Applying Smoothing Curves to AVHRR and MODIS Satellite Imagery to Detect the Onset of Spring in the Midwest and Northeast United States

Normalized Difference Vegetation Index (NDVI) values elicited from AVHRR and MODIS imagery have enabled the development of methods to identify start of spring (SOS) dates for multiple land cover types. This study evaluated the correspondence between field-based phenology observations and satellite-derived SOS measurements that utilize Fourier, sine square, and asymmetric Gaussian curves to smooth cloud-contaminated NDVI values. The relationships between the satellite-derived SOS dates and cloned lilac first bloom dates were measured using Pearson's correlation coefficient and regression analysis. The SOS dates were also compared to the average bud burst date of native species by computing the difference between the dates. The results indicate that the Fourier curve captures changes in vegetation during the beginning of the growing season and is superior to the sine square and asymmetric Gaussian curves. Moreover, the Fourier curve appears to be most effective in enhancing the seasonal trend of areas experiencing less vegetative change.