According to legend, Yakushima still has the giant tree that larger than Jomon-sugi. To search for this legendary giant tree, we used aerial LiDAR data to estimate the distribution area of the giant tree. In the analysis using low-density LiDAR data, the factors related to the distribution of the giant tree were elevation, tree height, tree crown area, tree species and topography, and the giant trees were extracted from these elements. In the analysis using the high-density laser, the factors related to the distribution of giant cedar were set as the gap in the forest and the gentle slope of the valley. We extracted these elements from voxel analysis and estimated the distribution of giant trees.

In response to the recent frequent occurrence of disasters, the program named “strengthening national resilience (disaster resilience/disaster mitigation)” was adopted in FY2018 as one of the Strategic Innovation Creation Program (SIP) by the Cabinet Office. In this project, we will make full-scale use of satellite observation data, through the constellation of satellite observations and utilization technology that uses multiple satellite instruments for disaster resilience. The heavy mission of this project is to provide disaster information by satellite to the disaster response headquarters within 2 hours after the disaster. This is because in the event of a large-scale disaster at the national level, the first response headquarters meeting will be held in about two hours after the disaster.

Then we are developing a system that enables targeting the observation area by analyzing the meteorological information, the prediction of flooding possibility simulator and the ordinally observed satellite data. Here, we will pick up the case of Typhoon No. 19 for an example that occurred in 2019 and introduce the processing results.

This case study improves the method of interpretation and damage analysis of landslide disaster were compared for emergency response. Compared with the previous artificial intelligence model, the new one improved both precision and recall. This is due to the improved teaching data and learning methods. It will continue to be necessary to develop a model that reduces mis-extraction by adding training data and devising additional layers.

In this paper, we report on a method for rapid extraction of inundation area and flood depth using the satellite SAR and the automated system. The inundation extent was extracted by the threshold method and the flood depth was estimated from the high resolution DEM based on the inundation extent. The method was applied to heavy rainfall disasters last year, and the properties and issues of the method were identified. In the future, we plan to build a cloud system to enable a quick response in the event of a disaster.

This study attempted to explore a faster and low cost solution for flood area extraction by integrating convolution neural networks (CNNs) with high resolution (1.5 m) SPOT satellite images. By consider the system requirement as a measure of cost, capabilities (speed and accuracy) of a deeper (ResNet101) and a shallower (MobileNetV2) CNNs on flood mapping were examined and compared. The models were trained and tested with satellite images captured during several flood events occurred in Japan. It is observed from the results that ResNet101 obtained better flood area mapping accuracy than MobileNetV2. Whereas, MobileNetV2 is having much higher capabilities in faster mapping in 0.3 sec/km² with a competitive accuracy and minimal system requirements than ResNet101.

To examine the relationship between the dust aerosol load and cloud thermodynamic phase, the fraction of ice phase (FIP) at the cloud top was analyzed using space-borne Cloud-Aerosol Lidar with Orthogonal Polarization (CALIOP) products in the summer (JJA) and winter (DJF) for one year. The target regions were east Eurasia (90˚E-130˚E, 40˚N-65˚N) and a part of America (70˚W-110˚W, 40˚N-65˚N). Unlike previous studies, we used the extinction coefficient (σ) as a more quantitative indicator of the dust aerosol load than relative frequency. Values of σ were larger in east Eurasia than in America, and the FIP between 250 K and 260 K was also approximately 20 % higher in east Eurasia than in America. As Kawamoto et al. (2020) point out, the difference in FIP would depend on the difference in σ. This means that FIP would be determined not only by temperature but also by σ. Next we compared FIPs in the two regions under consideration under the same temperature and σ conditions, and found again that FIP was considerably higher in east Eurasia than in America. There are two possible reasons for this phenomenon. First, dust aerosols might have different ice nucleating abilities depending on their locations of origin and chemical components, even if the aerosols are classified into the same ‘dust’ category by the satellite algorithm. The second explanation is that the analyzed sites might also contain other types of aerosols that could serve as ice nuclei, thereby changing the overall ice nucleating ability together with the dust aerosols. In-situ observations are required to investigate these reasons in detail, however. The present results will contribute to achieving more accurate climate model simulations by incorporating a more realistic relationship between σ and FIP.