Vegetation dynamics using AVHRR/NDVI: Regional climate, carbon dioxide fertilization and crop yield relations

Vegetation development is closely related to climate factors, and, therefore, it is important to understand how it responds to global climate changes. For the last two decades it has been possible to monitor vegetation development at continental or global scales utilizing remote sensing Normalized Difference Vegetation Index (NDVI) data. We have developed a frequency analysis method to investigate land's vegetation greenness change and its response to the El Niño Southern Oscillation (ENSO). We found an ENSO influence on a tropical forest, southern semi-deciduous forest and a northeastern mixed forest. Our analysis shows the annual trends in vegetation greenness respond more sensitively than averaging methods. Atmospheric CO₂ increase is another concern for climate change, for which fertilization effect on land vegetation has been suggested. Atmospheric CO₂ and NDVI have a seasonal pattern of negative correlation, which makes it difficult to discern any positive influence of CO₂ on vegetation. We adopted the concept of the rate of change in atmospheric CO₂ concentration and NDVI to overcome this set pattern, and to reveal undergoing fluctuations. We found evidence that suggests a CO₂ fertilization effect in some arctic and sub arctic regions and northern and inland parts of the eastern humid temperate zones in North America. Although NDVI reveals the vegetation greenness only at a fixed time and location, we have transformed NDVI effectively to describe the vegetation growth dynamics in the form of a new index, Normalized Growth Index (NGI). Utilizing NGI, we found the vegetation growth during the growing season is highly negatively correlated with the initial minimum vegetation greenness. One needs to be careful when comparing Net Primary Production (NPP) using NDVI between different types of vegetation, because the same NDVI value can imply the existence of different biomass due to different Leaf Area Index (LAI). To overcome this difficulty we have developed Vegetation Anomaly Index (VAI), which is not influenced by vegetation type and is almost perfectly correlated with spatially averaged NDVI over any eco-region. Finally, we examined a possibility of utilizing NDVI to forecast crop yield and crop market price. We found that National Agricultural Statistics Service (MASS) corn yield estimate for Iowa and August NDVI averaged over the selected counties of Iowa are fairly well correlated for the past two decades. The Iowa corn market price is better correlated with NASS yield estimate than the average August NDVI over the counties; however, the correlation is more stable with NDVI than the NASS estimates, which indicates a great possibility of utilizing NDVI to forecast crop related access by USDA.