Downloads: 159
Files in This Item:
File | Description | Size | Format | |
---|---|---|---|---|
a55b0p04.pdf | 3.3 MB | Adobe PDF | View/Open |
Title: | 複数の再解析データによる気候値の空間分布再現性とWCRP-CMIP3マルチ気候モデルにみる気候変動予測の不確実性 |
Other Titles: | Spatial reproducibility of the climate values by multiple reanalysis dataset and uncertainty of climate change projection using the CMIP3 Model Output |
Authors: | 辰己, 賢一 山敷, 庸亮 寶, 馨 https://orcid.org/0000-0001-5454-2989 (unconfirmed) |
Author's alias: | TATSUMI, Kenichi YAMASHIKI, Yosuke TAKARA, Kaoru |
Keywords: | 地上降水量 再解析データ グリッドデータ 再現性評価 observational precipitation re-analysis data gridded data validity of precipitation |
Issue Date: | 30-Sep-2012 |
Publisher: | 京都大学防災研究所 |
Journal title: | 京都大学防災研究所年報. B |
Volume: | 55 |
Issue: | B |
Start page: | 21 |
End page: | 30 |
Abstract: | 数値モデルから得られる出力値の不確実性には主にモデル自体が有する不確実性と入力データの不確実性が含まれることから,実際にモデルを用いた研究を行う際は,より精密な入力データの使用が望まれる。本研究では,ERA-40, CRU TS2.1, JRA-25, GPCP, CMAP, CMIP3の各種グリッドデータの降水量に着目し,世界の608地点における地上降水量値を基準とした統計的誤差解析を1979年から1999年を対象に行い,再現性を評価した。その結果,CRU TS2.1の降水量の値は他のデータソースと比べて,世界の大部分の領域において最も平均絶対誤差が小さく,かつデータのばらつきが小さいことがわかった。一方,再解析データは特に対流性降雨が卓越する熱帯域において誤差が大きく,再現性に課題があることがわかった。 Outputs obtained from the numerical model have the uncertainty, which includes 1) model uncertainty, 2) uncertainty of input data. Therefore, when conducting research using the numerical model, we need to focus attention on accuracy and characteristics of input data in particular. In this study, we assess the correspondence between precipitation products from ERA-40, CRU TS2.1, JRA-25, GPCP, CMAP and CMIP3 with adjusted observation precipitation from global (608 stations in total) for 1979-1999. In general, we conclude that CRU TS2.1 agrees more closely with observation precipitation than other datasets. Moreover, the mean absolute error of the precipitation differences (CRU TS2.1 - observation precipitation) and Standard deviation of the biases for CRU TS2.1 is the smallest in most regions. On the other hand, atmospheric reanalysis (ERA-40, JRA-25), especially a tropical region, have the large error, and it turned out that there is a problem in reproducibility. Taking into consideration the specific characteristics in datasets on each other, we should conduct research. |
URI: | http://hdl.handle.net/2433/161874 |
Related Link: | http://www.dpri.kyoto-u.ac.jp/nenpo/nenpo.html |
Appears in Collections: | Vol.55 B |
Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.