RELATIONSHIPS BETWEEN STABLE CARBON ISOTOPES AND SATELLITE DERIVED VEGETATION INDICES FOR USE IN PREDICTIVE VEGETATION DENSITY MODELS

Stable carbon isotope ratios vary predictably with environmental factors such as light and water availability within C₃ environments. Variability is greatest between wet, dense closed canopy forests, where δ¹³C values are markedly low, and dry open environments, where values are highest due to water-stress in plants. This variability creates a gradient in δ¹³C values that can be used to predict vegetation density.

Here we explore ways to quantify the relationship between δ¹³C values in vegetation and various vegetation parameters using satellite derived vegetation indices employed in modern remote sensing. These indices generally describe aspects of modern canopy cover, productivity, and climatic variables linked to the classification of different biomes. If a strong relationship exists between remote sensing data and regional δ¹³C values in C3 vegetation, this relationship could be used to interpret past environments with mean δ¹³C values from fossils, soil carbonates, bulk organics and other materials, potentially into deep time.

To quantify the relationships between δ¹³C values of modern C₃ dominated environments and the remotely-sensed vegetation indices, we used a large dataset of δ¹³C values from a variety of global C₃ environments. Data were collected from the literature and paired by location with several remote sensing products derived from NASA’s Moderate Resolution Imaging Spectroradiometer (MODIS). Several remote-sensing indices have strong linear relationships (r²≥0.65) with δ¹³C values. This indicates that the factors controlling the patterns in the satellite derived vegetation indices are similar to those that create the gradient in stable carbon isotopes and may be reliable across biomes. This approach has the potential to provide a way to broadly interpret and quantify mean δ¹³C values in terms of vegetation density and biome classification. Initial results show promising relationships between vegetation density, δ¹³C values, and remotely-sensed vegetation indices, but more work is necessary to develop and refine a predictive model.