Transport of nutrients from land to sea: Global modeling approaches and uncertainty analyses

Abstract

This thesis presents four examples of global models developed as part of the Integrated Model to Assess the Global Environment (IMAGE). They describe different components of global biogeochemical cycles of the nutrients nitrogen (N), phosphorus (P) and silicon (Si), with a focus on approaches to analyze model sensitivity and uncertainty. These models range from a lumped model at the scale of river basins to spatially explicit flow-path approaches, and are ultimately developed to analyze nutrient delivery to surface water, transport of and in-stream removal in the land-river continuum. The first example is a lumped global model to calculate the global annual river export of dissolved Si (DSi) to the ocean based on stepwise multiple regression. The model was cross-validated, and an extensive uncertainty analysis was made. Results show that the main factors determining river export of DSi include precipitation, soil bulk density, the occurrence of volcanic rock and relief. The importance of soil bulk density confirms the role of biological processes in the global Si cycle. Given the limitations of lumped regression models, a logical next step was toward a distributed and spatially explicit model. The soil nutrient budget is an important indicator for environmental losses of N and P, and is therefore the first step in the IMAGE distributed model. The N budget approach is illustrated with the uncertainty analysis of ammonia volatilization, a major term in the soil N budget. The animal N excretion rate is the largest determinant of uncertainty in the total agricultural NH3 emission in large parts of the world. However, in countries with industrial production systems, the contribution of animal houses and storage systems is more important. The third example is a model analysis of trends and uncertainty of the global terrestrial N removal process called denitrification (nitrate reduction to molecular N2 and the greenhouse gas N2O), which occurs in soils, groundwater, and riparian zones. Results show that denitrification has not kept pace with the increase in anthropogenic N sources (fertilizers and manure) and that riparian zones could be an important global N2O source. Sensitivity analysis showed that temperature, runoff and the soil N budget for natural ecosystems, cropland and grassland are important determinants for almost all global model variables. The fourth example is a global, spatially explicit distributed model to calculate changes in nutrient delivery to surface water, in-stream removal and export to the ocean during the 20th century. Results reveal that in the past century the increasing nutrient delivery by anthropogenic sources (expanding and intensifying agriculture, and sewage) and decreasing contributions from natural sources (deforestation) has led to increasing N and P export. The contribution of allochtonous organic matter inputs from vegetation in floodplains and wetlands to N and P delivery to surface water is probably important globally, but very uncertain and this source is among the major candidates for future research. The sensitivity of modeled delivery, retention and river export of N differs from that of P in many aspects.