透過您的圖書館登入
IP:3.17.150.89
  • 期刊

評估四種聯合克利金法整合雷達和雨量站觀測估計降雨空間分佈的誤差特性

Rain Field Error Characteristics of Four Co-Kriging Strategies Integrating Radar and Gauge Observations

摘要


本研究設計無誤差雨量站和五種不同雷達誤差架構的觀測資料,評估「普通」和「通用」兩類共四種聯合克利金法排除雷達觀測誤差影響、正確估計空間降雨分佈的能力。五種雷達觀測誤差分別是:(1)白噪音加法性誤差(WN),(2)含偏估的空間相關加法性誤差(AE),(3)空間相關乘法性誤差(ME),(4)空間趨勢誤差(TE),(5)包含AE、ME及TE三種誤差的綜合性誤差(AMTE)。兩種普通聯合克利金法,分別是權重平均所有雨量站和估計網格雷達觀測值的「普通聯合克利金法」(OCK),以及再增加各雨量站網格之雷達觀測值的「修正聯合克利金法」(MOCK)。「通用聯合克利金法」(UCK)的最小誤差平方估計式只包括雨量站觀測值,基於共網格雷達觀測和真實降雨量符合線性迴歸關係的假設,導出對於迴歸方程式任意未知截距和斜率都能滿足的不偏估式。「含空間趨勢的通用克利金法(UCKT)」,是UCK法再增加空間趨勢函數不偏估式的結果。評估結果證實,當雷達觀測誤差為白噪音加法性誤差時,四種整合方法都無法免除其影響,OCK法的表現略優於其他三種方法。四種方法中,僅OCK法整合估計不能避免受到AE誤差的影響。MOCK法和UCKT法都可有效避免AE、TE誤差,減輕ME誤差的影響,但權重方式不同。UCK法可以免除AE誤差的影響,降低ME及TE誤差的影響。本研究設計的誤差案例中,MOCK法是四種方法中估計誤差變異數最小者。UCK和UCKT二法的優勢,是不需要估計雷達觀測之間的半變異圖,以及雷達和雨量站觀測之間的半變異圖。

並列摘要


In this study, we use error-free rain gauge data and design five different error structured radar observations to examine the abilities of two ordinary co-kriging techniques and two universal co-kriging techniques to correctly estimate spatial distribution of rainfall. The five radar observation errors are (1) additive white noise error (WN), (2) additive correlative error with bias (AE), (3) multiplicative correlative error (ME), (4) trend error varying with radar range (TE), (5) combined error including AE, ME and TE (AMTE). Ordinary co-kriging (OCK) technique utilizes the linear combination of all rain gauge observations and the radar observation collocated with estimated grid. Modified ordinary co-kriging (MOCK) technique utilizes the radar observations on top of all rain-gauges in addition to the data used by OCK technique. The minimum error variance estimate of universal co-kriging (UCK) utilizes the gauge data only. Based on the collocated true rainfalls and radar observations follows a linear model assumption, the unbiased conditions are derived. UCKT is a UCK technique plus satisfying the spatial trend unbiased condition.Case study results illustrate that when radar error type is WN, all techniques could not avoid its influence. In this case, the estimates of OCK are slightly better than the other three. OCK is the only technique that cannot avoid AE error from going into rainfall rate estimates. Both MOCK and UCKT can effectively prevent AE and TE error from entering the estimates, and reduce the influence of ME error. According to the statistics of the case studies, MOCK had the lowest root mean square error. The major advantage of UCK and UCKT is that it is not necessary to provide the semi-variograms involving radar data.

被引用紀錄


呂玟潔(2018)。雷達降雨應用於農業災害預警之可行性研究〔碩士論文,淡江大學〕。華藝線上圖書館。https://doi.org/10.6846/TKU.2018.00396
黃裕翔(2013)。應用通用共克利金法結合不同雨量站網資料之空間變異推估〔碩士論文,國立臺北科技大學〕。華藝線上圖書館。https://doi.org/10.6841/NTUT.2013.00007
劉承昕(2014)。利用ABLER移流迴歸法估計颱風降雨回波移速之研究〔碩士論文,國立臺灣大學〕。華藝線上圖書館。https://doi.org/10.6342/NTU.2014.02314
鄭曉陽(2012)。受地形影響颱風的風雨關聯診斷和追蹤-模擬分析研究〔碩士論文,國立臺灣大學〕。華藝線上圖書館。https://doi.org/10.6342/NTU.2012.10358
鍾德霖(2005)。利用都卜勒雷達觀測與三維變分法進行颱風風場最佳化〔博士論文,國立臺灣大學〕。華藝線上圖書館。https://doi.org/10.6342/NTU.2005.00451

延伸閱讀