Changes of seagrass area in Beralas Pasir and Beralas Bakau Island observed from Sentinel-2 Satellite and verified by Unmanned Surface Vehicle (USV)

Abstract

Today, the area of seagrass ecosystems in Indonesia is estimated to have shrunk significantly. Bintan Island has quite a large seagrass ecosystems area. Along with the development of satellite technology, monitoring of conditions and changes to a coastal ecosystem can be carried out effectively through remote sensing technology. One satellite image that is relatively new and has good spatial quality is Sentinel-2 with a spatial resolution value of 10×10 m2 / pixel. Field data retrieval is facilitated by the use of Unmanned Surface Vehicle (USV). This research went through several stages such as image pre-processing, water column correction, masking, unsupervised classification, and detection of changes of seagrass area. The data obtained from the USV becomes the data for the accuracy-test in the supervised classification. Seagrass area was obtained in Beralas Pasir and Beralas Bakau Island is 84.27 ha (2016), 81.3 ha (2019) and 77.4 ha (2021). Detection of seagrass to non-seagrass area changes resulting 31.35 ha (2016-2019) and 30.91 ha (2019-2021). On the other hand non-seagrass to seagrass area is 24.84 ha (2016-2019) and 27.98 ha (2019-2021). The accuracy test of 2019 image classification and Unmanned Surface Vehicle data resulting overall accuracy at 62.20%.

Synergistic Use of Earth Observation Driven Techniques to Support the Implementation of Water Framework Directive in Europe: A Review

“…[5] integrated remotely sensed Chl-a into the WFD trophic state assessment for a lake case study, while Asper and Alikas (2019) [29] derived the ecological status of small lakes based on Chl-a for WFD reporting purposes in Estonia. Continuing in terms of the biological QEs, Elhag et al (2019) [30] utilized Sentinel-2 data to monitor Chl-a coupled with regression analysis in a lake case study while two studies by Zoffoli et al (2020) [31] and Yashira et al (2021) [32] used Sentinel-2 imagery data to provide high spatial resolution maps of seagrass at coastal zones; the first verified the results with USV data with an overall accuracy value of 62.2% and the second used one very high-resolution scene from WorldView-02 in order to evaluate the representativeness of the Sentinel-2 pixel size.…”

“…Special attention was given also to USV moving platforms as a water monitoring tool where it is noteworthy that four studies [32,49,74,75] used USV platforms with several sensors on board. It should be noted that 75% are custom-made solutions (see Table 2).…”

“…However, we should emphasize that research groups typically have legacy algorithms (e.g., deep learning) that are difficult to re-write and deploy on GEE or DC platforms, or researchers often find it difficult to deploy specific pre-processing routines (e.g., radiometric calibration, mosaicking). Hence, there are many software available for these needs such as the SNAP (Sentinel Application Platform) tool [110,111] and the ENVI image analysis software [32,54] that couple deep learning with all types of data including multi-spectral, hyperspectral, thermal, LiDAR and SAR and the PIX4D for UAV data processing [39,74,78] and the MEPHySto offering a hyperspectral pre-processed library [73].…”