Niel Bowerman (transfer report) - University of Oxford Department of ...

1 Project Description1.1 OverviewMany things in this world are uncertain. Much of science is devoted to assessing andreducing these uncertainties, and the field of climate prediction is no different. Thereare many types of uncertainty in climate prediction (Cox and Stephenson, 2007), andthese can be thought of in four broad categories:• Initial condition uncertainty arises from a lack of knowledge about the currentstate of the weather and climate. On shorter timescales, this is the dominantsource of uncertainty.• Process uncertainty arises from our inability to represent all of thecomponents of the process that drive climate in our models. This type ofuncertainty is the most difficult to quantify.• Parameter uncertainty is due to our lack of knowledge of the true values ofthe parameters in our models.• Scenario uncertainty arises because we cannot accurately predictanthropogenic emissions of greenhouse gases into the future. On longertimescales of the order of 100 years, this is the dominant source ofuncertainty.Much research is currently being done in the field of climate prediction, and most of itcan be described as quantifying or reducing one or more of these types ofuncertainty. The research that I am undertaking as part of my DPhil lies at theinterface between parameter and scenario uncertainty.Until recently, scenario uncertainty was the field of social scientists and IPCCWorking Group 3 (WG3). This started to change in 2009 with a series of papersintroducing the concept and features of cumulative carbon emissions (Allen et al.(2009), Meinshausen et al. (2009), Matthews et al. (2009), Zickfeld et al. (2009)).This concept allows an innovative method of simplifying scenario uncertainty, asthese papers show that when considering cumulative emissions, warming is pathindependent.Additionally, Allen et al. (2009) suggest that the uncertainty on warmingis also path independent for a given cumulative emissions total. I will be using theconcept of cumulative carbon emissions during my DPhil to explore scenariouncertainty from this new angle.1.2 ExperimentsOver the course of my DPhil I have been carrying out experiments exploringuncertainty, and I hope to do several more over the coming two years. I have giveneach experiment a code, which is listed in Table 1. The timetable that I hope tofollow over the coming two years is given in the Gantt chart in Table 2. Here I outlinethese experiments: past, present and future.1.2.1 N1: Simple Model ComparisonMuch of my initial work on cumulative carbon emissions has used simple climatemodels. I have employed simple models for their computational cheapness, and alsofor their ability to offer incites into the processes that govern the behaviour of theclimate system. I started my DPhil with a comparison between some simple climatemodels, which is documented in Section 2.3.2. This allowed me an insight intoprocess uncertainty, as it is likely that it was different representations of thePage 3 of 39