The exploitation of a geothermal field can be accompanied by both natural and induced seismicity. Hence the installation of a seismic network suitable for also locating low-magnitude earthquakes, is of great interest for geothermal development, especially for monitoring the activity related to the injection or production.

Most of the aspects that affect the location of earthquakes are data quality and phase picking, the inverse method chosen to solve the problem, the velocity model, and the seismic network configuration. We focus on the optimization of network configuration and the reduction of the location error is one of the alternatives to finding optimal station positions. Here we propose an improvement of the D-OPTIMAL algorithm (Tramelli et al., 2013) that tries and find optimal station positions minimizing the amount of the error ellipsoid of the event location using the D-criterion. This algorithm is computationally efficient and easy to be implemented. In this version, we also introduced the possibility to account for several prior information that is generally available when implementing a monitoring site permanently or temporarily. These a priori parameters introduced are: i) three-dimensional seismic velocity models, ii) seismic noise levels, iii) topographic gradient, and iv) H/V ratio values. The last three parameters are introduced in the station position selection using a weighted system scheme.

We applied this methodology to the Acoculco geothermal field (Mexico) where an injection test was planned and executed in 2021. The comparison between the network defined using the standard approach, which considers a 1D velocity model and the installation costs, and this improved version shows the importance of introducing a priori information during the selection of the network. Installation sites showed better distribution in the region, resulting in an overall increase in the sensitivity and a decreasing of the error location estimation in the target region.

The methodology presented here is easy to apply to other study cases such as active volcanoes, anthropogenic activities, or other kinds of location studies at local scale.

地热场的开发可能伴随着自然和诱发的地震活动。因此，安装适用于定位低震级地震的地震台网对于地热开发，尤其是监测与注入或生产相关的活动具有重要意义。

影响地震定位的大部分方面是数据质量和相位选择、解决问题所选择的反演方法、速度模型和地震台网配置。我们专注于优化网络配置，减少定位误差是寻找最佳站点位置的替代方法之一。在这里，我们提出了 D-OPTIMAL 算法的改进 (Tramelli et al., 2013)，该算法尝试使用 D 准则找到最佳站点位置，以最小化事件位置的误差椭球量。该算法计算效率高，易于实现。在这个版本中，我们还引入了考虑几个先前信息的可能性，这些信息在永久或临时实施监控站点时通常可用。地震噪声水平，iii) 地形梯度，以及 iv) H/V 比值。最后三个参数是在使用加权系统方案的站位选择中引入的。

我们将这种方法应用于 Acoculco 地热田（墨西哥），计划在 2021 年进行注入测试。使用标准方法定义的网络之间的比较考虑了一维速度模型和安装成本，这个改进的版本显示在选择网络时引入先验信息的重要性。安装地点在该区域内表现出更好的分布，导致灵敏度总体提高，目标区域的错误位置估计降低。

这里介绍的方法很容易应用于其他研究案例，例如活火山、人为活动或其他类型的局部规模的位置研究。