Tables S1 to S5: A data-driven approach to understanding the relations between geothermal exploration parameters: insights from Coso, Brady and Desert Peak, USA

Table S1. The K–S test result. The null hypothesis was rejected in most cases, showing that the distributions of datasets are primarily different. Table S2. The p-values of the coefficients and intercepts from the linear regression models of Fault Density Maps fit by Mineral Density Maps. Table S3. The p-values of the coefficients and intercepts from the linear regression models of Multiclass Temperature Maps fit by Fault Density Map and Mineral Density Maps. Table S4. The AIC values from the regression models of Fault Density Maps fit by Mineral Density Maps. Table S5. The AIC values from the regression models of Multiclass Temperature Maps fit by Fault Density Map and Mineral Density Maps.

补充表S1：柯尔莫哥洛夫-斯米尔诺夫（K–S）检验结果。绝大多数检验场景下均拒绝原假设，表明各数据集的分布整体存在显著差异。
补充表S2：以矿物密度图为拟合变量构建的断层密度图线性回归模型，其系数与截距的p值。
补充表S3：以断层密度图与矿物密度图为拟合变量构建的多分类温度图线性回归模型，其系数与截距的p值。
补充表S4：以矿物密度图为拟合变量构建的断层密度图回归模型的赤池信息准则（Akaike Information Criterion, AIC）值。
补充表S5：以断层密度图与矿物密度图为拟合变量构建的多分类温度图回归模型的赤池信息准则（AIC）值。