MCA Pilot PUI: Data Intensive Research Training (DIRT) in forecasting soil respiration at core terrestrial NEON sites

在此次NSF职业生涯中期推进奖(MCA)中，PI将把他在高级生态系统信息学方面的知识和技能扩展到一项名为“大数据”的环境数据科学问题上。生态预测科学对生态系统做出预测，以应对环境变化。反过来，生态预测又有助于开发基于自然的解决方案，以应对地球不断变化的气候。一些预测存在很大程度的不确定性，例如对陆地碳循环的预测。土壤中二氧化碳的得失，即所谓的土壤外流，是这种不确定性的一个重要来源，因为它们依赖于具有高度变异性的气候输入，如温度和降水。这项研究将利用数学和计算方法开发开放获取的土壤二氧化碳排放的实时预测。预报将在美国大陆的47个长期研究地点进行，这些地点是NSF支持的国家生态观测网络(NEON)的一部分。首席调查员(PI)将与NSF支持的生态预测倡议的一名导师合作，进一步开发一种工具，以应对社区对土壤排出物的预测挑战，导致其在生态社区中得到更广泛的采用。通过这个项目获得的技能将为PI提供长期和可持续的研究轨迹，并将帮助他发展新的本科生研究和培训经验。对陆地碳循环各组成部分(即总初级生产力和净生态系统碳交换)进行实时预测，对于了解和制定气候变化情况下的基于自然的解决办法十分重要。尽管土壤碳储量对气候缓解很重要，但由于缺乏土壤过程的结构性表示，陆地土壤碳排放的标准化预测滞后于其他陆地碳预测，因为(目前)缺乏土壤过程的结构性表示。该项目将对所有霓虹灯地面站点的季节和年际时间尺度上的土壤碳外流进行预报。项目成果包括将一套土壤碳模型(跨越一系列模型复杂性)纳入生态系统建模的综合信息学工具箱(预测生态系统分析仪或山核桃)，以生成土壤排放预测。在适当的情况下，将对照现有的测量或数据库(例如土壤呼吸数据库或连续土壤呼吸数据库)对模拟的排放进行验证。该项目将直接将霓虹灯数据和生态预测技术与更广泛的生态社区联系起来。开发的方法和工具将对科学界开放获取，特别是通过开放源码R包和将模型纳入山核桃生态信息学工作流程。该奖项反映了NSF的法定使命，并通过使用基金会的智力优势和更广泛的影响审查标准进行评估，被认为值得支持。

In this NSF Mid-Career Advancement (MCA) award, the PI will expand his knowledge and skills in advanced ecosystem informatics to a “big data” environmental data science problem. The science of ecological forecasting makes predictions about ecosystems in response to environmental change. Ecological forecasts in turn, aid in development of nature-based solutions for Earth's changing climate. Some forecasts have a large degree of uncertainty, such as those for terrestrial carbon cycling. Gains and losses of CO2 from the soil, known as soil efflux, are an important source of this uncertainty because they depend on climate inputs with a high degree of variability, such as temperature and precipitation. This research will develop open-access real-time forecasts of soil CO2 effluxes using mathematical and computational approaches. Forecasts will be developed across 47 long-term research sites in the continental United States that are part of the NSF-supported National Ecological Observatory Network (NEON). The principal investigator (PI), in partnership with a mentor from the NSF-supported Ecological Forecasting Initiative, will further develop a tool to address a community forecasting challenge for soil effluxes, leading to its broader adoption in the ecological community. The skills gained through this project will provide a long-term and sustainable research trajectory for the PI, and will help him develop new undergraduate research and training experiences. Real-time forecasts of components of the terrestrial carbon cycle (i.e. gross primary productivity and net ecosystem carbon exchange) are important for understanding and developing nature-based solutions in a changing climate. Despite the importance of soil carbon storage to climate mitigation, standardized prediction of terrestrial soil carbon efflux across seasonal and interannual timescales lags behind other terrestrial carbon forecasts with increased model forecast uncertainty from a (current) lack of structural representation of soil processes. This project will develop forecasts for soil carbon efflux across seasonal and interannual timescales at all NEON terrestrial sites. Project deliverables include incorporation of a suite of soil carbon models (spanning a range of model complexity) into an integrated informatics toolbox for ecosystem modeling (the Predictive Ecosystem Analyzer or PEcAn) to generate soil efflux forecasts. Modeled effluxes will be validated against existing measurements or databases (e.g. Soil Respiration Database or the Continuous Soil Respiration Database) where appropriate. This project will directly connect NEON data and ecological forecasting techniques with the broader ecological community. Developed methods and tools will be open-access for the scientific community,notably through an open-source R package and by incorporation of models into the PEcAn ecological informatics workflow.This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.