60 followers.
This project is aimed at reducing crop burning and air pollution by: 1) Identifying a) when & where farm waste is being burned, b) what and how much is getting burned, 2) Intervening: by creating a marketplace for crop residue and connecting farmers, collectors and buyers of crop residue to provide better alternatives and reduce burning, and 3) Doing Policy advocacy: inform the government about areas where crop waste is getting burned and also, do advocacy for policy changes required for utilization of crop waste
Improve the environment and health of residents by reducing crop burning
Create marketplace to provide alternatives to farmers to crop burning Do policy advocacy
Our primary sources of data include: 1. Satellite data: We have access to satellite data such as that from Sentinel satellites (specifically Sentinel 1 SAR and Sentinel 2 multispectral) as well as satellite data from Google Earth, NOAA, and USGS. We will also in​clude already created indices such as Normalized difference vegetation index (NDVI) and ​Normalized Difference Water Index (NDWI) as part of our initial data. 2. On the ground data for building and validation the AI models: This involves data on where the biomass is and what types of crops are being grown where. We have some limited data that has already been collected and as part of this project, we will be collecting additional data to build and validate our models initially on 300-400 acres of land in Punjab. 3. Fire data: that contains fires that have happened in a given area using both satellite (FIRMS data from NASA) and on the ground data from government agencies.
S1: Collect & Integrate data from satellite sources (e.g Sentinel, Google Earth, NOAA, and USGS) S2: Use satellite imagery to identify areas with fires (initially in the Patiala district and then scale to Punjab) S3: Identify areas where biomass was burned S4: Use data collected by Energy Harvest Trust with known crops in Punjab to create initial labeled data S5: Test feasibility of crop identification models using this initial labeled data S6: If unsuccessful, collect additional on the ground data to improve models S7: When successful, classify additional burned areas, and validate with on the ground data S8: Determine feasibility of marketplace based on accuracy of models in different areas and crops S9: Develop monitoring and alert system to intervene on new burnings S10: Plan to scale nationally