Vegetation phenological events, such as bud break, flowering and leaf coloring, are closely associated with lower atmospheric conditions as seasons change. Plant phenology during springtime is considered to be particularly sensitive to climatic factors, especially temperature variations. Therefore, the occurrence of specific plant events can be used to identify the start of spring. Advance or delay in these timings through interannual comparisons can serve as potential climate change detection measures over long periods.

However, two issues may prove problematic in monitoring spring shifts between years by means of plant phenology. First, phenological responses to identical climate change can vary among different kinds of species. Previous research has shown that, even at the same location and study period, different species often display dissimilar trends in spring phenology. Secondly, phenology has been observed only intermittently in most places around the world. Insufficient data collection limits efforts to compare species events across large geographical areas and over long time periods, let alone the variations of spring onset derived from plant phenology. Although continental-scale simulation models have been developed based on selected species, this approach alone cannot account for broader species diversity. Recently developed satellite metrics, have shown potential to monitor the onset of spring across large areas. However, the relationship between species-level phenology and remote sensing imagery remains unclear.

My intent in this study is to determine the onset of spring in Wisconsin from the first bloom event of several introduced and native species, which vary from herbs to shrubs and trees. Due to the incompleteness of the surface observations, satellite imagery and modeled plant phenology are employed to help reconstruct the first bloom time series for different plants. The overall plan is to create several coherent regions in which first bloom timing for each species can be regarded as identical. Therefore, within each so-derived region, observational stations at different locations can be treated as one site. Average values from all intra-regional stations can be assigned to this "site" and thus form a combined time series for individual species in each region.

Missing values may still be inevitable in the newly reconstructed time series. I will examine interspecies correlations based on the combined data in order to fill the gaps region by region. After testing the statistical validity of reconstructed data, integrated species indices will be created in each phenological region according to species interrelations. Trend analysis and departure tendencies will be examined finally to derive regional trends or differences of spring's onset in Wisconsin measured by species first bloom indices over the 1965-1998 period.

My research is designed to contribute to current phenology-climate research in the following ways: 1) it represents a new approach to creating "species indices", which reflect correlations between comprehensive species responses and seasonal climate change; 2) the integrated species indices can potentially be used to monitor long-term springtime climate change at the regional scale; and 3) the techniques of phenological regionalization and similar species interpolation may assist reconstruction efforts with other incomplete phenological datasets.