1. Introduction CMIP5 simulations are generally characterized by persistent biases affecting the quality and fidelity of climate projections for the Southern Hemisphere. Examples of such problems include the simulation of sea ice, whose recent trends and even whose extent and annual cycle are inadequately captured by models. Clouds in the Southern Ocean are a further such problem, with a satellite-based climatology indicating substantially more cloud cover than reproduced by CMIP5 models. The ocean circulation also poses issues, with low-resolution CMIP5 ocean models misrepresenting Antarctic bottom water formation. In view of these problems, the New Zealand Government has launched the Deep South National Science Challenge, whose mission is to improve climate predictions through an improved understanding and modelling of Antarctic and Southern-Ocean processes. Six projects focussing on modelling and observations have recently been given final approval; these projects are introduced here. 2. Clouds and Aerosols (contact PI: ) The project pursues a three-pronged approach: In-situ and ground-based observations of cloud and aerosol properties in the Southern Ocean region will provide basic validation data, adding to a relatively small set of existing data. These measurements will be upscaled using satellite remote-sensing. Substantial effort will then go into using the in-situ and satellite information to improve the simulation of Southern-Ocean clouds and aerosols in the model. Activities thus far include a deployment of a ceilometer on a ship-borne voyage to Antarctica and some initial sensitivity studies using the NIWA-UKCA chemistry-climate model (figs. 1,2). The Deep South National Science Challenge: Reducing Persistent Climate Model Biases in the Southern H emisphere O. Morgenstern 1, S. Dean 1, D. Frame 2,1, M. Williams 1, G. Bodeker 3, A. McDonald 4, M. Bowen 5, P. Langhorne 6 1 NIWA, NZ 2 Victoria U., Wellington, NZ 3 Bodeker Scientific, Alexandra, NZ 4 U. Canterbury, Christchurch, NZ 5 U. Auckland, NZ 6 U. Otago, Dunedin, NZ Figure 1: Contour plot of the range corrected backscattered power observed by the laser ceilometer on 23 rd February Grey dots identify the height of the marine boundary layer associated with enhanced backscatter. Black dots identify cloud base. Figure 2: Cloud-radiative forcing bias at the top of the atmosphere (W/m 2 ) in the NIWA-UKCA CCM, relative to the CERES-EBAF climatology. 3. Observations and modelling of Antarctic sea ice ( ) This project focusses on two processes at the inner and outer sea ice edges that remain inadequately incorporated in climate models: At the inner edge, melting of ice shelves supplies cold, fresh water which promotes sea ice growth. Global warming is thought to strengthen this process. At the outer edge, ocean surface wave propagate into the sea ice zone, breaking up ice floes and promoting ice melt. New and existing observations will inform new formulations of these processes, which will be incorporated into the NZESM. Figure 3: Trend in sea ice fraction versus trend in wave height (Kohout et al., Nature, 509, p604–607,doi: /nature13262 ) 4. Decadal prediction and extreme events ( ) This project will produce initialized decadal climate predictions using the NZESM. Such forecasts are useful as such, but also provide an opportunity to assess and track model skill using a timescale of forecast which has only recently begun to be assessed in climate modelling. In addition, the project will use a simple climate model, ran on volunteers’ home computers, to generate a large ensemble of projections which can be usefully analysed for trends in weather extremes. The ensemble will be tied to the NZESM through the provision of initial and boundary data. 5. Assessing and validating the NZESM using modern and historic observations ( ) This project will focus on providing advanced datasets for model validation. It will process and use historic observations to allow model validation from the 19 th century, update and improve ozone datasets for model validation, both total column and vertically resolved (e.g. fig. 4), and use upper-air climate observations to validate the NZESM. New datasets produced under this project will be made available to the international community. Figure 4: The NIWA-Bodeker Scientifc TCO climatology. Example fields for 21 March (a) total column ozone field (Dobson units), (b) the uncertainties on each value plotted in (a) and, (c) the number of values averaged to create the means plotted in panel (a). 6. The Southern Ocean in a Warming World ( ) The distribution and transport of heat in the Southern Ocean is a key focus of this project. The project will assess changes in these processes in oceanic reanalyses, observations, and existing, low-resolution (CMIP5) and high-resolution ocean simulations. This work, along with validation of the NZESM in these regards, will inform new parameterizations for ocean processes (such as Antarctic bottom water formation) that are hitherto poorly represented in GCMs. This may involve increasing resolution in the NZESM around the Antarctic continental shelf to better represent this process. 7. Establishing a New Zealand Earth System Model ( ) The five projects introduced above all inform the development of the New Zealand Earth System Model. This model will be developed in collaboration with our lead overseas partner, the UK MetOffice. This project is about coordinating and leading the development of the model in collaboration with our overseas partners and other Deep South projects. The project will also produce century-scale simulations to be used in downscaling and impacts assessments. Subject to additional funding, we will also contribute simulations to CMIP6; details of our participation in CMIP6 are yet to be confirmed. 8. Summary The Deep South National Science Challenge is an ambitious project to overcome long- standing issues in climate modelling affecting available climate predictions for the Southern Hemisphere. By partnering with the UK MetOffice, we will focus on those aspects of the NZESM where we are best placed to contribute. Any substantial model progress and also new observational datasets will be shared with the international community. In a first for New Zealand, we hope to contribute simulations to CMIP6.