<![CDATA[Traditional communication infrastructures are vulnerable to failure during natural disasters in dense urban areas, hindering early warning dissemination. Delay Tolerant Network (DTN) offers a resilience, yet most studies relies on simulations, overlooking physical layer constraints affecting on opportunistic routing in real-world urban scenarios. This study addresses that specific gap by empirically validating a DTN-LoRa prototype-based communication system for flood data monitoring in Surabaya. A 4-node field test simulated the Store-Carry-Forward mechanism. A fixed node in the Dukuhpakis flood zone (TMP) transmitted bundles to mobile nodes on public transport routes (FD9, FD4). These data mules relayed the bundles via an opportunistic contact at Marmoyo Shelter, delivering them to a gateway node at the BPBD Command Post (PMI Surabaya). Performance was evaluated by Packet Delivery Ratio (PDR) and Latency under varying LoRa parameters (Spreading Factor, Coding Rate). The experiment validated the functional DTN architecture, achieving 100% PDR in the optimal configuration (SF7, CR4/7). The key finding was a "Contact Window Bottleneck" as a critical failure factor. LoRa configurations with high Time on Air (ToA) failed to transfer the entire data bundle within the narrow opportunistic contact window between mobile nodes, causing PDR to drop as low as 60%. Implementation success depends on the physical layer throughput that must be high enough to complete data transfer during brief opportunistic contacts, rather than merely maximizing signal range. These findings provide a critical performance baseline for disaster management agencies, demonstrating a feasible, low-cost architecture that can enhance the reliability of urban disaster response communication.]]>
