基于Sobol'方法的释光热年代模型参数敏感性分析

地温随距地表深度减小而降低。地质体的剥露过程减小了地质体的埋深和温度，因此可通过地质体的温度变化历史重构其剥露过程。地质体中矿物释光信号积累受周围辐射剂量和环境温度的控制，因此可将矿物释光信号作为一种热年代计，通过测量矿物中积累的释光信号反演得到其热历史并推断相关地质过程。释光信号与时间、温度的关系由释光热年代模型所描述，模型中包含特征饱和剂量、环境剂量率、陷阱深度和频率因子等诸多参数，对矿物热历史的准确约束依赖于通过实验方法测定上述参数。实验测量的精度、测量方法的不确定性及样品本身特征对上述参数的测量带来了不确定性，影响了热年代和热历史重建的准确性。不同参数影响程度是否相同？哪些是更加关键的参数？从热年代实践的角度，需要我们获得对各参数影响的先验估计。然而，目前仍然缺乏相关的研究。为了定量评估不同参数的影响，本研究采用基于方差分解思想的Sobol'全局敏感性分析方法，使用现有文献报道的参数可能取值范围，对现有石英和钾长石的释光热年代模型进行参数的全局敏感性分析。考虑到参数敏感性可能受热历史、信号饱和水平影响，本研究比较了不同冷却速率(热历史)下不同冷却年龄(信号饱和水平)条件下的参数敏感性的差异，为有的放矢地对关键参数的测量方法进行优化约束提供了理论依据。

Sobol' sensitivity analysis of luminescence thermochronology model

Temperature decreases with the decrease of buried depth in the Earth's lithosphere. Exhumation process reduces the burial depth and temperature of geological body, so it is possible to reconstruct the exhumation process through the thermal history of the geological body. The accumulation of luminescence signals of minerals in geological body is controlled by ambient temperature, so the minerals can be used as a thermal chronometer, and the thermal history can be inversed by measuring the luminescence signals in the minerals. Luminescence thermochronology models contains many parameters such as characteristic saturation dose, environmental dose rate, trap depth, frequency factor and so on. The accurate constraint on thermal history depends on the measurement and the characteristics of samples bring uncertainty to the measurement of the above parameters, which affect the accuracy of thermal age. Do different parameters have the same effect?What are the more critical parameters?From the perspective of the practice of hot age, we need to have a prior estimate of the influence of each parameter. However, there is still a lack of relevant research at present. In this study, in order to quantitatively evaluate the effects of different parameters, Sobol' global sensitivity analysis method based on variance decomposition, was used to analyze the parameters of quartz and K-feldspar luminescence thermochronology models by using the value range of parameters reported in current references. Considering that the sensitivity of parameters may be affected by thermal history and signal saturation level, this study compared the difference of parameter sensitivity under different cooling rates(thermal history)and different cooling ages and different constant temperature conditions. The results show that: