CAREER: Elucidating Large-Scale Spatial Patterns of Ecosystem Traits with Data Assimilation

计算机模型被用来对地球上的生命状态及其周围的大气做出全球预测。这些模型对全球气候以及与地球上植物和微生物生命的联系做出了重要的预测。其中许多模型依赖于关于地球上的植物及其与大气的联系的简单关系。这个职业奖将探索发展这些非常重要的关系和导致它们不同的因素(土壤、光线、水等方面的变化)的新方法。由于没有足够的关于植被和土壤类型空间变异的信息，大多数模型假定植被响应类型只基于土地覆盖类型而不同。过去的研究表明，其他众所周知的特性会影响植被的敏感性，例如某个特定地点的干燥程度，或者土壤中有多少粘土。这项职业奖将使用新的建模框架和卫星数据来得出世界各地最佳工厂参数的地图，并测试这些关系，即使在现场测量稀缺的地区也是如此。这项研究还将确定，使用这些关系是否可以改善生态系统吸收多少二氧化碳的模型预测。该奖项的结果将提高对生态系统如何应对气候变化的预测，使对二氧化碳吸收、植物生长和土壤分解的预测更加准确。此外，该奖项包括从高中生(教师培训)到本科生(包括课程材料的重新设计，以及本科生的研究经验)到大学后(创建一个将观察纳入模型的数学技术研讨会)的几个教育组成部分。陆地生态系统的大规模模型是气候变化预测的主要不确定性来源之一。尽管几十年来一直在努力提高流程表示的复杂性，但它们仍然不确定。然而，对参数优化的研究却少之又少。生态系统模型参数的分配完全基于少数几种植物功能类型，而没有考虑到全球植物行为的巨大多样性。该项目将测试一种形成植物功能类型替代品的新途径：使用数据同化。拟议的工作将使用碳数据模型框架(豆蔻)，该框架结合了简单的生态系统模型、遥感数据和马尔可夫链蒙特卡罗模拟，以确定生态系统参数，从而在全球每个像素产生最真实的通量和碳库。得到的参数图不能直接用于其他模型，但将用于测试所谓的环境过滤关系，以根据其他已知的空间变化因素(如气候、土壤类型等)预测生态系统参数的可变性。该奖项将测试是否可以利用豆蔻的同化遥感数据，使用类似于最近现场分析的方法，在全球范围内得出环境过滤关系，但不依赖现场测量的质量和数量(在热带等传统上采样不足的地区尤其成问题)。它还将创造和展示异养呼吸的这种关系的价值，其空间变异性不能仅由现场测量来限制。该项目的教育部分包括为中学生物、化学和物理教师开发几个与生态系统过程和气候变化有关的主题的教学模块。该项目还将用于支持与CARDAMOM进行两年一次的数据同化研讨会。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。

Computer models are used to make global predictions about the state of life on earth and the atmosphere that surrounds it. These models make important predictions about global climate and the links to plant and microbial life on earth. Many of these models rely on simple relationships about plants on earth and their connections to the atmosphere. This CAREER award will explore new ways of developing these very important relationships and the factors (changes in soil, light, water, and more) that result in their differences. Because there is not enough information about spatial variations in vegetation and soil types, most models assume that vegetation response types vary only based on land cover types. Past research suggests that other well-known properties affect vegetation sensitivities, for e.g. how dry a particular location is, or how much clay the soil has. This CAREER award will use a new modelling framework together with satellite data to derive a map of optimal plant parameters around the world, and to test these relationships even in regions where field measurements are scarce. The research will also determine whether using these relationships can improve model predictions of how much carbon dioxide ecosystems absorb. The results of this award will improve predictions of how ecosystems respond to climatic changes by enabling more accurate predictions of carbon dioxide uptake, plant growth, and soil decomposition. Additionally, this award includes several educational components for high school students (teacher training) through undergraduates (including redesign of the material for a class, and undergraduate research experience) to post-collegiate (creating a workshop on mathematical techniques for incorporating observations into models). Large scale models of terrestrial ecosystems are one of the dominant sources of uncertainty in predictions of climate change. They have remained uncertain despite decades of effort to increase the sophistication of process representations. However, much less attention has been paid to parameter optimization. Ecosystem model parameters are assigned solely based on a handful of plant functional types, without accounting for the enormous variety of plant behavior across the globe. This project will test a new pathway for forming alternatives to plant functional types: using data assimilation. The proposed work will use the CARbon DAta MOdel fraMework (CARDAMOM), which combines a simple ecosystem model, remote sensing data, and Markov Chain Monte Carlo simulations to determine ecosystem parameters that result in the most realistic fluxes and carbon pools in each pixel across the globe. The resulting parameter maps cannot be used directly in other models but will be used to test so-called environmental filtering relationships to predict ecosystem parameter variability based on other factors whose spatial variation is well known (e.g. climate, soil type, etc). This award will test whether assimilating remote sensing data in CARDAMOM can be used to derive environmental filtering relationships across the globe using approaches similar to those from recent in situ analyses, but without relying on the quality and quantity of in situ measurements (particularly problematic in traditionally under-sampled regions like the tropics). It will also create and demonstrate the value of such relationships for heterotrophic respiration, whose spatial variability cannot be constrained by in situ measurements alone. The educational components of the project include development of several instructional modules on topics related to ecosystem processes and climate change for middle and high school biology, chemistry, and physics teachers. The project will also be used to support a bi-annual workshop on data assimilation with CARDAMOM.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.