of Australian landscapes through comparisons with finer resolution (half-hourly/ daily) OzFlux eddy covariance tower measures of gross primary productivity (GPP) and evapotranspiration (ET)
Tower measurements - Tower measurements of actual photosynthetic activity (GPP) primarily relate to onset, duration, magnitude, and growing season length. Tower measures of ET potentially provide more accurate measures of vegetation green-up and duration in the arid and semiarid landscapes
Regional levels - At regional levels, we plan validation activities focused on the use of phenocam networks that capture the seasonal dynamics and species-level phenology of overstory and understory plant functional types. This is being prototyped in NSW and thenceforth to be expanded to TERN supersites, and transects. We are exploring the availability, prevalence, and cost-effectiveness of using CSIRO's CSIRO Wireless Sensor Network technology for 'light' sensors. The combined use of a PAR sensor (400-700nm) and total radiation sensor (0.4 to 3.5um) enables the calculation of Visible/ Near-infrared based broadband vegetation indices yielding in situ measures of seasonal vegetation profiles.
MOD13C1 QA flags - Less than “good data” MOD13C1 QA flags occur more frequently during the monsoon season at tropical latitudes near the coast and during the southern hemisphere winter at higher latitudes. The west coast of Tasmania and high elevation areas in South Eastern Australia also have more frequent cloud cover causing data gaps. The gap filling scheme used by the phenology algorithm overcomes the issue of missing observation in most cases except for areas with persistent cloud cover.
MOD13C1 QA flags - Less than “good data” MOD13C1 QA flags occur with very high frequency over highly reflecting surfaces (salt crusts) in the continent’s interior. The gap filling method and the subsequently derived phenological metrics should not be considered reliable.
Seasons in Australia - Seasons in Australia do only in certain regions ‘fit’ within a calendar year. The implication for data completeness is that part of a season that occurred in the beginning of 2000 or in late 2012 may not be captured by the derived phenological metrics as the season’s complete trajectory is not covered by the input data time series.
Input data - MOD13C2 product EVI 0.05-degree 16 day input data.