Predicción de zonas susceptibles a incendios forestales aplicando técnicas de aprendizaje automático en el cantón Zapotillo, Loja – Ecuador
The effects of forest fires on ecosystems worldwide cause significant environmental damage. Ecuador has experienced this reality, as forest fires accounted for 38% of the natural and anthropogenic events that affected the country between 2010 and 2018. The Sierra region was the most affected by fire...
Guardat en:
| Autor principal: | |
|---|---|
| Format: | bachelorThesis |
| Idioma: | spa |
| Publicat: |
2024
|
| Matèries: | |
| Accés en línia: | https://dspace.unl.edu.ec/jspui/handle/123456789/30289 |
| Etiquetes: |
Afegir etiqueta
Sense etiquetes, Sigues el primer a etiquetar aquest registre!
|
Afegir etiqueta | Sumari: | The effects of forest fires on ecosystems worldwide cause significant environmental damage. Ecuador has experienced this reality, as forest fires accounted for 38% of the natural and anthropogenic events that affected the country between 2010 and 2018. The Sierra region was the most affected by fires, representing 80% of all fires at the national level. Among the provinces of this region, the province of Loja is one of the most affected, with 19% of the events registered. This event had a particular impact on the Zapotillo canton, due to its extensive areas of dry forest, which is home to a wide variety of flora and fauna. In response to this problem, the present research models the probability of the occurrence of forest fires in Zapotillo canton. Three supervised machine learning algorithms (RL, MARS and LMT) were applied to 164 points (fire and non-fire) corresponding to the period 2018 - 2021, which served to train (75 %) and validate (25 %) the models. Eight input variables associated with environmental, topographic and anthropogenic factors were used. The RL model obtained the best performance with an Area Under the Curve (AUC) of 0.65 (training set) and 0.63 (validation set). With this model, a map of probability of occurrence of forest fires (January - December 2020) was generated, which allowed automating the risk assessment through a spatio-temporal analysis, determining those urban areas are those with the highest probability of fires. |
|---|