Dynamic collocation between satellite and in-situ measurements in wave remote sensing

The validation and error analysis of remote sensing data are important for their application. Currently, due to the relative scarcity of in-situ observations in the ocean, the accumulation of collocations between remote sensing and in-situ data is slow. This study proposes an engineering trick to address this issue: using the output of numerical models as a "bridge" to connect remote sensing data with in-situ data, thereby expanding the spatiotemporal window of collocation and improving collocation efficiency. The basic idea of this method is that numerical models based on differential equations can partially simulate the local spatiotemporal variations of the parameter near in-situ data. These variations can be used to compensate for the representation errors caused by the spatiotemporal differences between remote sensing and in-situ observations. Therefore, this method can enlarge the spatiotemporal window for collocation when the error limit is given. This collocation process considers the dynamic processes through numerical models and is thus named "dynamic collocation." This study demonstrates through several simple experiments that this dynamic method is superior to traditional "static" windows and has application potential. In addition to improving data collocation efficiency in the open ocean, dynamic collocation can also address the issue of the difficulty of direct comparison between satellite and buoy data in coastal areas due to significant spatial gradients in sea waves. Besides data comparison, dynamic collocation can provide a larger sample size for the model training of narrow swath sensors' empirical algorithms. For example, we applied this method to SWIM data from CFOSAT and proposed a high-precision empirical retrieval algorithm for wave mean periods.