Estimations of aerosol radiate forcing through new microphysical processes and regime-based constraints

气溶胶-云相互作用产生的间接辐射强迫存在很大的不确定性，这些不确定性使人们对未来气候模式的可靠性产生了一些怀疑。在这些模式中，有大量的参数不能与观测量直接相关，这些参数通常以特别的方式选择，以提供历史气候模拟与卫星观测的最佳匹配。为了约束这些参数，我希望采用两种方法。帮助缩小这些不确定性的第一种方法是根据每个参数对气候模式的辐射强迫输出的直接影响，创建一个参数层次。这可以通过灵敏度分析来实现，将关键的基于云的观测值（如液态水路径、云光学深度和云滴数浓度）的响应与各种模型参数的变化联系起来。然而，当在全球范围内进行全面分析时，这可能会产生虚假的结果。这是因为由于云周围和云内的大气条件不同，不同云系的响应可能不同。因此，建议从基于状态的研究开始，其中处理模型输出以将云分类为不同的状态。这可以用标准的方法来完成，通过云顶压力和云光学深度对云进行分类（结果大致类似于标准的云分类）。或者，可能存在其他分类方法（新颖的或其他的），可以提供重要的发现。然后，可以在逐个政权的基础上进行敏感性分析。该分析有望揭示哪些模式参数对净辐射强迫提供了最强的反馈，从而将重点放在这些有问题的变量上。除此之外，上述参数还可能对降水效率等其他过程产生强烈影响，而降水效率反过来又会对入境辐射产生净影响。可以采取的第二种方法是利用现有卫星数据对参数约束进行逆向工程。将使用的气候模式是气象局的HadGEM3模式，该模式随CFMIP观测模拟器包（COSP）一起提供。该软件包提供在模型中模拟气候条件下的模拟模型诊断，这些诊断将从各种卫星（如MODIS）中检索。通过将COSP的产出与过去20-30年的现有卫星数据进行比较，有可能确定任何较大的差异，并回溯找出这些差异的根本原因。虽然这个过程需要对COSP所包含的算法有深入的了解，但我相信在时间限制下开发这个过程是可能的。通过了解差异的主要原因，可以通过强调逆向工程认为最重要的参数，而不是使用试错方法，专门调整模型以匹配卫星观测结果。

Large uncertainties exist for the indirect radiative forcing produced by aerosol-cloud interactions, and these uncertainties cast some doubt on the reliability of future climate models. Within these models there are a large number of parameters which cannot be related directly to observed quantities, and these parameters are often chosen in an ad hoc manner to provide the best match of historical climate simulations with satellite observations. There are two approaches I hope to take in order to constrain these parameters. The first approach to help narrow down these uncertainties would be to create a hierarchy of parameters based on the direct impact of each parameter on the radiative forcing output of a climate model. This can be done through a sensitivity analysis, relating the response of key cloud-based observables (such as liquid water path, cloud optical depth, and cloud droplet number concentration), to changes in various model parameters. However, when performed as a blanket analysis on a global scale, this may produce spurious results. This is because responses may differ between cloud regimes, owing to different atmospheric conditions both surrounding and within the cloud. Therefore, it is proposed to begin with a regime-based study, where model output is processed to classify clouds into various regimes. This can be done in the standard way, by binning clouds by cloud-top-pressure and cloud optical depth (with resultant regimes roughly analogous to standard cloud classifications). Alternatively, there may exist other classification methods (novel or otherwise) that may provide significant findings. Afterwards, a sensitivity analysis can then be performed on a regime-by-regime basis. This analysis would hopefully reveal which model parameters provide the strongest feedbacks to net-radiative forcing, allowing focus to be directed towards these variables in question. On top of this, said parameters may also have strong effects on other processes such as precipitation efficiency, which can in turn have net effects on inbound radiation. The second approach that could be taken would be to reverse-engineer parameter constraints using available satellite data. The climate model that will be used is the MET office HadGEM3 model, which comes with the CFMIP Observational Simulator Package (COSP). This package provides simulated model diagnostics under the simulated climatic conditions in the model, as would be retrieved from various satellites such as MODIS. By comparing the output from COSP with available satellite data from the last 20-30 years, it may be possible to identify any large disparities and work backwards to find the root cause of these differences. Although this process requires an in-depth knowledge of the algorithms that COSP is comprised of, I believe it would be possible to develop this process under the time constraints. By understanding the main cause of the disparities, the model can be specifically tuned to match the satellite observations, by placing emphasis on the parameters deemed most important by the reverse-engineering, rather than using a trial and error approach.