[H-CG27] [EE] 水―人間系の動態：観測、理解、モデル化とマネジメント

[HCG27-P12] Human in the Loop of Managing Early Warning of Coupled Dynamics and Risks with Poor Observations, Incomplete Understanding and Hybrid Modeling

*Mendiondo Eduardo Mario¹、Yamashiki Yosuke²、Horita Flavio³、Porto de Albuquerque Joao⁴、Seluchi Marcelo⁵ (1.Sao Carlos School of Engineering, University of Sao Paulo、2.Graduate School of Advanced Integrated Studies in Human Survivability (GSAIS), Kyoto University、3.Institute of Mathematical Sciences and Computation, University of Sao Paulo、4. Centre of Interdisciplinary Studies, The University of Warwick、5.National Center of Monitoring and Early Warning of Natural Disasters, Ministry of Science, Technology, Innovation and Communications, CEMADEN/MCTIC)

Here we address coupled human-water dynamics, especially with poor observation networks and incomplete understanding, by using emerging hybrid models. Human-in-the-loop (HITL) is defined as a model approach that requires human interaction. In HITL simulation, humans are usually a component, thereby influencing water-systems outcomes, sometimes in such a way difficult to reproduce exactly. From a Brazilian case study, we combine (1) a former application of a qualitative analysis of the early warning process in disaster management (see Horita et al, 2016) with (2) a novel, plausible HITL approach, viable for issuing anticipatory alerts on flood risks at a national scale. There are more than 40,000 flood-and-landslide prone areas officially mapped in Brazil. Also, 1 in 5 Brazilian municipalities depicted strong water-risks to population from hydrometeorological processes. Official alerts issued from CEMADEN-MCTIC are delivered at the municipality scale (community focused), with high uncertainty at the catchment scale (system driven). For that reason, a new generation of HITL depicts reasonable conditions for early warning of coupling dynamics and moving scales under complex realities of growing urbanization. In this contribution, we propose a new HITL in comparison with the observation network and alert database of the National Center of Monitoring and Alerts of Natural Disasters (CEMADEN/MCTIC), with more than 3,000 real time sensors, integrated at a 10-min time scale, installed in ca. 1000 municipalities. Novel indicators from this new HITL approach, based upon surrogate variables of hazard, vulnerability and exposure of flood-prone communities of Brazilian states, are here depicted. With emphasis in the North-East Region and South-East region, we derive HITL relations to help CEMADEN-MCTIC crisis room and the decision-making process of alerts at the national scale. Because the great nature of uncertainty, complexity of factors and cascade of decision-making rules, HITL should invoke promissory pathways for hybrid modeling under incomplete understanding of coupled human-water dynamics across scales significant to stakeholders. Reference and Supplementary Material: Horita, F., de Albuquerque, J., Marchezini, V., Mendiondo (2016) A qualitative analysis of the early warning process in disaster management, Short Paper, In: Community Engagement and Practitioner Studies Session, ISCRAM 2016 Conference (proceedings), Rio de Janeiro, 2016 (http://www.agora.icmc.usp.br/site/wp-content/uploads/2016/03/horita-iscram2016.pdf )