PROJECT SUMMARY
Interactions between the atmosphere and biosphere are a key component of the Earth’s physical system. Understanding this interaction is a crucial part of efforts to improve global change simulation models. While atmospheric data generally are available for most land areas, no such network exists to collect comparable information about biosphere activity. Recent satellite information can contribute to the development of global biospheric databases. In order to realize this potential, these remotely sensed data must be carefully calibrated with surface information. This approach is the most straightforward way of deducing the physical basis of, and developing a historical context for, the satellite’s observations. This study will take advantage of a unique surface vegetation data set to develop a physical interpretation of satellite data measuring the first appearance of spring foliage, commonly called the “green wave”, across eastern North America. The green wave is particularly important because it is one of the crucial biospheric variables necessary for accurate modeling of processes such as water balance and net primary productivity. The project will first develop empirical models to simulate surface green wave data (needed to supplement actual data limitations). Next, the relationship between satellite information and surface data will be determined for selected land cover types (biome). Lastly, detailed meteorological information from selected sites will produce a preliminary physical model of the green wave. These results will permit the satellite information to broaden existing surface green wave chronologies. In turn these lay the foundation for a future mid-latitude green wave model, which would assess the impact of climate variability on this aspect of the biosphere. Therefore, this project will serve as an example of how satellite bioclimatology in concert with surface phenology can produce physical interpretations of observed events, providing needed information for atmosphere-biosphere simulation models and global change monitoring.
RESEARCH FINDINGS
In the first phase of the project empirical multiple regression-based models were developed to simulate indicator (Lilac–Syringa chinensis ‘Red Rothomagensis’, Honeysuckle–Lonicera tatarica ‘Arnold Red’, and Honeysuckle–Lonicera korolkowii ‘Zabeli’) first leaf and first bloom events (needed to supplement limited actual phenological data). In subsequent phases, these models serve as bases of comparison between natural plant species and satellite-derived phenology, stratified by land cover type. An average date among the three indicator models, used to stabilize predictions, is termed the “spring index.” The first leaf spring index mean absolute prediction error was 5.5 days, four days less than error produced using average phenology dates from each site. Likewise, the first bloom spring index mean absolute error was 3.6 days, three days less than average phenology error. Bud burst phenology of fourteen natural tree species from a single Ohio, USA site were usefully predicted (model error less than average phenology error) by the first bloom spring index over the period 1883-1912.The objective of the second phase of the project was to document the utility of phenological data derived from satellite sensors by comparing them with modeled phenology. Surface phenological model outputs (spring index first leaf and first bloom dates) were correlated positively with satellite sensor-derived start of season (SOS) dates for 1991-1995, across the eastern United States. The relationship was highest for forest r .62 for deciduous trees and .64 for mixed woodland), and tall grass r .46), and lowest for short grass r .37). The correlation over all land cover types was .61. Average SOS dates were consistently earlier across all types than spring index dates. This finding and limited native tree phenology data suggest that the SOS technique is detecting understory green-up in the forest rather than overstory species. The biweekly temporal resolution of the satellite data placed an upper limit on prediction accuracy, thus, year-to-year variations at individual sites were typically small. Nevertheless, the correct biweek of satellite SOS could be identified from the surface models 61 percent of the time, and ±1 biweek 96 percent of the time. Further temporal refinement of the satellite measurements is necessary to connect them with surface phenology adequately and develop linkages among “green wave” components in selected biomes.Also in phase two, the surface spring index models were used to generate pseudo-onset dates from meteorological data back to the beginning of the 20th century, expanding and extending earlier efforts. Only last -2.2C frost date showed a significant trend (toward earlier dates) throughout the 1900-1995 period. Notably, however, first leaf dates did significantly move toward earlier values during the 1978-1990 period, in agreement with recent hemispheric-wide satellite studies of greenness onset. An independent satellite data study also confirmed this correlation between satellite-derived onset models and lilac phenology data. Thus, these spring indices can provide a historical context within which to assess recent satellite onset of greenness measurements in selected regions.The third and last phase of the project was concerned with assessing how well the spring index models pinpoint modifications in surface energy balance associated with the season’s onset. Well-defined changes were anticipated, as abrupt variations in meteorological data during this time have been documented in previous studies. Modeled flux data from twelve stations in the Oklahoma MESONET (energy balance model developed by Todd Crawford) over the 1996 and 1997 spring seasons were compared with values from the Lamont, OK ARM site (1994-1998), and the Harvard Forest, MA Ameriflux site (1991-1998). At all sites, daily average (daytime) latent heat flux began to rise rapidly after spring index first leaf date, and again after spring index first bloom date. Sensible heat flux leveled off and declined sharply after first bloom date. In addition, carbon dioxide flux at Harvard Forest turned negative and plummeted in connection with spring index first bloom date. Lastly, the average magnitudes of flux values were comparable at all sites when adjusted to the local “phenological” time scale (i.e., relative to spring index first leaf or first bloom date), despite the large difference in calendar date of the phenological events in Oklahoma and Massachusetts (roughly 50 days).In conclusion, this project has shown that continental-scale phenological models can be effectively linked and serve as intermediaries for satellite-derived metrics, native species data, and local changes in surface layer properties. These results further validate the feasibility and desirability of developing time lines that interconnect these various atmospheric and biospheric changes associated with the green wave phenomenon by land use-land cover type. Such connections will form the conceptual basis for improved future mid-latitude green wave modeling studies, and provide an effective means for monitoring one aspect of global change in mid-latitudes.