To determine which component corresponds to CME, we followed the criteria of (13). A component is considered to represent CME if >50% of stations exhibit a significant normalized spatial response (>25% of the maximum) and the eigenvalue exceeds 1% of total variance. In addition, we inspected the temporal pattern of each component to assess whether it exhibits changes corresponding to the timing of Hurricane Harvey, and estimated its spectrum, where previous studies have found CME to follow a flicker noise process (slope of −1) (15).

To determine CME from the GPS data, we initially applied ICA to the entire network (fig. S17). However, we found that this absorbed some of the real hydrologic signal (fig. S2, A and B). Instead, we estimated CME from a set of reference stations (n = 21) located in the northwest of the network, which are close enough to stations affected by Harvey but distal from regions of known precipitation (fig. S1) (20). Therefore, this approach is equivalent to most geodetic studies that take GPS positions relative to a group of reference stations distal to some target source (11).

CME was identified on IC1 (fig. S1) following the criteria described above, and which we found followed flicker noise (fig. S2C), consistent with a previous study of 259 GPS data in China (15). After this, we removed it from the rest of the network by simply multiplying the median spatial response of the 21 reference stations with IC1 and subtracting it from the time series of each GPS station. We found that, when applying ICA again, there was no remnant “CME component” found, indicating that it had been successfully removed (fig. S3).

For the horizontal component of the GPS motion, we followed the same approach as the vertical component, where we estimated CME from the same subset of stations and removed this from the data. We then reapplied the ICA analysis to extract the hydrologic loading signal. Similar to the vertical component of motion, we found two ICs that exhibited spatiotemporal changes consistent with the timing and spatial pattern of Harvey. For both the east and north components of the GPS motion, we interpreted the third and fourth ICs as reflecting hydrological loading (figs. S4 and S5), with clear deviations in time that corresponded to both landfalls of Harvey and spatial responses centered around Houston and west Louisiana.

Note: The content above has been extracted from a research article, so it may not display correctly.

Please log in to submit your questions online.
Your question will be posted on the Bio-101 website. We will send your questions to the authors of this protocol and Bio-protocol community members who are experienced with this method. you will be informed using the email address associated with your Bio-protocol account.

We use cookies on this site to enhance your user experience. By using our website, you are agreeing to allow the storage of cookies on your computer.