報告書番号

T91043

タイトル（和文）

わが国における気候変化の統計解析－多変量解析法を用いた地域気温のトレンド解析－

タイトル（英文）

CLIMATE CHANGE IN JAPAN -MULTIVARIATE ANALYSIS OF REGIONAL TEMPERATURE TRENDS-

概要 （図表や脚注は「報告書全文」に掲載しております）

気温アノマリーパターン（平均偏差）の時間・空間変動の体系的な解析結果をもとに，わが国における地域的な気温トレンドの解析を行った。抽出された３つの基本アノマリーパターン（主成分）の卓越指数には季節的に異なったトレンドが見られたので，このトレンドをもとに主成分合成法を用いて各地域毎，月毎に気温の有効トレンドを求めた。年平均気温でみると，過去７０年の間に日本全域とも気温の上昇トレンドが見られるが地域差があり，東北日本で０．５～１．０度／１００年，中部～西日本では１．２～１．４度／１００年（いずれも都市化の影響を含む）程度である。また季節的に見れば，いずれの地域でも気温の正のトレンドは冬季～春季（１２月～６月）に大きく，夏季～秋季（７月～１１月）に小さい。トレンドの地域差は夏季～秋季に大きく，気温の南北傾度は拡大している。

概要 （英文）

ANOMALIES (DEVIATIONS FORM MEAN) IN TEMPERATURE DISTRIBUTION PATTERNS OF MONTHLY MEAN TEMPERATURE DATA FOR THE PAST90 YEARS FROM 53 METEOROLOGICAL STATIONS IN JAPAN WERE EXAMINED USING MULTIVARIATE ANALYSIS. FUNDAMENTAL ANOMALY PATTERNS WERE DERIVED USING PRINCIPAL COMPONENT ANALYSIS. THE FIRST AND SECOND ANOMALY PATTERNS SHOW RESPECTIVE POSITIVE (NEGATIVE) ANOMALY PATTERNS FOR JAPAN AS A WHOLE AND AN ANOMALY IN THE NORTH-SOUTH TEMPERATURE GRADIENT. BOTH ANOMALY PATTERNS HAD POSITIVE TRENDS(SIGNIFICANT AT 1% LEVEL)FOR THE TIME SERIES OF ANNUAL MEAN Z-COMPONENT SCORES. THE TRENDS ARE MORE MARKED DURING THE LAST 70 YEARS. THE REGIONAL MEAN TEMPERATURE TRENDS WERE CALCULATED IN SEVEN REGIONS FROM THE TREND VALUES OF PRINCIPAL Z-COMPONENT SCORES. THE TREND VALUES BASED ON ANNUAL,MEAN Z-COMPONENT SCORES WERE 0.5 1.0DEGREE/100 YEARS IN NORTHEAST JAPAN AND 1.0 1.4DEGREE/100 YEARS IN SOUTHWEST JAPAN DURING THE 70-YEAR PERIOD. TEMPERATURE TRENDS IN WINTER AND SPRING WERE 1.5 2.0DEGREE/100 YEARS FOR THE PAST 70 YEARS WITH THE MAXIMUM VALUE IN MAY;THE REGIONAL DIFFERENCES WERE COMPARATIVELY SMALL. TEMPERATURE TRENDS IN SUMMER AND AUTUMN WERE -0.5 1.5DEGREE/100YEARS FOR THE PAST 70 YEARS;THE REGIONAL DIFFERENCES WERE COMPARATIVELY LARGE. THESE SEASONAL DIFFERENCES REFLECT THE PREDOMINANT POSITIVE TREND OF THE FIRST ANOMALY IN WINTER AND SPRING AND THE POSITIVE TREND OF THE SECOND ANOMALY IN SUMMER AND AUTUMN. THE ANNUAL DATA FOR THE 90 YEARS WAS CLASSIFIED INTO SIX GROUPS USING CLUSTER ANALYSIS DEPENDING ON (1)THREE Z-COMPONENT SCORES FOR EACH OF THE 12 MONTHS,AND ON (2)THE ANNUAL MEAN OF THREE Z-COMPONENT SCORES (I.E. DEPENDING ON THE SIMILARITY OF ANOMALY PATTERNS).

報告者

キーワード