著者
OTSUKA Shigenori KOTSUKI Shunji OHHIGASHI Marimo MIYOSHI Takemasa
出版者
Meteorological Society of Japan
雑誌
気象集誌. 第2輯 (ISSN:00261165)
巻号頁・発行日
pp.2019-061, (Released:2019-09-03)
被引用文献数
1

Since January 2016, RIKEN has been running an extrapolation-based nowcasting system of global precipitation in real time. Although our previous paper reported its advantage of the use of data assimilation in a limited verification period, long-term stability of its forecast accuracy through different seasons has not been investigated. In addition, the algorithm was updated seven times between January 2016 and March 2018. Therefore, this paper aims to present how motion vectors can be derived more accurately, and how data assimilation can constrain an advection-diffusion model for extrapolation stably for the long-term operation. The Japan Aerospace Exploration Agency's Global Satellite Mapping of Precipitation (GSMaP) Near-Real-Time product is the only input to the nowcasting system. Motion vectors of precipitation areas are computed by a cross-correlation method, and the Local Ensemble Transform Kalman Filter generates a smooth, complete set of motion vectors. Precipitation areas are moved by the motion vectors up to 12 hours, and the product, called “GSMaP RIKEN Nowcast”, is disseminated on a webpage in real time. Most of the algorithmic updates were related to better estimating motion vectors, and the forecast accuracy was gradually and consistently improved by these updates. Particularly, the threat scores increased the most around 40°S and 40°N. A performance drop in the northern hemisphere winter was also reduced by reducing noise in advection. The time series of ensemble spread showed that an increase in the number of available motion vectors by a system update led to a decrease in the ensemble spread, and vice versa.
著者
AWAZU Taeka OTSUKA Shigenori MIYOSHI Takemasa
出版者
Meteorological Society of Japan
雑誌
気象集誌. 第2輯 (ISSN:00261165)
巻号頁・発行日
pp.2019-066, (Released:2019-09-22)

This paper proposes a new verification metric that can evaluate location errors and shapes of rainfall areas simultaneously: the Pattern Similarity Index (PSI). Pixel-by-pixel verification methods such as the threat score and root mean squared error have difficulties in evaluating location errors and shapes of rainfall areas, and in evaluating small rainfall areas. To address these difficulties, various object-based methods have been developed. However, object-based methods tend to be complicated and computationally expensive. Therefore, PSI adopts a simpler, computationally more efficient algorithm as follows. First, bounding rectangles of individual rainfall areas are computed, and neighboring rectangles are combined so that they are treated as a single precipitation system to mimic the human recognition. Next, shape parameters are computed for each integrated bounding rectangle. For each pair of the observed and forecasted rainfall areas, the location error weighted by the differences of the shape parameters is used as the verification score. If no observed rainfall area with a similar size exists near a forecasted rainfall area, this distance- based score of the forecasted area is set to a large value. The integration method of the bounding rectangle and the precipitation threshold are the only tunable parameters in this method, and we repeat computing the verification score by varying these parameters. The best value is used as the final verification score. Idealized cases showed the ability of PSI to evaluate location errors and differences in the shape parameters. A real case with global precipitation nowcasting showed that the proposed evaluation value increased almost linearly with the forecast time, whereas the threat score and root mean squared error tended to saturate as the forecast time increases, showing a potential advantage of PSI. Comparison with another object-based method revealed the advantage of PSI in its computational efficiency while providing similar verification scores.