INVESTIGATING THE OPTIMAL TEMPORAL RESOLUTION OF SATELLITE DATA TO DETECT CLIMATE-INDUCED GRASSLAND PHENOLOGY CHANGES

Phenology is as an effective indicator of vegetation response to climate change and variability, and satellite remote sensing is a promising approach to monitoring large-scale phenology changes across grasslands and other terrestrial biomes. Most satellite data to detect land surface phenology are composited at coarse time intervals to minimize noise and cloud contamination. Trading detailed seasonal dynamics for higher data quality of satellite composite data may affect the accuracy of estimated phenology and its response to climate variability. However, not many studies have investigated this issue. The purpose of this research is to evaluate the effect of the temporal resolutions of satellite data on the estimated climate-induced changes in grassland phenology. The research was conducted in the Canadian prairie grasslands. Satellite vegetation indices (VIs) including AVHRR and MODIS VIs produced at a wide range of temporal resolutions were used to estimate the start of growing season (SOG), the end of greenness (EOG) and other phenological metrics and their response to climate variability. The near-surface PhenoCam phenology data were used to validate the satellite-based phenology due to the lack of in-situ grassland phenology measurement in the Canadian prairies. The results show that the prairie grassland SOG and EOG are sensitive to the variability of the preseason drought, precipitation and temperature. Also, the temporal resolutions of MODIS VIs were indicated to influence the quality of estimated phenological metrics and ability of MODIS VIs to depict the detailed grassland seasonal dynamics. Moreover, the results reveal the different response of prairie grassland SOG and EOG to climate variability measured by MODIS and PhenoCam and demonstrate the 10-day time interval is the optimal temporal resolution of MODIS EVI2 to predict the climate-induced changes in PhenoCam-based grassland phenology. This research addressed a critical gap in satellite-based phenology detection, investigating the uncertainties of land surface phenology and its response to climate variability related to the temporal resolution of input satellite data in the mixed prairie. This research also improves the understanding of the variability of biome functions in relation to climate variability