Search

Forest fires cause temporary disappearance of vegetation. In the province of Azuay until October 2018, approximately 1,339.4 hectares of burned plant cover have been registered. In the present work, the degree of severity caused by a fire on the pine forest cover was estimated. This occurred on October 24, 2018 on the farm of the University of Cuenca, Irquis Azuay. Sentinel 2 satellite images (pre and post fire) were used, 55 ROIs were taken (30 training zones and 25 validation zones). The classification methods, Maximum Likelihood (ML) and Support Vector Machines (SVM) were used, but only the second one was used because it presented better results in its confusion matrix and Kappa index. Three indices were calculated to estimate the loss of vegetation cover; the normalized differential vegetation index (NDVI), the degree of severity index (NBR) and its differential version (dNBR). The NDVI post fire established that 91,87% of the surface contained sparse vegetation that was not vigorous; the rest was bare soil or dead vegetation, reducing the types of classification that existed before the fire. The pre and post fire NBR was performed to obtain its dNBR difference, it presented 4 levels of severity; 66,07% of the surface was intact (unburned), 22,93% burns of low severity, 10,07% low moderate severity and 0,93% moderate high severity. The dNBR results were visually contrasted with an orthophoto obtained with a drone by the university's geometrics department, reaching 4,9 ha burned compared to 7,9 ha obtained by the other method.

Forest fires cause temporary disappearance of vegetation. In the province of Azuay until October 2018, approximately 1,339.4 hectares of burned plant cover have been registered. In the present work, the degree of severity caused by a fire on the pine forest cover was estimated. This occurred on October 24, 2018 on the farm of the University of Cuenca, Irquis Azuay. Sentinel 2 satellite images (pre and post fire) were used, 55 ROIs were taken (30 training zones and 25 validation zones). The classification methods, Maximum Likelihood (ML) and Support Vector Machines (SVM) were used, but only the second one was used because it presented better results in its confusion matrix and Kappa index. Three indices were calculated to estimate the loss of vegetation cover; the normalized differential vegetation index (NDVI), the degree of severity index (NBR) and its differential version (dNBR). The NDVI post fire established that 91,87% of the surface contained sparse vegetation that was not vigorous; the rest was bare soil or dead vegetation, reducing the types of classification that existed before the fire. The pre and post fire NBR was performed to obtain its dNBR difference, it presented 4 levels of severity; 66,07% of the surface was intact (unburned), 22,93% burns of low severity, 10,07% low moderate severity and 0,93% moderate high severity. The dNBR results were visually contrasted with an orthophoto obtained with a drone by the university's geometrics department, reaching 4,9 ha burned compared to 7,9 ha obtained by the other method.

Forest fires cause temporary disappearance of vegetation. In the province of Azuay until October 2018, approximately 1,339.4 hectares of burned plant cover have been registered. In the present work, the degree of severity caused by a fire on the pine forest cover was estimated. This occurred on October 24, 2018 on the farm of the University of Cuenca, Irquis Azuay. Sentinel 2 satellite images (pre and post fire) were used, 55 ROIs were taken (30 training zones and 25 validation zones). The classification methods, Maximum Likelihood (ML) and Support Vector Machines (SVM) were used, but only the second one was used because it presented better results in its confusion matrix and Kappa index. Three indices were calculated to estimate the loss of vegetation cover; the normalized differential vegetation index (NDVI), the degree of severity index (NBR) and its differential version (dNBR). The NDVI post fire established that 91,87% of the surface contained sparse vegetation that was not vigorous; the rest was bare soil or dead vegetation, reducing the types of classification that existed before the fire. The pre and post fire NBR was performed to obtain its dNBR difference, it presented 4 levels of severity; 66,07% of the surface was intact (unburned), 22,93% burns of low severity, 10,07% low moderate severity and 0,93% moderate high severity. The dNBR results were visually contrasted with an orthophoto obtained with a drone by the university's geometrics department, reaching 4,9 ha burned compared to 7,9 ha obtained by the other method.