A multi-scale assessment of forest primary production across the eastern USA using Forest Inventory and Analysis (FIA) and MODIS data
MetadataShow full item record
As evidence of global warming continues to increase, being able to predict the relationship between forest growth rate and climate factors will be vital to maintain the sustainability and productivity of forests. Comprehensive analyses of forest primary production across the eastern US were conducted using remotely sensed MODIS and field-based FIA datasets. This dissertation primarily explored spatial patterns of gross and net carbon uptake in the eastern USA, and addressed three objectives. 1) Examine the use of pixel- and plot-scale screening variables to validate MODIS GPP predictions with Forest Inventory and Analysis (FIA) NPP measures. 2) Assess the net primary production (NPP) from MODIS and FIA at increasing levels of spatial aggregation using a hexagonal tiling system. 3) Assess the carbon use efficiency (CUE) calculated using a direct ratio of MODIS NPP to MODIS GPP and a standardized ratio of FIA NPP to MODIS GPP. The first objective was analyzed using total of 54,969 MODIS pixels and co-located FIA plots to validate MODIS GPP estimates. Eight SVs were used to test six hypotheses about the conditions under which MODIS GPP would be most strongly validated. SVs were assessed in terms of the tradeoff between improved relations and reduced number of samples. MODIS seasonal variation and FIA tree density were the two most efficient SVs followed by basic quality checks for each data set. The sequential application of SVs provided an efficient dataset of 17,090 co-located MODIS pixels and FIA plots, that raised the Pearson's correlation coefficient from 0.01 for the complete dataset of 54,969 plots to 0.48 for this screened subset of 17,090 plots. The second objective was addressed by aggregating data over increasing spatial extents so as to not lose plot- and pixel-level information. These data were then analyzed to determine the optimal scale with which to represent the spatial pattern of NPP. The results suggested an optimal scale of 390 km 2 . At that scale MODIS and FIA were most strongly correlated while maximizing the number of observation. The maps conveyed both local-scale spatial structure from FIA and broad-scale climatic trends from MODIS. The third objective examined whether carbon use efficiency (CUE) was constant or variable in relation to forest types, and to geographic and climatic variables. The results indicated that while CUEs exhibited unclear patterns by forest types, CUEs are variable to other environmental variables. CUEs are most strongly related to the climatic factors of precipitation followed by temperature. More complex and weaker relationships were found for the geographic factors of latitude and altitude, as they reflected a combination of phenomenological driving forces. The results of the three objectives will help us to identify factors that control carbon cycles and to quantify forest productivity. This will help improve our knowledge about how forest primary productivity may change in relation to ongoing climate change.