Elucidating the temporal variability of glacial organic carbon concentration and composition toward determining carbon export via discharge separation and machine learning techniques (Falljökull, Iceland)

与冰川径流相关的有机碳输出预测非常有限，现有的有机碳输出研究主要基于综合方法，使用单个冰采样点和质量平衡来计算冰川衍生的有机碳的平均年输出。这种质量平衡方法没有考虑到潜在的OC日变化和季节变化，因此可能不能准确反映冰川OC输出率。因此，为了进一步阐明冰川在河流OC输出中的作用，在高时间分辨率（季节、事件、日）下考虑冰川水文的时间变化，并考虑不同的相关径流成分是很重要的。本项目旨在以高时间分辨率系统地研究冰川衍生OC（浓度、组成和生物利用度）的输出，并考虑冰川产流变化的生物化学时间变异。只有详细了解冰川水文对OC输出的影响，才能对未来冰川退缩导致的OC释放做出可靠的预测。调查将在温和的冰岛冰川Falljökull （Öraefajökull和Vatnajökull冰帽的一部分）进行，自1932年以来每年测量一次（冰锋）。使用创新的方法，如机器学习方法，结合流量水流线分离，将有助于了解冰川流量不同源区的时间相互作用及其日变化和季节变化。将涉及OC和排放动力学的过程理解联系起来，可以在考虑OC组成的同时模拟冰川OC的输出。在冰川终点站安装的经校准的便携式水质测量仪，将采集972个冰、雪和水的样本，包括季节性、日间和事件抽样，以每隔60分钟自动测量水温、电导率、浊度、水位和荧光DOM。使用一系列最先进的实验室设备和方法（C/N-和toc分析仪，Picarro），我们将分析BDOC， DOC， POC，光学性质（荧光，吸光度），营养物质（PO4, NO3, NO2, NH4）和稳定同位素（18O, 2H）。多元统计技术（如PCA、CCA）的使用将有助于确定时间模式、过程和驱动因素。将模拟冰川OC的通量和输出（PARAFAC、SIMMR、LOADest）。结合排放分离和机器学习技术，对有机碳出口进行系统调查，将推进目前关于冰川衍生有机碳浓度、组成和生物利用度的日变化和季节变化的过程知识。此外，它将有助于可靠地预测由于气候变化引起的冰川融化过程变化而产生的（未来）冰川OC输出动态。

Predictions of organic carbon (OC) export related to glacier runoff is very limited and existing studies about OC export are primarily based on an integrated approach, using single ice sampling points and mass balances to calculate an average annual export of glacier derived OC. This mass balance approach does not account for potential diurnal and seasonal changes in OC, and may therefore not accurately reflect glacial OC export rates. Therefore, it is important to consider temporal shifts in glacial hydrology at a high temporal resolution (seasonal, event, diurnal), and to account for different relevant runoff components, in order to further elucidate the role of glaciers in riverine OC export. This project aims to systematically investigate the export of glacier derived OC (concentration, composition and bioavailability) at a high temporal resolution, and consider the biochemical temporal variability with variations in glacier runoff generation. Only by understanding the effects of glacier hydrology on OC export in detail reliable predictions for future release of OC due to glacier retreat can be made. The investigations will be carried out at the temperate Icelandic glacier Falljökull, part of Öraefajökull and Vatnajökull ice cap and measured annually since 1932 (ice front). The use of innovative methods such as machine learning methods, in combination with a discharge hydrograph separation, will help to understand the temporal interplay of different source areas of glacier discharge and its diurnal and seasonal variability. Connecting process understanding involving OC and discharge dynamics will enable the modelling of the export of glacial OC, while taking OC composition into consideration. Altogether 972 ice, snow and water samples will be taken including seasonal, diurnal and event sampling in connection with continuous measurements of water temperature, conductivity, turbidity, water level and fluorescent DOM automatically at 60 min intervals using a calibrated portable water quality meter installed directly at the glacier terminus. Using an array of state-of-the-art laboratory equipment and methods (C/N- and TOC-Analyzer, Picarro), we will analyze BDOC, DOC, POC, optical properties (fluorescence, absorbance), nutrients (PO4, NO3, NO2, NH4) and stable isotopes (18O, 2H). The use of multivariate statistical techniques (e.g., PCA, CCA) will help to identify temporal patterns, processes and drivers. Fluxes and export of glacial OC will be modelled (PARAFAC, SIMMR, LOADest). This systematically investigation of OC export combined with discharge separation and machine learning techniques will advance current process-knowledge about diurnal and seasonal changes in the concentration, composition and bioavailability of glacier derived organic carbon. Moreover, it will contribute to the ability to reliably predict the (future) dynamics of glacier derived OC export due to climate-change induced variations in glacier melting processes.