复垦后煤矸石山的复燃问题是矿区土地复垦与生态环境保护中的巨大挑战。超前和及时的自燃监测和预警对治理至关重要，同时也一直是研究和治理中的难题。研究基于无人机遥感技术，提出了一种在复垦后利用植被紫花苜蓿评估煤矸石山自燃风险的方法，并对该方法在自燃监测和预警中的可行性进行评估。以中国山西省某复垦后自燃煤矸石山为例，利用无人机搭载可见光、热红外相机获取了某复垦后煤矸石山的无人机影像，并基于无人机影像提取的图像特征估算了紫花苜蓿地上生物量、株高和植被水分信息；在此基础上提出了自燃风险评估方法，在已发生自燃的研究区A1开展预警可行性探究；同时，利用上述方法对一未知（潜在）自燃研究区A2的自燃风险进行评估。研究结果表明：（1）无人机是煤矸石山植被监测的有效工具，基于无人机遥感图像特征能够实现紫花苜蓿长势参数的精准估算。基于随机森林的紫花苜蓿生物量和植被水分估算模型在验证集上的R2分别为0.92和0.78，RMSE分别90.58 g/cm²和4.29%；基于作物高度模型的株高预测结果的R2为0.92，RMSE为7.58 cm。（2）三种紫花苜蓿长势参数均表现出了对自燃的解释能力，在空间分布上与25 cm深度土壤温度ST25cm均表现出一定的负相关性（R2= -0.43~ -0.51），且地上生物量的解释能力最优（R²= -0.51）。（3）基于紫花苜蓿生物量能够在一定程度上掌握地下自燃过程的影响范围、强度和变化方向，从而可能对煤矸石山的潜在自燃风险实现监测和预警。研究意为矿区复垦后煤矸石山的复燃防治工作提供新思路和方法支撑。

Spontaneous combustion of coal waste dumps is a huge challenge in land reclamation and ecological environment protection in mining areas. Advance and timely monitoring and early warning in spontaneous combustion process are crucial, and have always been a difficult issue in research and governance. Based on unmanned aerial vehicle (UAV) remote sensing technology, this study proposed a method for assessing the spontaneous combustion risk of coal waste dumps by using the reclaimed vegetation, alfalfa (Medicago sativa L.), and evaluated the feasibility of the method in potential spontaneous combustion monitoring and warning. Taking a coal waste dump after reclamation in Shanxi province, China, as an example, this study obtained the images of the coal waste dump by using an UAV equipped with visible and thermal infrared cameras. Then, the imagery features were extracted from the UAV images and used to estimate the alfalfa growth parameters, aboveground biomass (AGB), plant height (PH), and plant water content (PWC). On this basis, a spontaneous combustion risk assessment method was developed, and was applied to explore the feasibility in the study area A1 where spontaneous combustion had occurred. Then, the above method was used to assess the risk of study area A2, where the spontaneous combustion was unknown (or potential). The research results indicated that: ① UAV is an effective tool for vegetation monitoring in coal waste dumps, and alfalfa growth information can be accurately estimated based on the UAV remote sensing imagery features. The determination of coefficient (R2) of the alfalfa AGB and PH estimation model based on random forest (RF) was 0.92 and 0.78, respectively, and the root mean square error (RMSE) was 90.58 g/cm2 and 4.29%, respectively. The alfalfa PH estimation based on crop height model (CHM) resulted in an R2 of 0.92 and an RMSE of 7.58 cm. ② The three alfalfa growth parameters indicated the explanatory ability to the spontaneous combustion of coal waste dumps, which showed a certain negative correlation with the soil temperature at a depth of 25 cm (Ts,25) in spatial distribution (R2= −0.43−−0.51). Furthermore, alfalfa AGB showed the best performance (R2= −0.51). ③ The assessment result based on alfalfa AGB can grasp the scope, intensity and change direction of the underground spontaneous combustion process to some extent, so as to realize the monitoring and early warning of the potential spontaneous combustion risk of coal waste dump. Our research aimed at providing a new idea and the method support for the spontaneous combustion prevention of coal waste dumps after reclamation in mining areas.