Inland waterways play a key role within the freight transportation system by connecting productive heartland areas to international gateways, while keeping costs competitive. Quantifying commodity flow is important because it affects cost-based supply chain decision-making. However, data on commodity movements to inform investment and planning decisions is elusive. Publicly available commodity data on U.S. inland waterways is limited in its spatial aggregation to the location of locks, which is insufficient to identify inter-port commodity flows. Automatic Identification System (AIS) data has the potential to disaggregate freight-flows to the port and river segment levels but it does not identify the commodity carried. This paper characterizes and quantifies vessel trips by port of origin-destination, timestamp, commodity carried, and path (mapped to an inland waterway network), allowing for disaggregated commodity flow analysis, previously unavailable in the public domain in the U.S.

This is accomplished through a multi-commodity assignment model which conflates AIS vessel movement data with commodity-specific port throughput. A stochastic approach is introduced to handle uncertainty in cargo-to-vessel ratios. Validation using data from the Arkansas River show agreement between model predictions and aggregated commodity volumes with differences lower than 1.82% by commodity and lock. Ubiquitous AIS data permit the transferability of the proposed work.

内陆水道在货运系统中发挥着关键作用，将生产性中心地带连接到国际门户，同时保持成本竞争力。量化商品流量很重要，因为它会影响基于成本的供应链决策。然而，为投资和规划决策提供信息的商品流动数据却难以捉摸。美国内陆水道上公开的商品数据在其空间聚合上仅限于船闸的位置，这不足以识别港口间的商品流动。自动识别系统 (AIS) 数据有可能将货运流分解到港口和河段级别，但它不能识别所载商品。本文通过始发港-目的地、时间戳、所载商品、

这是通过一个多商品分配模型来实现的，该模型将 AIS 船舶移动数据与特定商品的港口吞吐量相结合。引入了一种随机方法来处理货船比的不确定性。使用来自阿肯色河的数据进行的验证显示，模型预测与商品总量之间的一致性，商品和锁定的差异低于 1.82%。无处不在的 AIS 数据允许拟议工作的可转移性。