Extrapolating canopy phenology information using Sentinel-2 data and the Google Earth Engine platform to identify the optimal dates for remotely sensed image acquisition of semiarid mangroves.

Continuum monitoring of mangrove ecosystems is required to maintain and improve upon national mangrove conservation strategies. In particular, mangrove canopy assessments using remote sensing methods can be undertaken rapidly and, if freely available, optimize costs. Although such spaceborne data have been used for such purposes, their application to map mangroves at the species level has been limited by the capacity to provide continuous data. The objective of this study was to assess mangrove seasonal patterns using seven multispectral vegetation indices based on a Sentinel-2 (S2) time series (July 2018 to October 2019) to assess phenological trajectories of various semiarid mangrove classes in the Google Earth Engine platform using Fourier analysis for an area located in Western Mexico. The results indicate that the months from November through December and from May through July were critical in mangrove species discrimination using the EVI2, NDVI, and VARI series. The Random Forest classification accuracy for the S2 image was calculated at 79% during the optimal acquisition period (June 25, 2019), whereas only 55% accuracy was calculated for the non-optimal image acquired date (March 2, 2019). Although mangroves are considered evergreen forests, the phenological pattern of various mangrove canopies, based on these indices, were shown to be very similar to the surrounding land-based semiarid deciduous forest. Consequently, it is believed that the rainfall pattern is likely to be the key environmental factor driving mangrove phenology in this semiarid coastal system and thus the degree of success in mangrove remote sensing classification endeavors. Identifying the optimal dates when canopy spectral conditions are ideal in achieving mangrove species discrimination could be of utmost importance when purchasing more expensive very-high spatial resolution satellite images or collecting spatial data from UAVs.