Research (En)

We are exploring the new and useful methods to integrate computer simulation and observation data. Computer simulation, which generates the hypothetical earth and artifacts, has been widely used in meteorology and civil engineering. Although computer simulation is a powerful tool to estimate the state of the earth system and predict its future, the skill of simulation to mimic the real earth system is generally limited. On the other hand, the observation data generally have high accuracy but it is difficult to observe the entire earth system directly and predict its future using observation only. Therefore it is strongly needed to integrate simulation and observation and produce a synergistic effect. Specifically, we focus on mathematics related to data assimilation and uncertainty quantification.

We have been developing data assimilation and uncertainty quantification which integrate a land model and satellite observation. Furthermore, we are exploring the coupled data assimilation system between atmosphere and land, and data assimilation of social dynamics. We recently focus on the application of machine learning to the simulation-observation integration.

Drought is one of the costliest natural disasters all over the world so that monitoring and prediction of droughts are the grand challenges in hydrometeorology. Drought can be categorized as meteorological drought quantified by precipitation, agricultural drought quantified by soil moisture, and hydrological drought quantified by stream flow and groundwater. Drought is complicated natural disaster which propagates various sectors in various spatiotemporal scales. We are calculating not only water but ecosystem damages by drought and exploring the accurate description of drought (called "ecohydrological drought"). In addition, we are developing the real-time ecohydrological drought monitoring and prediction system based on land data assimilation in the developing world.

To mitigate the risk of severe weather events, we need to effectively use the accurate prediction of these events. It is generally difficult to predict the rapid growth of convective precipitation systems and the associated local severe rainfall and flood events. This is because of extremely high spatiotemporal scales and nonlinear dynamics. We are exploring the globally-applicable method to predict not only the sudden local severe weather but also associated flood events using the rapid satellite observations and the other useful observations. In addition, we are developing the method for uncertainty quantification of this prediction and discussing how to effectively use the information of uncertainty to induce the preparedness actions. Based on the statistical mathematics for simulation-observation integration, we are solving the issues in the coupled systems of natural and social domains. We call this interdisciplinary area "socio-meteorology". The fundamental studies on socio-meteorology are expected to increase the resilience of our society.