Intellectual Merit: The origin, function, and fate of dissolved organic carbon (DOC) in terrestrial ecosystems and its transport processes from landscapes to sea are only partially understood. In this proposed research, statistical and GIS-based transport models are used to study DOC dynamics in terrestrial ecosystems and rivers. The research objectives are three-fold: 1) To analyze the spatial variability of DOC contributions from land surfaces to streams at sub-basin scales. 2) To identify the transport and transformation mechanisms behind long (decades) and short (days to months) term DOC fluctuations due to natural and anthropogenic influences. 3) To predict DOC concentration from remote sensing reflectance, turbidity and chlorophyll in rivers.
In the past several years, a large amount of field data, has been collected including 5-years of monthly DOC measurements in 30 sub-basins, high spatial and temporal resolution riverine observations, and remote sensing imagery. Equipped with these data, the team will rely mainly on an approach that integrates statistical modeling and GIS-based transport modeling. Specifically, they will 1) Use an adaptive varying coefficient mixture model to derive monthly DOC loads from uniform land use types. 2) Construct a GIS-based transport model for routing DOC mass in the drainage stream network. 3) Use a varying coefficient functional linear model and nonparametric functional model to predict DOC through remote sensing 4) Validate the above models with independent data sets from two specific river watersheds.
The geosciences contribution from this project lies in the improved understanding of the DOC export processes influenced by interactions among human activities, natural events, watershed characteristics, and climate. The statistical contribution lies in the innovative estimation and inference procedures that are particularly adapted to this complex geoscience data.
Broader impacts: This proposed research will improve the scientific understanding of carbon export processes from land to coastal waters. Funding of this project will potentially change the public?s perception of the influence of their actions in coastal communities. This research will result in a modeling technique that is suitable for studying interactions among watershed properties, human activities, natural events and climate change. The project provides learning opportunities for all participating students. The methods, models, and data from this research will be integrated into undergraduate and graduate teaching programs in a broad range of disciplines (statistics, biogeochemistry, remote sensing, hydrology, and modeling). The collaboration offers students access to cutting edge modeling techniques and environmental sciences through an equal partnership of two diverse research institutions. Especially, the two female PIs will be valuable assets in guiding female graduate students to become future fellow scientists (statisticians and geo-scientists) with their own experiences. Coastal carbon research-related instructional materials will be developed and disseminated through existing educational programs and The Bridge website. These outreach efforts will strongly benefit from early experience on the COSEE-New England and Watershed-Integrated Sciences Partnership (GK-12) programs.
Results from this project improved our understanding of the origin, function, and fate of dissolved organic carbon (DOC) in terrestrial ecosystems and the associated transport processes from landscapes to waters. We quantitatively identified that terrestrial DOC export is an interactive processes among climate (temperature), terrestrial ecology (hydrological processes), and human activities (land cover/land use). The key intellectual merits from this study are 1) the discovery of spectral reflectance for freshwater CDOM and non-algal particles are separable from hyperspectral satellite images, 2) first quantitative evidence of DOC production rate changes driven by environmental factors, and 3) identification of the differences between DOC degradation rate changes from individual vegetation species. These research progresses are resulted from multiple years of field observation and mathematical modeling in different spatial and temporal scales. The major contributions in method innovation include the first effective remote sensing algorithm for extracting dissolved organic matters from freshwater, a feasible controlled experiment for examining DOC production and leaching processes, a single index varying coefficient model (SIVCM) for dynamic unmixing of DOC export attributable to each land cover type, and the development of a process-based model (ArcTeDOC) for describing spatial and temporal DOC cycles from land to rivers. The significant breakthrough is that a new semi-analytical remote sensing algorithm was developed to retrieve CDOM from satellite images in fresh and coastal regions. Conventionally, the combined absorption, adg(440) of CDOM and non-algal particles have been used as surrogates for CDOM in open seas. Unfortunately, this surrogate does not work well for river or near-shore waters due to the high levels of non-algal particles from terrestrial ecosystems. This study revealed that remote sense reflectance at four wavelengths provides sufficient information to separate the levels of freshwater CDOM and non-algal particles. This algorithm (QAA-CDOM) was tested effective in world-wide river systems exhibited a broad range (0.02–7.2 m-1) of CDOM levels (Figure 1). This research result shows the promise of assessing spatial and temporal carbon exchange between terrestrial and aquatic environments. A more significant scientific finding is that the relationships between DOC concentration and corresponding optical properties (CDOM) are linear but with distinctly different slopes between three vegetation types: deciduous forest, evergreen forest, and agricultural plants (Figure 2). This explains why the quantitative relationship between freshwater CDOM and DOC has been found inconsistent at a range of sub-basin outlets. This finding indicates substantive potential for improving remote sensing of freshwater DOC flux by coupling QAA-CDOM algorithm with terrestrial sources. Field measurements and experimental results provide evidence that spatial variability of terrestrial DOC dynamics mainly governed by the production and leaching rates that are functions of vegetation type and density, surface temperature, soil water content at sub-basin scales (Figure 3). The temperatures and soil hydrology are dominant factors for annual terrestrial DOC flux to rivers in large watersheds of different climate zones (Figure 4). Our mathematical model SIVCM also identifies temperature as the single most important factor affecting DOC production rate among all relevant environmental factors considered. It also reveals that in watersheds with mixed land cover types, wetland produces DOC at the highest rate (Figure 5). This result has a large implication of terrestrial DOC exports responding to climate change. The DOC leaching rates from leaf litters is in descending order of deciduous, evergreen, and forest land uses. It concluded that land cover type is a significant variable driving the spatial variation of soil DOC and heterogeneous landscapes potentially affect terrestrial carbon loss. These DOC production and degradation processes are important scientific basis for integrative modeling studies of carbon cycling from land to water. The main broad impacts can be highlighted in three aspects. 1) This research will improve the scientific understanding of carbon export processes from land to coastal waters. The results are vital to calibrate and validate ecosystem models of carbon cycling in large scales. The resulted product, ArcTeDOC, will be used to help to change public’s perception of the influence of their practices in land uses and hydrological connectivity on the earth system. The technical innovation is applicable to scientific fields for studying interactions between watershed properties, human activities, natural events and climate change. Our efforts in training REU students and environmental modeling classes showed that research results will strongly benefit to K-12 and college students’ science and technological education as well as the communities in Watershed-Integrated Sciences Partnership programs.
