3/13/2023 0 Comments Airbeam sensorWhat was the hardest part of this project? The combination of deeply informed on-the-ground reporting and creative data-gathering and visualization helps make the project distinctive. It is important to note that just as important as specific technologies used were the months of on-the-ground reporting from reporters in the New Delhi bureau to identify appropriate families, gain entry to their homes and schools, and understand the broader social context behind Monu’s and Aamya’s lives. The timelines and side-by-side photos, using this new data, were made interactive in the browser with JavaScript, leaning heavily on the D3 JavaScript library. This new, combined data allowed us to generate all the time code-based visuals you see in the story. We used Google Sheets and a custom Node.js app to parse the various metadata and sync it all via their timecodes. The video files included sidecar metadata generated by the cameras. We recorded the air pollution data as CSV files. We worked with researchers from ILK Labs in Bangalore to design a data collection and processing protocol involving three types of portable air quality sensors and custom software running on a battery-powered Raspberry Pi computer. The visual contrast reflects a dispiriting reality: A long-term, consistent disparity like we observed that day could steal around five years more life from someone in Monu’s position, compared with an upper-middle-class child like Aamya.Īrden Pope, one of the world’s foremost experts on health and air pollution, called the piece “an engaging, important, and sobering story.” Scott Murray called it “the finest piece of data-driven visual journalism I have seen, ever, hands-down.” Most importantly, it is impossible to address health inequalities if they are not understood, and this piece provides an opening. We watch as both children brush their hair, hang out with friends and sit down for dinner, and see overlaid the spikes and valleys of their real-time pollution exposure. Monu, who lives in a slum and attends school outside, is exposed to about four times as much pollution as Aamya, whose school and home are guarded by air purifiers. We showed, moment to moment, what that exposure looks like. The most harmful pollutants are commonplace, legal and largely invisible. ![]() Few researchers have collected this data. While Delhi’s poor air quality is well-known, disparities in individual exposure based on class or circumstances are poorly understood. This project went to extraordinary lengths to make visible this dangerous reality. Impact reached:Įveryone does not breathe the same air. Performance remained strong during cross-validation, which included both a random sample approach (RMSE = 1.52 μg/m 3) and a spatial blocking cross-validation method (RMSE = 2.8 μg/m 3).We measured the air pollution that two children in New Delhi breathed as we followed them around the city on a normal day to see how wealth inequality affected their exposure. A stacked ensemble model was developed using machine learning, which had a training 5-fold cross-validation mean residual deviance of 3.82 μg/m 3, a root mean squared error of 1.95 μg/m 3, and a mean absolute error of 0.95 μg/m 3. The linear regression model performed poorly with a training R 2 of 0.15 and a cross-validation R 2 of 0.15. Both linear regression and machine learning approaches were applied. The long-term air pollution surface for the city was generated with a land use regression model. We applied a temporal adjustment algorithm to account for this imbalance in observation periods with the intention of producing long-term estimates representing the sampling period of 20. Cycling times were not balanced throughout the day nor the year. The adjustment was able to reduce the root mean squared error from 12 μg/m 3 to 3.8 μg/m 3, and the mean bias was reduced to −0.5 μg/m 3. A machine learning model was developed to adjust the sensor observations, which demonstrated their highest errors during periods of high humidity. We evaluated the accuracy of the sensor through a collocation study for 3203 h, which identified the sensor had a mean bias of 7.25 μg/m 3 and a correlation of r = 0.77 with an US EPA Federal Equivalent Monitor. The air pollution observations were obtained with a low-cost sensor. Mapping is completed using a land use regression model for Charlotte, North Carolina. In our paper, particulate matter less than 2.5 μm (PM 2.5) air pollution data obtained by community scientists while cycling is used to develop high-resolution spatial air pollution maps. ![]() Fine particulate matter air pollution is a global issue cycling is a global activity.
0 Comments
Leave a Reply. |
AuthorWrite something about yourself. No need to be fancy, just an overview. ArchivesCategories |