论文标题

部分可观测时空混沌系统的无模型预测

The Importance of Co-located VLBI Intensive Stations and GNSS Receivers: A case study of the Maunakea VLBI and GNSS stations during the 2018 Hawai`i earthquake

论文作者

Dieck, Christopher, Johnson, Megan C., MacMillan, Daniel S.

论文摘要

储层计算是预测湍流的有力工具,其简单的架构具有处理大型系统的计算效率。然而,其实现通常需要完整的状态向量测量和系统非线性知识。我们使用非线性投影函数将系统测量扩展到高维空间,然后将其输入到储层中以获得预测。我们展示了这种储层计算网络在时空混沌系统上的应用,该系统模拟了湍流的若干特征。我们表明,使用径向基函数作为非线性投影器,即使只有部分观测并且不知道控制方程,也能稳健地捕捉复杂的系统非线性。最后,我们表明,当测量稀疏、不完整且带有噪声,甚至控制方程变得不准确时,我们的网络仍然可以产生相当准确的预测,从而为实际湍流系统的无模型预测铺平了道路。

Frequent, low-latency measurements of the Earth's rotation phase, UT1$-$UTC, critically support the current estimate and short-term prediction of this highly variable Earth Orientation Parameter (EOP). Very Long Baseline Interferometry (VLBI) Intensive sessions provide the required data. However, the Intensive UT1$-$UTC measurement accuracy depends on the accuracy of numerous models, including the VLBI station position. Intensives observed with the Maunakea (Mk) and Pie Town (Pt) stations of the Very Long Baseline Array (VLBA) illustrate how a geologic event (i.e., the $M_w$ 6.9 Hawai`i Earthquake of May 4th, 2018) can cause a station displacement and an associated offset in the values of UT1$-$UTC measured by that baseline, rendering the data from the series useless until it is corrected. Using the non-parametric Nadaraya-Watson estimator to smooth the measured UT1$-$UTC values before and after the earthquake, we calculate the offset in the measurement to be 75.7 $\pm$ 4.6 $μ$s. Analysis of the sensitivity of the Mk-Pt baseline's UT1$-$UTC measurement to station position changes shows that the measured offset is consistent with the 67.2 $\pm$ 5.9 $μ$s expected offset based on the 12.4 $\pm$ 0.6 mm total coseismic displacement of the Maunakea VLBA station determined from the displacement of the co-located global navigation satellite system (GNSS) station. GNSS station position information is known with a latency on the order of tens of hours, and thus can be used to correct the a priori position model of a co-located VLBI station such that it can continue to provide accurate measurements of the critical EOP UT1$-$UTC as part of Intensive sessions. The VLBI station position model would likely not be updated for several months. This contrast highlights the benefit of co-located GNSS and VLBI stations in support of the monitoring of UT1$-$UTC with single baseline Intensives. Abridged.

扫码加入交流群

加入微信交流群

微信交流群二维码

扫码加入学术交流群,获取更多资源