ABSTRACT
In early 2020s, NASA’s Earthdata open access portal has released a high-resolution SST data as a part of the NASA AWS Space Act Agreement. This paves the way for multidimensional transformative technology, wherein we can build upon it by using data analytic tools without seeking support for data acquisition. Recent developments in the domain of edge computing have made it possible to provide services overcoming the physical boundaries. Updates pertaining to disturbed marine lives and threats to the marine ecosystem are quite a common affair nowadays. Apart from other man-made factors, it is the sea surface temperature (SST) that plays a very significant role not only in marine life ecosystem monitoring but also as an early indicator of the overall health of the entire global ecosystem as well. However, the temporal and spatial dimensions of the data being immense, unlike physics-based numerical modelling methods that were used in the earlier days, a data-driven site-specific approach is preferred nowadays, thanks to the advancement of machine learning algorithms. The reconstructed Hadley Centre Sea Ice and Sea Surface Temperature data set (HadISST) obtained from National Center for Atmospheric Research (NCAR) across the Niño 1+2, Niño3, Niño3.4, and Niño 4 are used for analysis. These are calculated from the Kaplan SST, area averaged since 1870. The National Oceanic and Atmospheric Administration (NOAA)’s Earth System Research Laboratory (ESRL) provides such data to researchers and users. A comparison of forecasting analysis using MATLAB tool and Python is presented in the proposed work for prediction. In view of recent developments contributed by cloud services, the paper proposes a model that may make it possible for climate and marine ecosystem researchers to perceive the forecasted values and take decisive actions specially pertaining to small and large marine ecosystems that experience threats arising due to various oceanic perturbations. A set of error parameters strengthens the validity of the proposed prediction algorithms. Warning systems can be developed based on it.
