特定場址的通道模擬可藉由可重建實驗、可稽核原生證據與穩定介 面形成可重用研究資產。本文提出一套用於 AODT 1.5.1 資料生成、 證據驗證與 contract-gated Network-AI integration 的自動化十元件空 中數位分身流程。主要評估依發布的 Lake Wheeler workflow，從來源 註冊一路執行到 physical validation；3，530 個 UAV positions 與五個 tower sites 以 paper-aligned 3.3-GHz SISO profile 映射到 AODT。RSS 在 11 個獨立確認位置比較，原論文在同一 Lake Wheeler trajectory 上 發布的 Sionna RT result 也取樣至相同 anchors，作為 reference-paper curve。AODT-to-measurement 最大空間誤差為 1.769 m。LW2–LW3 呈 中度一致，LW4 保留最強趨勢但有 residual level bias，LW1 較弱，LW5 受 outage 主導。另一項 contract-selected case 使用獨立的版本化 input profile 執行 Network-AI integration path，並保存回傳 artifacts。結果 建立具有 field evidence、same-trajectory Sionna RT reference 與已實作 Network-AI integration path 的可重建開放流程；Apache-2.0 專案原始 碼、runbooks、configurations 與 release metadata 已發布於 GitHub。;Site-specific channel simulation is reusable when researchers can reconstruct an experiment， audit its native evidence， and connect downstream tools through stable interfaces. We present an automated ten-component aerial digital twin pipeline for AODT 1.5.1 dataset generation， evidence validation， and contract-gated Network-AI integration. The primary evaluation follows the released Lake Wheeler workflow from source registration through physical validation. It maps 3，530 UAV positions and five tower sites into AODT using a paper-aligned 3.3-GHz SISO profile. Received signal strength (RSS) was compared at 11 independently verified positions after deterministic alignment to the 10-W transmit-power reference. The published Sionna RT result for the same Lake Wheeler trajectory was sampled at the same anchors as a reference-paper curve. The maximum AODT-to-measurement spatial discrepancy was 1.769 m. Agreement varied by link: LW2–LW3 were moderate， LW4 retained the strongest trend with a residual level bias， LW1 was weak， and LW5 was outage dominated. A separate contract-selected case exercised the Network-AI integration path through an independent versioned input profile and recorded the returned artifacts. These results establish a reconstructable open pipeline with field evidence， a same-trajectory Sionna RT reference， and an implemented Network-AI integration path. The Apache-2.0 project source， runbooks， configurations， and release metadata are published on GitHub.
