Predictions of floodplain and streambank geomorphic change and flux of sediment and nutrients, and streambed characteristics, for stream reaches in the Chesapeake Bay and Delaware River watersheds

Input predictor variables and output predictions from statistical modeling of floodplains, streambanks, and streambeds for each NHDPlusV2 stream reach in the Chesapeake Bay and Delaware River watersheds of the U.S. Mid-Atlantic. Random Forest statistical models using either 1) characteristics of upstream drainage area, or 2) characteristics of upstream drainage area (Wieczorek et al. 2018, https://doi.org/10.5066/f7765d7v) and reach geomorphometry (Hopkins et al. 2020, https://doi.org/10.5066/P9RQJPT1), were used to explain and predict spatial variation in measured floodplain and streambank flux of sediment, fine sediment, sediment-C, sediment-N, and sediment-P and rates of geomorphic change, and streambed sediment characteristics (d50, cover by fine sediment, cover by fine and sand sediment) (Noe et al. 2020,https://doi.org/10.5066/P9QLJYPX). These predictions were compared to published estimates of upland erosion and delivery of upland erosion to streams (Chesapeake Bay Program, https://cast.chesapeakebay.net/Documentation/ModelDocumentation), and catchment export (Ator 2019, https://doi.org/10.3133/sir20195118).

本数据集包含美国中大西洋地区切萨皮克湾与特拉华河流域内，每一条NHDPlusV2河段的泛洪平原、河岸与河床统计建模所用的输入预测变量，以及模型输出的预测结果。本研究采用两类随机森林（Random Forest）统计模型：其一仅以上游汇水区特征作为输入变量；其二则结合上游汇水区特征（Wieczorek等，2018，https://doi.org/10.5066/f7765d7v）与河段地貌参数（Hopkins等，2020，https://doi.org/10.5066/P9RQJPT1）作为建模依据，用于解释并预测实测泛洪平原与河岸的沉积物、细沉积物、沉积物碳、沉积物氮、沉积物磷通量、地貌变化速率以及河床沉积物特征（d50中值粒径、细沉积物覆盖度、细砂混合沉积物覆盖度）的空间异质性（Noe等，2020，https://doi.org/10.5066/P9QLJYPX）。本数据集的预测结果还与已发表的高地侵蚀估算值、高地侵蚀向河道的输运量（切萨皮克湾项目，https://cast.chesapeakebay.net/Documentation/ModelDocumentation）以及流域出口通量（Ator，2019，https://doi.org/10.3133/sir20195118）进行了对比分析。