Detection of Forest Fires using Satellite Imagery and Machine Learning.
Monitoring and prediction proves to be seemingly important for prevention as it will help in threat identification and better problem solving. Therefore we came up with a user-friendly monitoring application. With this app users can monitor any particular green area in real time just by giving its coordinates. Using Google Earth Engine we are collecting a real time data set and it will be stored in IBM Cloud. IBM cloud is being used for its flexible interface and privacy will be maintained. Users having unique user id and password can monitor at any instance. Also an alarm will be there to alert in urgency.
The global climate crisis calls for an urgent action to combat the climatic change and its impacts.Forest exploitations and forest fires on large scale contribute to the degradation of not only land but also the climate extensively. Adhering to which, the principal intent of the project is to develop an online platform to get real-time information about the forests in the most affected areas of the country and avert any disastrous emergency.
Most forest fire control departments are informed of fires when it's already too late. For this we introduce RTMA (Real Time Mesoscale Analysis) Data based Machine Learning. So that it can be detected early and prevented easily. Supervising the entire area for forest fire prediction is really challenging. So our idea will allow users for real time monitoring and predicting. Every data will be available at the user's fingertip.
Forest covers in India have been under threat for the past few years and there can be numerous undeniable reasons for it. The impacts of reduction in forest covers cannot be ignored any further. Forest fires not only cause damage to the forest and its livelihood but also to the atmosphere releasing an abundance of harmful gases into the atmosphere which is harmful to human health. They ultimately lead to loss of property, crops, animals, resources and people.
Loss of vegetation further causes climate change, desertification, soil erosion, floods, increase in greenhouse gases in the atmosphere and also economic hardships to those small communities who are entirely dependent on the resources from forests.
To predict the threat at an early stage therefore we come up with a real-time monitoring web application through which we can monitor the nearby forests and identify any alarming situation.
This project is focused on a technical solution towards sustainability. Forestfire is one of the major concerns as it is getting worse day by day. The problem with forest fires is their uncontrolled spreading nature and their sheer sudden eruptability. Most forest fire control departments are informed of fires when it's already too late. For this we introduce an UI where the user will give user credentials with the latitude & longitude of the target area. We are sending these data in the form of API with SQL queries to Cloud. With the help of IBM Cloud these information will be stored safely. We have also created a ML model which is capable of doing instance segmentation. This model will segregate fire in three parts on the basis of intensity of the fire (eg.- HIGH, MODERATE, LOW). Also an alarm will be there to alert in urgency.
Forest fire can occur due to either human behavior or by nature itself. Man-made causes are reduced fairly in recent years but natural causes are quite challenging to predict beforehand. So our idea will help users to predict and prevent fire at a very early stage. So gamaged can be minimised and the ecosystem nurtured within it can be protected.
Authors :
- Srijani Das (Team Lead)
- Hadeeqa Nishat
- Sayanti Dutta
- Arbhijay Saha
- Farhan Hai Khan
