Developing a statistical methodology for the assessment and management of peatland (StAMP)

在良好的条件下，泥炭地是所有土壤中最有效的碳储存。它们调节淡水供应（泥炭地95%是水）和质量，通过储存温室气体来减缓气候变化，并保持生物多样性。土地使用管理干预措施（例如，将泥炭用于农业、排水、林业、燃烧用于狩猎管理和娱乐）可能会破坏泥炭数千年来锁定的巨大碳储存，从而损害所有这些服务的提供。英国有2 Mha的泥炭地（10%的土地面积），然而，高达80%的泥炭地在某种程度上受到破坏。据估计，退化的英国泥炭地排放10吨碳的a-1，类似的规模炼油厂或垃圾填埋场，使英国的前20个国家的碳排放量从降解泥炭。恢复退化的泥炭地以阻止碳损失是应对气候变化全球战略的重要组成部分。然而，到目前为止，我们还没有一个工具来帮助我们评估土地利用如何以具有成本效益的方式影响泥炭地状况，在大型和偏远地区，这使得很难确定哪些地区应该优先进行管理干预。在英国，数百万英镑的公共资金已经投资于大规模的泥炭地恢复项目，但我们没有一个可靠和强大的方法来评估恢复的有效性。这些是我们知识中的重要空白，使我们无法在泥炭地管理方面做出具有成本效益的选择。通过该项目，我们将开发新的统计方法，从卫星收集的数据中检测泥炭地景观状况的变化。在以前的一个研究项目中，我们表明，泥炭地的条件可以从测量泥炭表面运动的卫星数据中找到。湿泥炭在良好的条件下表现出非常不同的特点，干泥炭在恶劣的条件。然而，我们基于卫星的方法产生了太多复杂的数据，无法通过肉眼进行可靠和一致的分析。我们的目标是通过开发一种新的统计方法来为泥炭地管理决策提供信息，该方法可以从卫星数据中稳健和一致地量化泥炭地景观的变化。这需要能够处理非常大和复杂的结构化数据集的方法。在统计学中，一种新的框架，称为面向对象的数据分析（OODA），非常适合通过基于适当的数据对象选择构建模型来实现这一目的。OODA可用于开发用于检测变化的简约模型，并用于量化预测中的不确定性。作为空间和时间函数的卫星数据的OODA将能够对不同区域的趋势和变化进行建模，并检测泥炭地的变化。我们的项目将进一步发展OODA方法比它目前的能力，并将这种方法应用于泥炭表面运动的卫星数据集。其结果将是一系列地图，说明泥炭地景观随时间的变化，旨在供土地管理者和政策制定者使用，以指导决策。这将有助于减少不必要的支出，并优先考虑最紧迫和最具战略意义的泥炭恢复地区。我们将最先进的统计方法与卫星数据相结合的新方法将为评估泥炭恢复投资并向资助机构报告提供可靠的工具。利用统计数据量化泥炭景观变化的能力应该为泥炭地管理者和那些资助和投资泥炭地恢复的人提供信心，使他们能够为泥炭地做出更好的选择。

In good condition, peatlands are the most efficient carbon store of all soils. They regulate freshwater supply (peatlands are 95% water) and quality, mitigate climate change by storing greenhouse gases, and maintain biodiversity. Land use management interventions (e.g. use of peat for agriculture, drainage, forestry, burning for game management and recreation) can compromise the delivery of all these services by destabilising the vast carbon store that peat has locked away over thousands of years. The UK has 2 Mha of peatlands (10% land area), however, up to 80% of these peatlands are damaged to some degree. It is estimated that degraded UK peatlands emit 10 Mt C a-1, a similar magnitude to oil refineries or landfill sites, placing the UK among the top 20 countries for emissions of carbon from degrading peat. Restoring degraded peatlands to halt carbon losses is an essential part of a global strategy to fight climate change. However, to date, we do not have a tool to help us assess how land use affects peatland condition in a cost effective manner over large and often remote areas, making it difficult to identify which areas should be prioritised for management intervention. In the UK, several millions of pounds of public money have already been invested in large-scale peatland restoration projects yet we do not have a reliable and robust way to evaluate the effectiveness of restoration. These are important gaps in our knowledge that prevent us from being able to make cost-effective choices when it comes to peatland managementWith this project, we will develop new statistical methods to detect change in the condition of peatland landscapes from data collected by satellites. In a previous research project, we showed that peatland condition can be found from satellite data that measures surface motion of the peat. A wet peat in good condition displays very different characteristics to dry peat in poor condition. However, our satellite-based approach produces too much complex data that cannot be reliably and consistently analysed by eye.We aim to inform peatland management decisions by developing a new statistical method that can robustly and consistently quantify the changes in the peatland landscape from the satellite data. This requires methods capable of handling extremely large and complex structured datasets. In statistics, a new framework, known as Object-Oriented Data Analysis (OODA), is ideally suited to achieve this purpose by building models based on suitable choices of data objects. OODA can be used for developing parsimonious models for detecting change, and for quantifying uncertainty in predictions. OODA of the satellite data as functions of space and time will enable the modelling of trends and variability in the different regions, and the detection of reg change in the peatland. Our project will develop the OODA method further than its current capabilities and apply this method to the satellite datasets of peat surface motion. The result will be a series of maps that illustrate the change in peatland landscape over time that are designed to be used by land managers and policy makers to guide decision making. This will help reduce unnecessary spending and prioritise the most urgent and strategic areas for peat restoration. Our novel approach combining state-of-the-art statistical methods with satellite data will provide a reliable tool to evaluate investments in peat restoration and report to funding bodies. The ability to quantify changes in the peat landscape using statistics should provide confidence to peatland managers and to those who fund and invest in peatland restoration, enabling them to make better choices for peatlands.