Ocean pattern analysis and weather forecasting

• Main Authors: Jishnu, N. M., Githin, Varghese, Thejus, T., Abraham Jisha, P., Pristy, Paul T.
• Format: Conference Proceeding
• Language: English
• Subjects: Atmosphere; Climate change; Climatology; Data processing; Modules; Numerical weather forecasting; Ocean currents; Ocean models; Oceans; Pattern analysis; Predictions; Regression models; Sea currents; Sea surface temperature; Ship terminals; Spatial data; Statistical analysis; Statistical methods; Statistical models; Technology assessment; Weather forecasting
• ISSN: 0094-243X; 1551-7616
• DOI: 10.1063/5.0227457

Weather forecasting is the operation of science and technology to prognosticate the conditions of the atmosphere for a given location and time. The weather can be predicted through application of the principles of physics, supported by a variety of statistical and empirical methods. The current Numerical Weather Prediction (NWP) models mostly depend upon atmospheric data. The current methodology overlooks the oscillations in the ocean. The oceans play a major role in defining the weather we experience on land. The sun heats the surface of Earth, resulting in warming the atmosphere. The heat is distributed from the warmer tropics to the poles with the help of ocean currents. The proposed model ensamples vector ocean and atmospheric data represented in NetCDF format that forecasts the pattern of trigger points resulting in drastic climate change. The global satellite data is obtained from two sources, ICODAS (International Comprehensive Ocean-Atmosphere Data Set) and GPCP(Global Positioning Climatology Project). The ICODAS data is collected from ships and gauge terminals. The former sources provide information like sea surface temperature (SST), scalar wind, sea level pressure (SLP), humidity and the latter one provides precipitation data. The model is divided into a data-processing module, forecasting module, trigger point estimation module. In the data processing module daily satellite data is taken and ocean and climate features are extracted and loaded into the spatial database. The forecasting methodology employed is the Vector Auto-Regression (VAR) model, which is a statistical model used to determine the relationship between multiple quantities as they vary over time. The model forecasts the short-term or long-term future values of the time-series data. Vector autoregression (VAR) is a stochastic procedure model used to find the linear interdependencies among various time arrangements. The Trigger point estimation module is based on the climate risk index factor calculated using the data from the spatial database. The proposed model outputs a graph with trigger points plotted in a given ocean region.