Predicting the Future of Kelp Forest Ecosystems in Southern California

Abstract

Climate change has caused a continued increase in the frequency and intensity of heat waves. This temperature increase is expected to cause a decline in kelp forest ecosystems across California. Kelp forests are needed economically by industries such as fisheries and a loss of these ecosystems would devastate California’s economy and communities. To determine the future of California’s kelp forests, we aimed to create a model based on existing data in this ecosystem. I acquired data on bottom sea temperatures from the Santa Barbara Coastal Long-term Ecological Research program and fish species abundance from the National Park Service (NPS). Once I accumulated and organized the data, I used Amazon Forecast's AutoPredictor to predict the future of California’s kelp forests between the years of January 1st, 2017, to January 1st, 2020 for the temperature data and September 1st, 2003 to June 1st, 2005 for the fish species data. I then evaluated the accuracy of the model and cross-validated it with existing data for the time periods. From this, I found that the model’s 10-percentile data, 50-percentile data, and 90-percentile data had accurate forecasts for the lower, middle, and upper parts of the data respectively. However, the model couldn’t predict an accurate and consistent amount of a species of fish due to either the non-daily data or due to the acquisition of data that had many extraneous variables that the algorithm couldn’t predict. From this it can be concluded that the algorithm works best for forecasting the parts of a kelp forest that have large amounts of consistent data and that the model struggles with observational data with extraneous factors.