Tesis Doctorado
Permanent URI for this collection
Browse
Browsing Tesis Doctorado by Author "Armas Lorenzo, Elier"
Now showing 1 - 1 of 1
Results Per Page
Sort Options
Item Pronóstico de corto plazo de zonas de pesca de anchoveta (engraulis ringens) en el norte de Chile.(Universidad de Concepción, 2024) Armas Lorenzo, Elier; Neira Alarcón, SergioThis doctoral thesis is focused on short-term forecasting of potential fishing zones for E. ringens in northern Chile using a neural network model. The variables most associated with the spatial distribution of E. ringens were determined and utilized to train the neural network model. Additionally, the impact of forecasting E. ringens fishing zones on the efficiency of the northern Chilean anchovy fleet was studied. Engraulis ringens (anchovy) is a small pelagic fish species belonging to the Engraulidae family, inhabiting the neritic-coastal zone and ranging from northern Peru (3°N) to Chiloé Island in southern Chile (43°S). Its geographic and bathymetric distribution is influenced by fluctuations in oceanographic conditions at various temporal scales (daily, weekly, monthly, annual, supra-annual) and by fishing activities. In northern Chile, E. ringens is a key industrial fishing resource, representing up to 80% of annual landings by the purse seine fleet. The fishery's history (1985-2023) shows a significant decline in annual industrial landings, especially during extreme El Niño-Southern Oscillation (ENSO) events (1997–1998 (very intense), 2002–2003 (moderate), and 2015–2016 (very intense)). The most substantial declines in landings over the last two decades occurred in 2015 (El Niño) and 2020 (La Niña), decreasing by 65% and 88%, respectively, below the historical average. Given the decline in E. ringens landings in northern Chile during ENSO events, particularly during El Niño 2015 (65% below historical average) and La Niña 2020 (88% below historical average), fishing fleets must allocate more effort to locate E. ringens schools accessible to fishing gear. This has resulted in a decrease in catch relative to the fishing effort exerted on E. ringens during typical ENSO events. Catch per trip decreased between 9% to 33% during the 2006-2010 period, characterized by La Niña-like cold conditions, 12% during El Niño 2015, and 46% during La Niña 2020. Catch per fishing set also decreased between 5% to 13% during the 2006-2010 period and 40% during La Niña 2020. Catch relative to the distance traveled by industrial vessels declined between 13% to 40% during the 2006-2010 period, 53% during El Niño 2015, and 78% during La Niña 2020. This implies the need to increase effort (trips, sets, distance traveled) to maintain average catch levels. Currently, the primary challenge for the northern Chilean purse seine fleet is to find ways to reduce costs in locating and catching E. ringens, thereby increasing economic profitability. This necessitates optimizing/reducing fishing trips. An interesting alternative to explore is to identify potential fishing zones and to understand how the probability of E. ringens catch changes under oceanographic variability and during ENSO events, and to determine whether this probability is related to annual landings of E. ringens in the study area. In this context, this doctoral thesis is focused on studying the impact of using E. ringens fishing zone forecasts on the efficiency of the industrial fleet in northern Chilean. The hypotheses of this thesis indicates that if the spatial distribution of E. ringens in northern Chile is primarily determined by oceanographic variables, and these relationships are accurately captured by a neural network model, then the developed model will appropriately forecast E. ringens fishing zones in the short term, increasing the performance of the purse seine anchovy fleet. The research results are presented in three chapters briefly described below. In the first chapter, the spatial distribution of E. ringens in northern Chile (18°21” S−27° S) is analyzed, along with its association with available oceanographic variables for the study area for the period 2003 to 2020. Georeferenced catch data for this species, obtained from a database built by Corpesca S.A. Fishing Company, were used to approximate E. ringens distribution. Areas where the fleet transited without capturing E. ringens were considered absence records. Oceanographic variables were obtained from the Copernicus program's Nemo model, providing historical and daily forecast data. A neural network model was applied to address the relationship between oceanographic variables and the distribution of E. ringens, determining the variables that best explained the location of fishing zones for this species during the study period. In terms of the explained variance percentage, geographical longitude (23%) was the most relevant variable for identifying potential fishing zones, followed by the depth of the mixing layer (18%), geographical latitude (15%), sea surface temperature (12%), month (12%), sea level (9%), salinity (9%), and zonal and meridional components of current velocity (1% each). In the second chapter, a predictive model of fishing zones based on neural networks is implemented, trained with georeferenced daily catches from industrial purse seine vessels from 2003 to 2020, along with oceanographic variables (sea surface temperature, salinity, mixing layer depth, sea level, and currents), obtained from Copernicus (https://marine.copernicus.eu). The neural network model achieved 86% performance, and correctly classifying the most of areas with and without fishing. Therefore, its use is recommended for predicting fishing zones for E. ringens in the study area. In the third chapter, the thesis analyzes whether the current decline in annual E. ringens landings is associated with oceanographic changes in northern Chile during El Niño or La Niña events. The neural network model developed in the second chapter is applied to identify the spatial and temporal distribution of E. ringens fishing probabilities, particularly for El Niño (2015), La Niña (2007, 2013, 2020), and Neutral (2004) years. It was found that the probability of E. ringens catch extended further west during La Niña events (except in 2020), covering a larger area that is, however, limited to a coastal strip of 10 nautical miles during the El Niño 2015 event. The highest catch probabilities in Neutral conditions are near the coast, although not as restricted as during the El Niño 2015 event. The highest catch probabilities in the La Niña event of 2020 are near the coast, in contrast to the previous events of 2007 and 2013, due to the restriction of the optimal habitat of E. ringens by changes in oceanographic conditions. The annual quantity of potential fishing zones identified by the neural network model is an indicator of the environment's suitability for encountering E. ringens schools. The application of these results should enable better management of the industrial fishing fleet, reducing operational costs. The results of this thesis support the hypotheses, as the implemented neural network model successfully forecasted short-term fishing zones for E. ringens. The implemented neural network model was able to replicate the spatial distribution of E. ringens observed during ENSO events. Simulations showed that using the fishing forecasts from the neural network model in fleet operations would have resulted in a decrease in effort and an increase in fleet performance. The application of the results of this study will allow understanding, and likely anticipating, the consequences that extreme ENSO events could have on the performance of the northern Chilean anchovy purse seine fleet. The neural network model developed in this study provides a valuable tool for the management of this fleet.