Early well-coordinated response during unexpected catastrophes can define the near future of the stricken regions. Beirut city, Lebanon, was one of the unfortunate regions to endure the horrific ordeal of an unexpected explosion that caused thousands of human casualties, billions of dollars’ worth of property damage, and destroyed its main maritime entry point. In this paper, we identify damaged regions and classify their severity using a simple and robust SAR correlation technique. We employ phase coherence and amplitude correlation of a SAR stack to estimate pixels’ damage probability using hypothesis testing. We use a spatial phase filter applied in the frequency domain to improve the estimated coherence by removing the spatial decorrelation component of the total estimated coherence. Using this filter improved the coherence of nearly 44.2% of pixels identified with coherence less than 0.25 in our study area. The estimated damaged regions are presented and compared against a damage map issued by Advanced Rapid Imaging and Analysis (ARIA) which shows an average agreement of 68.3%. Also, a fine agreement was observed when compared to optical satellite images.

在意料之外的灾难中，早期协调良好的响应可以确定受灾地区的近期。黎巴嫩贝鲁特市是不幸的地区之一，遭受了一场意想不到的爆炸的可怕折磨，爆炸造成数千人伤亡，数十亿美元的财产损失，并破坏了其主要海上入口。在本文中，我们使用简单而强大的SAR相关技术来识别受损区域并对其严重程度进行分类。我们使用假设检验，使用SAR堆栈的相位相干性和幅度相关性来估计像素的损坏概率。我们使用了在频域中应用的空间相位滤波器，以通过去除总估计相干性的空间去相关分量来改善估计相干性。使用此滤波器可将相干性提高近44。在我们的研究区域中，有2％的像素被确定为相干小于0.25。呈现估计的损坏区域，并将其与由Advanced Rapid Imaging and Analysis（ARIA）发布的损坏图进行比较，后者显示平均一致性为68.3％。另外，与光学卫星图像相比，观察到很好的一致性。