Developing an integrated surface-age remote sensing spectral property model to enable slip rate studies in inaccessible regions: Examples and challenges from the Mojave Desert

Slip rate studies often rely upon several in situ geochronological observations from mapped geomorphological surfaces. While this methodology is quite robust and works well in most cases, it can limit detailed studies in certain circumstances. Sites might be thoroughly observed with remote sensors such as LiDAR, SAR, and multispectral VNIR-SWIR, but lack physical accessibility due to factors such as unsafe conditions, political instability, and global events like pandemics. One such example of this is the Garlock fault in the Mojave Desert, California, where a large portion of eastern end of the fault is located on the artillery range of Fort Irwin with 50+ years of unexploded ordinance. Despite limited safe physical access, there is ample aerial LiDAR and satellite-based remote sensing products that could be used to establish a chronology on dated surfaces, and thus slip rates that could help answer questions about the Garlock faults role in North American tectonics. In order to address this and other similar problems, we are developing a model that will integrate the numerous remotely sensed properties for geomorphic surfaces of known age throughout the Mojave Desert region. While prior works have established links between surface age and specific sensed properties, it is possible to construct a more robust model using modern machine learning techniques. Ensemble learning models combine multiple simple models to reduce variance, bias, and increase accuracy, depending on the algorithms used. We have started assembling a database of published surface ages in the region and establishing relationships between various bands / sensors and surface age. Ultimately, hundreds of surface ages will be necessary to build a sufficiently robust model. After bootstrap analysis and cross-validation, the resulting model will be applied to surfaces of unknown age along the eastern Garlock fault to estimate slip rates in these inaccessible areas. Additionally, this model should also be useful to better contextualize solitary / sparse geochronology results from accessible field sites, and determine a more accurate minimum age.