download

北日本における4月と8月気温の
強い相関関係とその時間変動
(3:1998年は climate-shiftか?)
(独)農業・食品産業技術総合研究機構
東北農業研究センター
菅野洋光
目的
• 北日本の農業にとって、作付時期の春と生長・
稔実時期の夏の気温は非常に重要である。しか
しながら、近年、夏季の気温変動が大きく、冷夏
や暑夏が頻発している。冷害や高温による米の
品質低下が発生し、気温の変動要因の把握が
重要である。
• 特に2010年は春の気温が低く、高温の夏とあわ
せて極端な季節変化となった。そのような気温
変動について、季節変動の規則性の観点から統
計的に検討する。
データと方法
• 北日本の平均気温偏差は、東北と北海道の
気象官署月平均気温偏差を平均して求めた。
• 平年値は全期間に1981年~2010年平年値を
用いている。
• 客観解析データはJRA25を、統計解析にはR
および気象庁のITACSver.4を用いた。
• 各季節、月の平均値でそれぞれの相関関係
を検討し、特に関係の明瞭な4月と8月の事例
を中心に客観解析データを用いて解析した。
Fig. 1. Map of northern Japan, which consists of Hokkaido and the Tohoku
region.
Table 1. Correlation coefficients between each possible pair of monthly mean
temperatures in the last 13 years (from 1999 to 2011) in northern Japan.
Shaded boxes indicate statistical significance at the 5% level or better.
Fig. 2. Thirteen-year running correlation coefficients between April and June,
July, August, and September monthly mean temperatures in northern Japan
from 1950 to 2011. Values are reported for the year that is the midpoint of the
13-year period. Broken horizontal lines indicate a significance level of 5%
where the correlations exceeded an absolute value of approximately +0.6.
Fig. 3. Time series of the separate monthly mean temperature deviations in April
and August in northern Japan, and the 13-year running correlation coefficients
between these two months during 1950–2011. Only those few correlation
coefficients greater than 0.6 or less than –0.6 (horizontal dashed lines) were
significant at the 5% level.
Fig. 4. Correlation between northern
Japan’s temperature deviation in August
and the 200-hPa u-component of wind in
the same August from 1998 to 2011, and
the normal u-component wind in August.
Fig. 5. Correlation between northern Japan’s
temperature deviation in August and the 850hPa air temperature in the same August from
1998 to 2011.
Fig. 6. Correlation between northern
Japan’s temperature deviation in August
and the 200-hPa u-component of wind in
April, from 1998 to 2011 (dashed
contours and shaded areas), overlaid on
the normal u-component distribution of
wind in April (thin solid contours).
Fig. 7. Correlation between northern Japan’s
temperature deviation in August and the
850-hPa air temperature in the preceding
April from 1998 to 2011.
Fig. 8. Correlation coefficients between the surface air temperature deviation
in August in northern Japan and the 200-hPa geopotential height in the
preceding April, from 1998 to 2011. Shading indicates correlations significant
at the 5% level. Dotted lines indicate the EOF analyzed area from 20°N to
70°N latitude and from 0° to 140°W longitude.
Table 2. Correlation coefficients between each PCs in April and August in last 13
years (from 1999 to 2011).Numerals in parenthesis indicate proportion of variance
(%).Shaded boxes indicate statistical significance at the 5% level or better.
Fig. 9. Time series of PC2 in April and PC3 in August at the 200-hPa level. The
solid white line indicates the 13-year running correlation coefficients
between PC2 in April and PC3 in August, and the dashed white line indicates
the 13-year running correlation coefficients between temperatures in April
and August in northern Japan.
(a)
(b)
Fig. 10. (a) EOF2 in April, and (b) EOF3 in August on the 200-hPa height field from
1979 to 2011. Unit is m.
(a)
(b)
Fig. 11. Regression of (a) the PC2 time series of the 200-hPa height field in April
and (b) the PC3 time series of the 200-hPa height field in August against the 200hPa u-component field.
(a)
(b)
Fig.12. First SVD mode of FMA (February- April) 200 hPa height and SST for 1979-2011.
a) Heterogeneous correlation map for 200hPa height of mode 1,
b) Heterogeneous correlation map for SST of mode 1.
(c)
(d)
Fig.12. Second SVD mode of FMA (February- April) 200 hPa height and SST for 1979-2011.
c) Heterogeneous correlation map for 200hPa height of mode 2,
d) Heterogeneous correlation map for SST of mode 2.
(e)
Fig.12. First and second SVD modes of FMA (February- April) 200 hPa height and SST
for 1979-2011.
e) Time series of the normalized expansion coefficients for 200hPa height (white solid
curve; mode 1, black solid curve; mode 2) and SST (white dashed curve; mode 1, black
dashed curve; mode 2). Thin solid and broken lines indicate the average from 1979 to
1997 and from 1998 to 2011 of mode2.
(a)
(b)
Fig.13. First SVD mode of JJA (June-August) 200 hPa height and SST for 1979-2011.
a) Heterogeneous correlation map for 200hPa height of mode 1,
b) Heterogeneous correlation map for SST of mode 1.
(c)
(d)
Fig.13. Second SVD modes of JJA (June-August) 200 hPa height and SST for 1979-2011.
c) Heterogeneous correlation map for 200hPa height of mode 2,
d) Heterogeneous correlation map for SST of mode 2.
(e)
Fig.13. First and second SVD modes of JJA (June-August) 200 hPa height and SST for
1979-2011.
e) Time series of the normalized expansion coefficients for 200hPa height (black solid
curve; mode 1, white solid curve; mode 2) and SST (black dashed curve; mode 1, white
dashed curve; mode 2). ). Thin solid and broken lines indicate the average from 1979 to
1997 and from 1998 to 2011 of mode1.
(a)
(b)
Fig. 15. Tine-latitude cross section along 140°E for monthly mean 200 hPa
u-component from 1998 to 2011. Contours indicate wind speed (m/s) and
shadings indicate standard deviation (m/s). a)1979-1997, b)1998-2011.
結論
北日本4月8月気温負の相関関係は、
200hPa高度の4月第2主成分、8月第3主成分
の出現およびSST変動で説明できた。さらに
200hPa高度とSSTのSVD解析結果より、1998
年は、北日本における気温変動のみならず、
大規模な高度場・SST変動のレジームシフト
期間として把握できると思われる。
(a)
(b)
Fig. 14. Regression of the 200-hPa PC3 in August against the SST anomaly in (a) April
and (b) August.