Methane emission monitoring and reduction in dairy farms using an optical nose on chip

减少温室气体排放是减缓气候变化的优先事项。在其2030年减排计划中，加拿大的目标是到2030年将排放量比2005年的水平减少40%，到2050年达到净零排放。尽管甲烷占温室气体的11%，但在20年的时间里，它的变暖能力是二氧化碳的86倍。因此，减少人为甲烷排放显得尤为重要。2019年，农业占加拿大甲烷排放总量的29%，其中大部分来自生物来源，如通过肠道发酵的畜牧业生产。这项提议的全球目标是开发一个基于芯片光学鼻子的平台，允许监测奶牛场的甲烷排放，以验证和改进农场干预措施的效果，以减少奶牛在哺乳期产生的甲烷。我们的合作伙伴Lactanet的使命是为加拿大各地的6800家生产商提供建议，以提高奶牛场的可持续性和盈利能力。光学鼻子将使用与微电子工业中使用的类似工艺在硅芯片上制造，并在聚合物中具有一系列光学微谐振器。这种方法允许以低成本制造大量在生产奶牛场部署所需的传感器。事实上，市场上商用传感器的高昂成本阻碍了它们的大规模使用。在这个项目中，我们将把机器学习策略应用到光学鼻子上，它将被放置在挤奶机器人的喂食箱中。为了开发可靠的算法来呈现24小时的甲烷排放，有必要开发模型，将来自饲料箱中的甲烷传感器的数据与从代谢室记录的连续甲烷排放进行关联，每天3次，每次5分钟。后者将与加拿大农业和农业食品公司合作实现。更好地了解和准确监测甲烷排放，不仅可以减少对环境的影响，还可以改善动物健康，提高生产率。事实上，甲烷生产可以消耗牛所摄取的高达12%的能量。

Reducing greenhouse gas emissions is a priority for mitigating climate changes. In its 2030 emissions reduction plan, Canada targets to reduce its emissions 40% below 2005 levels by 2030 and reach net-zero emissions by 2050. Although methane represents 11% of greenhouse gases, it has 86 times the warming power of carbon dioxide over a 20-year period. Therefore, it is particularly important to reduce anthropogenic methane emissions. Agriculture contributed to 29% Canada's total methane emissions in 2019, most of it coming from biological sources, such as livestock production through enteric fermentation. The global objective of this proposal is to develop a platform, based on an optical nose on chip, allowing the monitoring of methane emissions in dairy farms to validate and refine the effect of on-farm interventions to reduce methane produced by cows in lactation. The mission of our partner Lactanet is to advise 6800 producers across Canada to improve the sustainability and profitability of dairy farms. The optical nose will be fabricated on a silicon chip using similar processes as the ones used in microelectronics industry and have an array of optical microresonators in polymers. Such approach allows low-cost fabrication of large numbers of sensors necessary for the deployment in production dairy farms. Indeed, the high cost of commercial sensors available on the market are preventing their use at large scale. In this project, we will apply machine learning strategies to the optical nose, which will be placed in the feeding box of the milking robot. To develop reliable algorithms rendering a 24h methane emission, it is necessary to develop models correlating data from methane sensors in the feeding box, which are taken for 5 minutes 3 times a day, with continuous methane emission recorded from a metabolic chamber. The latter will be realized in collaboration with Agriculture and Agri-Food Canada. Better understanding and accurate monitoring of methane emissions have a great potential not only to reduce the environmental impact, but also for improved animal health, and increased productivity. Indeed, methane production can consume up to 12% of the energy ingested by cattle.