Variable River Inputs over Time

Constraining River Fluxes over Geological Time

Experiments with varying the initial conditions and perturbing the nutrient models show that they exert a fairly robust regulation of the nutrient levels in the ocean. A further experiment is to try varying a "forcing" or "driving" input to the system, to see how well the system automatically compensates for variations in its forcing. In the case of the nutrient models this can be done by varying the amounts of nutrients flowing down rivers and into the sea. Unfortunately it is not easy to accurately reconstruct these fluxes from geological data. As nutrients flowed down ancient rivers they didn't leave behind little tell-tale indicators in sediments. Therefore we don't have any clear idea about how much phosphorus, nitrogen, silicate or carbon were flowing down rivers in earlier times. There are however less direct methods. Following an idea by Tim Lenton, we have used an alternative dataset to reconstruct river input. 

What goes in, must come out. The residence time of phosphorus in the oceans is order 50,000 years, and therefore, over timescales longer than this, we would expect any variations in phosphorus inputs to be balanced by corresponding changes in phosphorus outputs. Fortunately Karl Follmi, a Swiss geologist (see picture above), has compiled a dataset of phosphorus outputs from the ocean. From numerous measurements of the phosphorus contents in marine rocks through the ages and spread across different continents, he has obtained an estimate of how the global burial flux of phosphorus has changed over geological time. The image to the right shows Follmi's estimate of how P burial changed over time. We have used this record of P burial as an indication of P inputs.

Using the JModels to Examine the Effect of Time-Variable River Inputs

All of the JModels include the Follmi river inputs as one of their scenarios. The pictures below illustrate the steps to run this scenario in a model. Follmi's dataset extends over 100 million years. It is feasible to run the models for this length of time, but we have also shrunk the dataset into 10 million years to allow for faster run times. Select "Fast Follmi" to try these. 

 

External Links

Global Dimming