SCANNING THE GLOBE WITH REMOTE SENSING
We may be living in the information age, but ecological modelers continue to confront a scarcity of appropriate data for driving and validating their theoretical models. For scientists Steve Prince, Sam Goward, Scott Goetz, and Kevin Czajkowski of the Department of Geography at the University of Maryland in Col-lege Park, part of the solution lies in developing a model that can exploit an existing data set.
The group is seeking to estimate a parameter central to forecasting trends in climate change-global net primary productivity. This is the rate at which the biosphere assimilates atmospheric carbon through photosynthesis minus the rate at which plants release carbon through respiration. The mother lode of data used to drive the group's Global Produc-tion Efficiency Model, or GLO-PEM, lies in the optical and thermal remote-sensing measurements that have been continuously collected by weather satellites for more than 15 years. Traditionally, scientists have derived primary productivity from observations on the ground, through estimates of rates of biomass production, or through gas-exchange measurements. In the 1980s, however, ecologists discovered that weather satellites were fortuitously collecting data in the spectral regions relevant to monitoring the productivity of Earth's vegetation. The AVHRR (Advanced Very High Resolution Radiometer) sensor of the National Oceanic and Atmospheric Agency's polar-orbiting satellite, for example, has five channels, ranging from the visible to the thermal infra-red spectral regions. This range allows researchers to estimate key pro-ductivity parameters, including surface temperature and the absorption of photosynthetically active ra-diation. Moreover, Goetz says, "you're not extrapolating between point values. You actually have a measurement for every area on a global scale."
Given the wide array of physical and biological factors that influence photosynthesis and plant respiration, can a model driven solely by satellite data adequately estimate global pri-mary productivity? "Yes," Goetz said as he described GLO-PEM at the annual meeting of the International Society for Ecological Modelling.
The model starts with the amount of absorbed photosynthetically active radiation, a value derived from the interplay between the near-infrared and red spectral reflectance. The modelers convert this absorbed radiation into biomass production by factoring in variables such as air temperature, soil moisture, vapor pressure deficit, and standing aboveground biomass. The values of these variables are likewise derived from the remote-sensor measurements of near-infrared and red reflectance and of radiometric surface temperature.
A unique feature of GLO-PEM is that, in relying exclusively on remote-sensing data to drive the model, GLO-PEM does not require calibration with ground measurements of primary productivity. The relatively limited ground data base can therefore be used to validate model results, which is the current focus of the work of Goetz and his colleagues. At the same time, they hope to refine their model in preparation for more advanced measurements that will come from Earth observation satellites scheduled for launching over the next few years.
GLO-PEM is not without its competitors. At least 15 models for estimating global primary productivity, based on either remote sensing or ground measurements, appear in the scientific literature. A comparison of these models reveals considerable differences in both the regional patterns and magnitudes of the primary productivity estimates. What does this variability say about the global productivity modeling process? "I think the models that are purely statistically based are less reliable," Goetz says, adding that these models generally are calibrated for specific areas and then used at a global scale. He also believes, however, that estimates from mechanistically based models "are now starting to converge, giving a realistic range of results." -DS