報告書番号

T97003

タイトル（和文）

統計気候モデルを用いた地域気候変化予測手法の開発（その１）－わが国の気温場に対する手法の適用妥当性の検討－

タイトル（英文）

DEVELOPMENT OF REGIONAL CLIMATE CHANGE PREDICTION METHOD USING STATISTICAL DOWNSCALING(1) -APPLICATION TO TEMPERATURE FIELD IN JAPAN-

概要 （図表や脚注は「報告書全文」に掲載しております）

全球気候モデルの粗い出力結果を地域スケールに翻訳する目的で構築された統計気候モデルを，日本の地上および上層５高度の月平均気温偏差データ（１３地点，３７年間）に適用し，妥当性を検討した。主成分分析により抽出された地上気温変化パターンのうち，全分散の６～８割を説明する第１主成分スコアと上層気温場からの推定スコアの相関係数は高く，全ての月で１％水準で統計的に有意であった。また第２主成分についても，相関係数は第１主成分の場合より低いものの，ほとんどの月で１％水準で有意であった。正準相関分析を用い，上層１，２主成分と地上第１，２主成分を関連付けた解析では，南部の夏季という例外を除き，個々の地点の地上気温の観測値と推定値の相関係数は１％水準で有意であった。以上の検討結果により，日本周辺の地域における上層気温場と地上気温場の強い相関が明らかになり，統計気候モデルの当該地域への適用の妥当性が示された。

概要 （英文）

STATISTICAL DOWNSCALING METHODS, WHICH WERE DEVELOPED TO INTERPRET THE COARSE RESOLUTION GENERAL CIRCULATION MODEL (GCM) OUTPUT INTO REGIONAL SCALE CLIMATE INFORMATION STATISTICALLY, WERE EVALUATED BY APPLICATION TO THE TEMPERATURE FIELD IN JAPAN. MONTHLY ANOMALY DATA FOR THE SURFACE AND UPPER (5 PRESSURE LEVELS) AIR TEMPERATURES OBSERVED AT 13 STATIONS IN JAPAN FOR 37 YEARS WERE ANALYZED USING MULTIVARIATE ANALYSIS. FIRST, TEMPERATURE CHANGE PATTERNS WERE DERIVED FROM THE DATA SET USING PRINCIPAL COMPONENT ANALYSIS. SECOND, THE 1ST AND 2ND PRINCIPAL COMPONENTS OF THE SURFACE AIR TEMPERATURE WERE ESTIMATED FROM THE 1ST AND 2ND PRINCIPAL COMPONENTS OF THE UPPER AIR TEMPERATURE. WITH REGARD TO THE 1ST COMPONENT OF THE SURFACE AIR TEMPERATURE, WHICH ACCOUNTS FOR 60% TO 80% OF THE DATA VARIANCE, THE CORRELATION COEFFICIENTS BETWEEN THE SCORES CALCULATED FROM OBSERVATION AND ESTIMATION ARE STATISTICALLY SIGNIFICANT (P<0.01) FOR ALL MONTHS. ALTHOUGH THE CORRELATION COEFFICIENTS FOR THE 2ND PRINCIPAL COMPONENT OF THE SURFACE AIR TEMPERATURE ARE SMALLER THAN THOSE FOR THE 1ST PRINCIPAL COMPONENT, THEY ARE STATISTICALLY SIGNIFICANT (P<0.01) FOR MOST MONTHS. THIRD, THE SURFACE AIR TEMPERATURE AT EACH STATION WAS ESTIMATED USING CANONICAL CORRELATION RELATIONSHIP BETWEEN THE 1ST AND 2ND PRINCIPAL COMPONENTS OF THE SURFACE AIR TEMPERATURE AND THOSE OF THE UPPER AIR TEMPERATURE. ALTHOUGH MORE PRINCIPAL COMPONENTS ARE REQUIRED TO REPRODUCE THE SUMMER SURFACE TEMPERATURE FIELD IN THE SOUTHERN REGION, THE CORRELATION COEFFICIENTS BETWEEN OBSERVATION AND ESTIMATION ARE STATISTICALLY SIGNIFICANT (P<0.01) FOR THE OTHER SEASONS AND STATIONS. THEREFORE, THERE IS A STRONG RELATIONSHIP BETWEEN THE SURFACE AND UPPER AIR TEMPERATURE FIELDS OVER JAPAN AND STATISTICAL DOWNSCALING METHODS CAN BE APPLIED TO THE AREA.

報告者

キーワード