Introduction Three major features: Monitor vegetation growth and notify Predict and simulate future vegetation growth areas Risk scoring based on fire threat Users: Community members. Monitor vegetation growth around their community and alert authorities. Power company. Plan vegetation management around power lines today and in future Problem statement When trees grow close to power lines, they cause electric spark due to a phenomenon called di-electric breakdown. This triggers wildfire in high fire risk areas and causes power equipment damage which is very expensive and time consuming to replace. Power line operators struggle to manage vegetation getting close to their powerline as these lines are spread across thousands of miles often over hills and forest which is difficult to access. Solution Using open source geo spatial, powerline infrastructure, and satellite imagery data we can analyse a macro geographical area and determine high risk areas where vegetation is getting too close to power lines. This helps power line operators to plan and prioritize tree pruning activities at high risk areas. In addition, using historical canopy height and weather data, we can predict vegetation growth using advanced AI techniques, at various locations for the next few weeks or months. This helps operators plan their vegetation management activities for the future. App Web application Landing page is a US Map or California Map. Based on open source Google Map or Arch GIS etc. Add Powerline infrastructure layer to the Map. Transmission lines, Distribution lines, Towers etc. Obtain transmission line height data. When not available, estimate based on line KV rating Add vegetation layer for the entire area being supported (US or CA) Get real time canopy height data for the supported Area (US or CA) Maintain a list of cities/locations where vegetation detection algorithm needs to run. Take 10 or so cities / locations. Can be expanded if needed. Dataset Canopy Height / Lidar data from USGS: https://apps.nationalmap.gov/3depdem/services California power line: https://cecgis-caenergy.opendata.arcgis.com/datasets/CAEnergy::california-electric-transmission-lines-1/explore Satellite imagery for vegetation: Sentinel-2 (ESA) - https://scihub.copernicus.eu/ Vegetation Detection When one of the locations / Cities is selected, do the following: Take 50 mile radius of the selected location Get the power line segment in the above circle Obtain a list of geo locations along the line segment 10 cm apart. Exclude segments where there is no vegetation. Calculate clearance of each of the geo locations selected using the Lidar canopy height data and transmission line height data at that location. Maintain a clearance threshold table for different KV rating of transmission lines If the calculated clearance is less than the threshold, display an alert / balloon popup at that location. Prediction and Simulation Model Building Dataset: For the chosen (10) locations, get the past few years (3?) canopy height data. Convert to average height for each week Future: Add weather data at the location (Temp, Humidity, pressure, wind) Model: Using the Location, and Height as input build a model using few ML techniques: Linear Regression Ensamble (Binary tree, Random Forest) Neural Networks. Use confusion matrix to calculate Precision, Recall and F1 scores for each technique. Use highest F1 score to choose the model deploy model Predict Canopy Height Implement a prediction function. Takes 2 inputs: Location Week of the year Output : Canopy height at the location Predict Clearance Input: Location, Canopy Height Output: Green, Orange, Red; Clearance in feet User Interface Use the play button to increment the week of the year(WOY), starting from the current date. For each week, send the WOY and location to the predicted canopy height function. Use canopy height output to predict clearance Use the green, orange, red output to plot on the map. Display clearance and canopy height data Risk scoring