南海海面風(fēng)的動力降尺度模擬
發(fā)布時(shí)間：2018-11-05 11:51
【摘要】：現(xiàn)有的全球大氣觀測資料在時(shí)空分布上存在著非常嚴(yán)重的非均一性問題,而全球大氣模式的時(shí)空分辨率一般較低,難以滿足研究人員對區(qū)域氣候精細(xì)化研究的需求。用全球模式的結(jié)果驅(qū)動適合區(qū)域模擬的有限區(qū)域模式,從而得到時(shí)空分辨率較高,符合區(qū)域氣候特征的氣象要素分布是一個(gè)不錯(cuò)的選擇。目前動力降尺度在氣溫和降水上的應(yīng)用比較廣泛,但是對海面風(fēng)的研究還不夠深入,尤其是南海海面風(fēng)。本文首先通過對2003年南海海面風(fēng)場的模擬探究了中尺度天氣預(yù)報(bào)模式WRF區(qū)域氣候模擬的能力,并比較分析了兩種驅(qū)動方式與多種全球(再)分析資料用于南海海面風(fēng)的降尺度研究,最后選擇質(zhì)量較高的驅(qū)動場和驅(qū)動方式對2001-2008年南海進(jìn)行降尺度,分析降尺度后的海面風(fēng)場相比于驅(qū)動場的增加價(jià)值,并探究譜逼近方法在南海海面風(fēng)降尺度中的作用,主要結(jié)論如下：1)WRF連續(xù)長期積分模擬會導(dǎo)致系統(tǒng)誤差累積,使模擬結(jié)果偏離觀測值,定期更新初始場的方法能夠有效改善長期積分積累的誤差,每個(gè)月風(fēng)向風(fēng)速的均方根誤差(RMSE)和空間相關(guān)系數(shù)(SCC)都得到優(yōu)化。2) ERA-Interim作為模式強(qiáng)迫場降尺度后的結(jié)果優(yōu)于FNL和CFSR,主要體現(xiàn)在ERA-Interim作為強(qiáng)迫場得到的年平均風(fēng)速SCC最大,達(dá)到0.909,年平均風(fēng)向空間RMSE最小,僅17.99°,比表現(xiàn)最差的CFSR改善了8.3%°同時(shí),區(qū)域平均的風(fēng)速風(fēng)向RMSE和CC都表現(xiàn)最優(yōu),相比風(fēng)速表現(xiàn)最差的FNL,風(fēng)速ROSE改善了8.5%,CC提高了9.2%,相比風(fēng)向表現(xiàn)最差的CFSR RMSE改善了13.8%,CC提高了5.8%。3)降尺度能較好緩解再分析資料存在整體風(fēng)速強(qiáng)度過低的問題,同時(shí)一定程度上修正再分析資料對風(fēng)速的空間分布模態(tài)刻畫不夠準(zhǔn)確,出現(xiàn)虛假高值中心或觀測上存在的高值中心沒有刻畫出來的問題,但是值得注意的是,降尺度偶爾會使得原本已刻畫較好的風(fēng)速提高到高于觀測的水平。進(jìn)一步對不同風(fēng)速等級的降尺度效果進(jìn)行檢驗(yàn)發(fā)現(xiàn),降尺度對于高風(fēng)速能產(chǎn)生增加價(jià)值,但是對于低風(fēng)速的刻畫能力不足,會導(dǎo)致低風(fēng)速模擬過高。4)譜逼近在本次試驗(yàn)中影響較小,表現(xiàn)偏中立,其效果不顯著的原因與采用5天更新一次初始場的方式驅(qū)動有關(guān),每次積分時(shí)間過短,初始場的作用還較強(qiáng),譜逼近的作用未能得到體現(xiàn)。5)降尺度結(jié)果顯示,2001-2008年南海大風(fēng)發(fā)生次數(shù)分布存在兩個(gè)極大值中心,分別位于呂宋海峽和越南東南沿海,極大值為125天。南海北部的大風(fēng)日數(shù)高于75天的高值區(qū)在這8年的時(shí)間內(nèi)呈現(xiàn)先減小后增長的趨勢,拐點(diǎn)出現(xiàn)在2004年,從2004年往后,該區(qū)域范圍一直呈增大趨勢,到2008年甚至與南海南部的高值區(qū)連通。南海大風(fēng)頻次表現(xiàn)出明顯的季節(jié)變化特征,冬季大風(fēng)發(fā)生頻次最高,春季大風(fēng)發(fā)生頻次最低,且在2001到2008年這8年時(shí)間,南海冬季大風(fēng)頻次呈增加趨勢,而夏季大風(fēng)頻次呈減小趨勢。
[Abstract]:There is a very serious problem of heterogeneity in the temporal and spatial distribution of the existing global atmospheric observation data, but the spatial and temporal resolution of the global atmospheric model is generally low, so it is difficult to meet the needs of researchers for the fine study of regional climate. It is a good choice to use the results of the global model to drive the finite regional model which is suitable for regional simulation so that the spatial and temporal resolution is higher and the distribution of meteorological elements in accordance with the regional climate characteristics is a good choice. At present, dynamic downscaling is widely used in temperature and precipitation, but the study of sea surface wind is not deep enough, especially in the South China Sea. In this paper, the capability of the mesoscale weather forecast model WRF regional climate simulation is studied by simulating the sea surface wind field in the South China Sea in 2003. Two driving modes and several global (reanalysis) data are compared to study the downscaling of the South China Sea (SCS) wind. Finally, the high quality driving field and the driving mode are selected to downscale the South China Sea from 2001 to 2008. The value of downscaling sea surface wind field compared with driving field is analyzed, and the function of spectral approximation method in South China Sea sea surface wind downscaling is explored. The main conclusions are as follows: 1) WRF continuous long-term integral simulation will lead to the accumulation of system errors. The method that the simulation results deviate from the observed values and the initial field is updated regularly can effectively improve the error of long-term integral accumulation. The root mean square error (RMSE) and spatial correlation coefficient (SCC) of monthly wind direction and wind speed are optimized. 2) ERA-Interim is superior to FNL and CFSR, as the result of model forced field downscaling. The main results are as follows: the annual mean wind speed (SCC) obtained by ERA-Interim as a forced field is the largest (0.909), and the annual mean wind direction spatial RMSE (RMSE) is the smallest, only 17.99 擄, which is 8.3% better than the worst performing CFSR. The regional average wind speed and wind direction RMSE and CC both performed best. Compared with FNL, wind speed ROSE, which had the worst wind speed performance, the regional average wind speed direction increased by 9.2% compared with FNL, wind speed ROSE, and improved by 13.8% compared with CFSR RMSE, which had the worst wind direction performance. CC improved 5.8.3) downscaling can better alleviate the problem that the overall wind speed intensity is too low in the reanalysis data, and at the same time, to some extent, the reanalysis data is not accurate enough to describe the spatial distribution mode of wind speed. The problem of false high value center or observed high value center does not exist, but it is worth noting that the downscaling occasionally makes the wind speed which has already been well described raise to higher than the observed level. Further tests on the downscaling effect of different wind speed grades show that downscaling can add value to high wind speed, but the ability to depict low wind speed is insufficient. The results show that the low wind speed simulation is too high. 4) the effect of spectral approximation is relatively small and neutral in this experiment. The reason why the effect is not significant is driven by the method of updating the initial field once in 5 days, and the time of integration is too short. The effect of the initial field is still strong, and the effect of spectral approximation is not reflected. 5) the downscaling results show that there are two maximum centers in the frequency distribution of strong winds in the South China Sea from 2001 to 2008, located in Luzon Strait and off the southeast coast of Vietnam, respectively. The maximum value is 125 days. The high value area with gale days above 75 days in the northern part of the South China Sea showed a trend of first decreasing and then increasing in the period of 8 years. The inflection point appeared in 2004, and since 2004, the area has been increasing. By 2008 even with the South China Sea south of the high value of the region connected. The frequency of strong wind in the South China Sea shows obvious seasonal variation characteristics, the frequency of strong wind in winter is the highest, and the frequency of strong wind in spring is the lowest, and the frequency of strong wind in winter of the South China Sea increased in the 8 years from 2001 to 2008. However, the frequency of strong wind decreased in summer.
【學(xué)位授予單位】：中國海洋大學(xué)
【學(xué)位級別】：碩士
【學(xué)位授予年份】：2014
【分類號】：P732
本文編號：2312000
[Abstract]:There is a very serious problem of heterogeneity in the temporal and spatial distribution of the existing global atmospheric observation data, but the spatial and temporal resolution of the global atmospheric model is generally low, so it is difficult to meet the needs of researchers for the fine study of regional climate. It is a good choice to use the results of the global model to drive the finite regional model which is suitable for regional simulation so that the spatial and temporal resolution is higher and the distribution of meteorological elements in accordance with the regional climate characteristics is a good choice. At present, dynamic downscaling is widely used in temperature and precipitation, but the study of sea surface wind is not deep enough, especially in the South China Sea. In this paper, the capability of the mesoscale weather forecast model WRF regional climate simulation is studied by simulating the sea surface wind field in the South China Sea in 2003. Two driving modes and several global (reanalysis) data are compared to study the downscaling of the South China Sea (SCS) wind. Finally, the high quality driving field and the driving mode are selected to downscale the South China Sea from 2001 to 2008. The value of downscaling sea surface wind field compared with driving field is analyzed, and the function of spectral approximation method in South China Sea sea surface wind downscaling is explored. The main conclusions are as follows: 1) WRF continuous long-term integral simulation will lead to the accumulation of system errors. The method that the simulation results deviate from the observed values and the initial field is updated regularly can effectively improve the error of long-term integral accumulation. The root mean square error (RMSE) and spatial correlation coefficient (SCC) of monthly wind direction and wind speed are optimized. 2) ERA-Interim is superior to FNL and CFSR, as the result of model forced field downscaling. The main results are as follows: the annual mean wind speed (SCC) obtained by ERA-Interim as a forced field is the largest (0.909), and the annual mean wind direction spatial RMSE (RMSE) is the smallest, only 17.99 擄, which is 8.3% better than the worst performing CFSR. The regional average wind speed and wind direction RMSE and CC both performed best. Compared with FNL, wind speed ROSE, which had the worst wind speed performance, the regional average wind speed direction increased by 9.2% compared with FNL, wind speed ROSE, and improved by 13.8% compared with CFSR RMSE, which had the worst wind direction performance. CC improved 5.8.3) downscaling can better alleviate the problem that the overall wind speed intensity is too low in the reanalysis data, and at the same time, to some extent, the reanalysis data is not accurate enough to describe the spatial distribution mode of wind speed. The problem of false high value center or observed high value center does not exist, but it is worth noting that the downscaling occasionally makes the wind speed which has already been well described raise to higher than the observed level. Further tests on the downscaling effect of different wind speed grades show that downscaling can add value to high wind speed, but the ability to depict low wind speed is insufficient. The results show that the low wind speed simulation is too high. 4) the effect of spectral approximation is relatively small and neutral in this experiment. The reason why the effect is not significant is driven by the method of updating the initial field once in 5 days, and the time of integration is too short. The effect of the initial field is still strong, and the effect of spectral approximation is not reflected. 5) the downscaling results show that there are two maximum centers in the frequency distribution of strong winds in the South China Sea from 2001 to 2008, located in Luzon Strait and off the southeast coast of Vietnam, respectively. The maximum value is 125 days. The high value area with gale days above 75 days in the northern part of the South China Sea showed a trend of first decreasing and then increasing in the period of 8 years. The inflection point appeared in 2004, and since 2004, the area has been increasing. By 2008 even with the South China Sea south of the high value of the region connected. The frequency of strong wind in the South China Sea shows obvious seasonal variation characteristics, the frequency of strong wind in winter is the highest, and the frequency of strong wind in spring is the lowest, and the frequency of strong wind in winter of the South China Sea increased in the 8 years from 2001 to 2008. However, the frequency of strong wind decreased in summer.
【學(xué)位授予單位】：中國海洋大學(xué)
【學(xué)位級別】：碩士
【學(xué)位授予年份】：2014
【分類號】：P732
【相似文獻(xiàn)】
相關(guān)碩士學(xué)位論文 前1條
1 張萌;南海海面風(fēng)的動力降尺度模擬[D];中國海洋大學(xué);2014年
，本文編號：2312000
本文鏈接：http://sikaile.net/kejilunwen/haiyang/2312000.html
最近更新
教材專著
