跳到主要內容

臺灣博碩士論文加值系統

(44.210.83.132) 您好!臺灣時間:2024/05/29 14:07
字體大小: 字級放大   字級縮小   預設字形  
回查詢結果 :::

詳目顯示

我願授權國圖
: 
twitterline
研究生:林尚農
研究生(外文):LIN SHANG-NONG
論文名稱:評估西北太平洋季風指數與臺北及高雄夏季氣溫之關聯性
論文名稱(外文):Assessment on Association between Western North Pacific Monsoon Index and Air Temperatures during summers in Taipei and Kaohsiung Cities
指導教授:王玉純王玉純引用關係
指導教授(外文):WANG, YU-CHUN
口試委員:江謝令涵林旭信
口試委員(外文):CHIANG HSIEH, LIN-HANLIN, SHIU-SHIN
口試日期:2022-07-21
學位類別:碩士
校院名稱:中原大學
系所名稱:環境工程學系
學門:工程學門
學類:環境工程學類
論文種類:學術論文
論文出版年:2022
畢業學年度:110
語文別:中文
論文頁數:90
中文關鍵詞:氣象測站都市氣溫氣象因子季風指數主成分分析隨機森林XGBoost
外文關鍵詞:meteorological stationsurban air temperaturemeteorological factorsmonsoon indexprincipal component analysisrandom forestXGBoost
DOI:10.6840/cycu202201456
相關次數:
  • 被引用被引用:0
  • 點閱點閱:83
  • 評分評分:
  • 下載下載:0
  • 收藏至我的研究室書目清單書目收藏:0
大氣環境溫度變化對於人體健康是關鍵的指標,近年隨都市化程度發展,探討季風盛行區域下都市中溫度為重要研究課題之一。本研究為探討西北太平洋季風 (western north pacific monsoon, WNPM)作用下,臺北與高雄測站的氣象因子之變化,並以機器學習建模進行溫度預測評估。
本研究蒐集並彙整交通部中央氣象局局屬氣象測站 (Central Weather Bureau, MOTC) 以及國際太平洋研究中心 (International Pacific Research Center, IPRC) 1980年至2015年間的局屬氣象測站資料,以及西北太平洋季風指數資料,使用氣象參數包含臺北站 (編號:466920) 及高雄站 (編號:467440) 的平均溫度、最高溫度、最低溫度、日較差溫度、相對濕度、降雨量、降雨時數和季風指數。首先進行資料彙整與勘誤,接續透過敘述性統計 (descriptive statistics) 了解氣象參數基本特性與長期趨勢,再以主成分分析 (principal component analysis, PCA) 將多氣象變數之間的關聯性聚合成一個特徵特性,以觀察數據之相關性。建立1980年至2015年氣象因子與西北太平洋季風指數夏季 (6月至9月) 的歷史資料非延遲天數的A (自變數:每日平均溫)、B (自變數:每日平均溫與每日西北太平洋季風指數)、C (自變數:每日西北太平洋季風指數與每日其他氣象因子)、D (每日其他氣象因子) 組合之模型並進行成效評估;建立1980年至2015年氣象因子與西北太平洋季風指數夏季的歷史資料延遲一天至三十天 (延遲為兩個相關事件或現象在發生時的時間點有所不同,如前置時間的概念,夏季延遲約為25-35天,因此本研究設定最大延遲天數為30天) 的E (自變數:每日平均溫)、F (自變數:每日最高溫與最低溫)、G (自變數:每日西北太平洋季風指數) 組合之模型並進行成效評估;為了更深入了解季風指數的影響,從A、B、C、D之組合篩選出季風指數具有最高變數重要性 (對此模型貢獻度最大的參數) 之模型並進行成效評估。A至G組合輸出的參數皆為平均氣溫。先將氣象因子與西北太平洋季風指數依照時間排序分為訓練資料與測試資料 (以資料總量之7:3分配),並採用隨機森林 (random forest, RF) 與 XBGoost (extreme gradient boosting)模擬氣象資料研究的機器學習算法,隨機森林與XGBoost皆使用迴歸作為預測進行建模,並調控機器學習中的最大深度,使機器學習找出最佳模型與最高變數重要性之參數。最後,藉由平均絕對誤差(mean absolute error, MAE)、均方誤差(mean square error, MSE)、平均絕對百分比誤差(mean absolute percentage error, MAPE) 和對稱性平均絕對百分比誤差 (symmetric mean absolute percentage error, SMAPE) 評估隨機森林與XGBoost的成效。
1980至2015年氣象參數敘述性統計結果顯示,整體溫度呈現上升之趨勢,測站之最低溫度中上升趨勢最為顯著的測站為高雄測站。臺北與高雄測站的相對濕度皆呈現下降趨勢,此結果是因為夜晚最低溫度的上升導致相對濕度的下降。透過主成分分析結果,高雄測站其平均溫度、最高溫度、最低溫度皆與季風指數呈現正相關,而臺北其平均溫度、最高溫度、最低溫度皆與季風指數呈現負相關,由此可知在臺北與高雄測站季風指數對溫度變化影響較其他氣象因子大。

使用隨機森林迴歸進行預測時,臺北與高雄測站從A、B、C、D之組合篩選出季風指數具有最高變數重要性之組合位於C,最佳模型之參數依序為季風指數、日較差、降雨量最後是降雨時數;臺北測站非延遲天數預測在C與D組合成效較好,延遲天數預測則在E組合較好;高雄測站非延遲天數預測在A與B組合成效較好,延遲天數預測則在E組合較好。
使用XGBoost迴歸進行預測時,臺北與高雄測站從A、B、C、D之組合篩選出季風指數具有最高變數重要性之組合位於C,最佳模型之參數依序為季風指數、日較差、降雨量最後是降雨時數;臺北測站非延遲天數預測在A與B組合成效較好,延遲天數預測則在E組合較好;高雄測站非延遲天數預測在A與B組合成效較好,延遲天數預測則在E組合較好。
本研究透過敘述性統計分析結果發現近年來平均溫度以及最低溫度呈上升趨勢,而相對濕度則呈下降趨勢,此結果代表都市高溫伴隨都市化程度的上升而有更顯著的提升;主成分分析結果發現溫度相關因子與季風指數呈現正相關,並在夏季時兩者的關聯會更強烈,此結果代表西北太平洋季風槽加深會對臺灣都市的溫度上升產生影響;機器學習法評估A、B、C與D組合結果發現,溫度相關因子皆有最高變數重要性,此結果代表溫度預測溫度在非延遲天數預測上具有更好的成效,且XGBoost的成效評估相較隨機森林來的更加精準;機器學習法評估E、F與G組合結果發現,雖然E與F組合的精準度較高但與G組合相比不超過2%,而又因E、F組合的延遲天數 (皆為一天) 較低,G組合的延遲天數 (十二天以上) 較高,此結果代表在應用於提前預警夏季高溫之情況下,季風指數可以獲得比溫度相關因子更前期的成效。建議後續研究可新增2015年後的氣象資料以及納入其他類型的氣象資料進行預測,並可使用延遲天數三十天以上的參數進行預測以觀測,亦可使用不同地區的季風指數資料,或是使用更適合研究地區之機器學習算法以更貼近研究地區特性。

With increasing urbanization and economic developments in Taiwan, the issue of the urban high temperature is gradually being emphasized in recent years. The study of urban temperature in the monsoon region is one of the most important issues nowadays. The objective of this study is to investigate the changes of meteorological factors in Taipei and Kaohsiung stations under the western north pacific monsoon (WNPM), and project the long-term air temperature (AT) using machine learning modelings.
This study collected and compiled data from the Central Weather Bureau (MOTC) and the International Pacific Research Center (IPRC) from 1980 to 2015, as well as the western north pacific monsoon index (WNPMI). The meteorological parameters used, include mean AT, maximum AT, minimum AT, diurnal AT range, relative humidity, precipitation, rain hours recording in the Taipei (No. 466920) and Kaohsiung (No. 467440) stations, and daily monsoon index. The data were firstly compiled and harmonized, followed by descriptive statistics was used to observe the basic characteristics and long-term trends of meteorological parameters. Further, principal component analysis was used to aggregate the correlations among multiple meteorological variables into a characteristic feature to observe the correlation of the data. Models A (independent variable: daily mean AT), B (independent variable: daily mean AT and daily WNPMI), C (independent variable: daily WNPMI and daily other meteorological factors), and D (daily other meteorological factors) were evaluated for the non-lagged days using the historical data of meteorological factors and WNPMI during summers (June to September) from 1980 to 2015. And models E (independent variable: daily mean AT), F (independent variable: daily maximum and minimum AT), and G (independent variable: daily WNPMI) for the period 1980 to 2015 with a lagged of one day to 30 days (the lagged is the difference in time point between two related events or phenomena, such as the concept of leading time, the lagged in summer is about 25-35 days, so the maximum lagged is set to 30 days in this study) were developed and evaluated for AT projection effectiveness as well. To better understand the impact of the monsoon index, the models with the highest importance of variables (The parameters majorly contributed the variance of models) were selected from the combinations of A, B, C, and D and evaluated for their effectiveness. The output parameters of combinations A to G are the mean AT. The temporal meteorological factors and the WNPMI were divided into training data and testing data (7:3 distribution of the total data), and the machine learning algorithms of random forest (RF) and extreme gradient boosting (XBGoost) were used to simulate the meteorological data. Both RF and XGBoost used regression as a prediction for modeling and modulating the maximum depth in machine learning to identify the best model setting and the parameters with the highest variable importance. Finally, the mean absolute error (MAE), mean square error (MSE), mean absolute percentage error (MAPE), and symmetric mean absolute percentage error (SMAPE) was used to evaluate the effectiveness of RF and XGBoost.
The descriptive statistics of meteorological parameters from 1980 to 2015 showed an increasing trends of overall AT variables with the most significant increasing trend for daily minimum AT in Kaohsiung station. The relative humidity of both Taipei and Kaohsiung stations showed a decreasing trend that resulted from the increase in nighttime minimum AT. The results of principal component analysis showed that, the mean, maximum, and minimum ATs of Kaohsiung station were all correlated with the WNPMI, while the mean, maximum, and minimum ATs of Taipei were all negatively correlated with the WNPMI indicating the monsoon index has a greater influence on the AT change than on other weather variables in Taipei and Kaohsiung stations.
Assessing the projection performance using RF regression, the Model C was identified the best model with the highest variable importance from the combinations of Modles A, B, C, and D in the Taipei and Kaohsiung stations, and the parameters of the best model were in the order of WNPMI, diurnal AT change, precipitation, and rain hours. The prediction of non-lagged days in Taipei station was better in Models C and D, while the prediction of lagged days was better in Model E. The prediction of non-lagged days in Kaohsiung station was better in Models A and B, while the prediction of delayed days was better in Model E.
Assessing the projection performance using XGBoost regression, the Model C was identified the best model with the highest variable importance from the combinations of Modles A, B, C, and D in the Taipei and Kaohsiung stations, and the parameters of the best model were in the order of WNPMI, diurnal AT change, precipitation, and rain hours. The prediction of non-lagged days in Taipei station was better in Models A and B, while the prediction of lagged days was better in Model E. The prediction of non-lagged days in Kaohsiung station was better in Models A and B, while the prediction of delayed days was better in Model E.
This study summarized that descriptive statistical analysis revealed that the mean AT and minimum AT have been increasing in recent years, while the relative humidity has been decreasing that means that the urban high AT has been increasing more significantly with the increase of urbanization. The principal component analysis showed that the AT correlation factor and WNPMI are correlated and the correlations are stronger in summers that means that The deepening of the western north pacific monsoon trough had an impact on AT rise in Taiwan's cities. The machine learning method evaluated the results of Models A, B, C, and D. It was found that the AT-related factors all had the highest variable importance that means that a better AT prediction in non-lagged day’s models is to use ATs as inpact parameter. Moreover, this study found evaluation effectiveness of XGBoost was more accurate in compare to RF. This study found the accuracy of the Models E and F are higher than Model G with a difference of predicting accuracy less than 2% , but the best lagged day setting is 1 day for Models E and F and more than 12 days for Model G. This result implying the WNPMI can be more effective than the AT-related factors in operating the warning system for summer high ATs. This study suggests that subsequent studies can extend meteorological data after 2015 and incorporate other types of meteorological data for AT forecasting, and evaluate the performance of weather parameters with lags more than 30 days for AT forecasting. Futhermore, assessment on various monsoon index data from different regions, or use machine learning algorithms that are more suitable for the study area is recommended.


目次
摘要 I
Abstract III
目次 V
圖目次 VII
表目次 VIII
中英對照表 IX
第一章、緒論 1
1.1 研究緣起 1
1.2 研究目的 2
第二章 、文獻回顧 3
2.1 全球季風氣候 3
2.1.1 季風區定義 3
2.2 季風指數 3
2.2.1 西北太平洋季風 3
2.2.2 西北太平洋季風指數 4
2.2.3 熱帶氣旋降雨與季節性季風降雨 4
2.2.4 其他季風指數 6
2.3 都市高溫之成因與影響 9
2.3.1 季風與都市高溫 9
2.3.2 臺灣高溫研究彙整 10
2.3.3 都市高溫與健康 11
第三章、研究材料與方法 12
3.1 研究架構 12
3.2 研究材料 14
3.2.1 中央氣象局地面局屬氣象站觀測資料 14
3.2.2 氣象觀測資料之可分析筆數篩選 16
3.2.3 國際太平洋研究中心西北太平洋季風指數 16
3.3 研究方法 18
3.3.1 主成分分析 (principal component analysis, PCA) 18
3.3.2 XGBoost (extreme gradient boosting) 21
3.3.3 隨機森林(random forest, RF) 25
3.3.4 機器學習成效評估指標 27
3.4 分析步驟 29
3.4.1 中央氣象局地面局屬氣象站觀測資料整理與分析 29
3.4.2 國際太平洋研究中心季風指數整理 29
3.4.3 主成分分析設定 30
3.4.4 XGBoost流程 30
3.4.5 隨機森林流程 31
3.4.6 機器學習成效評估 31
3.4.7 模型建立預測溫度組合成效評估 32
3.4.8 分析軟體 33
第四章、結果與討論 34
4.1 研究地區氣象與季風指數資料概況 34
4.1.1 研究地區氣象資料概況 34
4.1.2 研究地區季風指數概況 35
4.1.3 小結與討論 40
4.2 氣象因子與季風指數主成分分析 41
4.2.1 氣象因子與季風指數標準化結果 41
4.2.2 氣象因子與季風指數主成分分析結果 42
4.2.3 小結與討論 45
4.3 氣象因子與季風指數機器學習成效評估 46
4.3.1 隨機森林參數設定 46
4.3.2 XGBoost參數設定 47
4.3.3 臺灣臺北與高雄測站氣象因子對季風指數機器學習成效評估 49
4.3.4 小結與討論 53
4.4 氣象因子與季風指數延遲天數機器學習成效評估 54
4.4.1 臺北測站氣象因子與季風指數延遲天數機器學習成效評估 62
4.4.2 高雄測站氣象因子與季風指數延遲天數機器學習成效評估 63
4.4.3 臺北與高雄測站季風指數延遲天數決策樹結構 64
4.4.4 小結與討論 66
4.5 研究限制 66
第五章、結論與建議 67
5.1 結論 67
5.1.1 主成分分析結果 67
5.1.2 隨機森林成效評估 67
5.1.3 XGBoost成效評估 67
5.1.4 實際應用面 68
5.2 建議 69
參考文獻 70
口試委員建議及回覆 74

圖目次
圖2-1 西北太平洋季風指數北部與南部地區涵蓋範圍[10] 8
圖2-2 (A) 1948年至1997年間在850-HPA下西北太平洋-東亞夏季季風區領先多變量的EOF模式 (B) 時間係數[23] 9
圖3-1 研究架構圖 14
圖3-2 本研究選用之氣象測站位置圖 16
圖3-3 隨機森林示意圖 26
圖3-4 XGBoost流程圖 31
圖3-5 隨機森林流程圖 32
圖4-1 中央氣象局局署臺北測站1980年至2015年每日平均溫度箱型圖 37
圖4-2 中央氣象局局署高雄測站1980年至2015年每日平均溫度箱型圖 37
圖4-3 臺灣臺北與高雄測站1980年至2015年間每年平均最低溫度趨勢圖 39
圖4-4 臺灣臺北與高雄測站1980年至2015年間每年平均相對濕度趨勢圖 39
圖4-5 臺灣1980年至2015年每日平均西北太平洋季風指數 40
圖4-6 臺灣1980年至2015年每月平均西北太平洋季風指數 40
圖4-7 臺灣1980年至2015年每年平均西北太平洋季風指數 41
圖4-8 臺灣1980年至2015年每日平均西北太平洋季風指數箱型圖 41
圖4-9 臺灣臺北與高雄測站第一與第二主成分結果相較顯著表徵圖 45
圖4-10 臺灣地區臺北與高雄測站氣象因子與季風指數主成分分析散佈圖 45
圖4-11 臺灣臺北與高雄測站氣象因子對季風指數延遲天數資料切割圖 56
圖4-12 隨機森林臺北測站延遲30天E組合最佳模型變數重要程度圖 58
圖4-13 隨機森林高雄測站延遲30天E組合最佳模型變數重要程度圖 59
圖4-14 隨機森林臺北測站延遲30天F組合最佳模型變數重要程度圖 60
圖4-15 隨機森林高雄測站延遲30天F組合最佳模型變數重要程度圖 61
圖4-16 隨機森林臺北測站延遲30天G組合最佳模型變數重要程度圖 62
圖4-17 隨機森林高雄測站延遲30天G組合最佳模型變數重要程度圖 63
圖4-18 XGBoost臺北測站延遲30天E組合最佳模型變數重要程度圖 64
圖4-19 XGBoost高雄測站延遲30天E組合最佳模型變數重要程度圖 65
圖4-20 XGBoost臺北測站延遲30天F組合最佳模型變數重要程度圖 66
圖4-21 XGBoost高雄測站延遲30天F組合最佳模型變數重要程度圖 67
圖4-22 XGBoost臺北測站延遲30天G組合最佳模型變數重要程度圖 68
圖4-23 XGBoost高雄測站延遲30天G組合最佳模型變數重要程度圖 69
圖4-24 隨機森林臺北測站G組合最佳模型決策樹結構 72
圖4-25 隨機森林高雄測站G組合最佳模型決策樹結構 72
圖4-26 XGBoost臺北測站G組合最佳模型決策樹結構 73
圖4-27 XGBoost高雄測站G組合最佳模型決策樹結構 73

表目次
表2-1 應用於西太平洋季風指數之彙整 6
表2-2 近10年臺灣13站年平均氣溫比較表(單位:攝氏度)[37] 11
表3-1 本研究選用之氣象測站資訊 15
表3-2 臺北測站於1980年至2015年氣象與季風指數觀測資料樣本數概況 18
表3-3 高雄測站於1980年至2015年氣象與季風指數觀測資料樣本數概況 18
表3-4 臺北測站主成分分析結果示意表格 21
表3-5 1980年至2015年非延遲天數隨機森林與XGBoost模型之組合 33
表3-6 1980年至2015年延遲天數隨機森林與XGBoost模型之組合 33
表4-1 臺灣地面局屬臺北與高雄測站1980年至2015年間氣象因子敘述性統計表 36
表4-2 臺灣地區1980年至2015年間西北太平洋季風指數敘述性統計表 36
表4-3 中央氣象局局屬臺灣地區臺北與高雄測站標準化敘述性統計表 43
表4-4 中央氣象局局屬臺灣地區臺北與高雄測站標準化敘述性統計表 (續) 44
表4-5 臺灣地區臺北與高雄測站氣象因子與季風指數主成分分析結果 44
表4-6 臺灣臺北與高雄測站氣象因子對季風指數隨機森林算法成效評估彙整 51
表4-7 臺灣臺北與高雄測站氣象因子對季風指數XGBoost算法成效評估彙整 51
表4-8 臺灣臺北與高雄測站氣象因子對季風指數隨機森林算法成效評估彙整 52
表4-9 臺灣臺北與高雄測站氣象因子對季風指數XGBoost算法成效評估彙整 52
表4-10 臺灣臺北與高雄測站季風指數最高特徵占比機器學習算法成效評估彙整 52
表4-11 臺灣臺北與高雄測站氣象因子與季風指數延遲30天隨機森林算法MAE、MSE、MAPE與SMAPE成效評估彙整 57
表4-12 臺灣臺北與高雄測站氣象因子與季風指數延遲30天XGBoost算法MAE、MSE、MAPE與SMAPE成效評估彙整 57


1.Gallo, K.P., et al., Satellite‐based detection of global urban heat‐island temperature influence. Journal of Geophysical Research: Atmospheres, 2002. 107(D24): p. ACL 16-1-ACL 16-6.
2.Gu, Y. and D. Li, A modeling study of the sensitivity of urban heat islands to precipitation at climate scales. Urban Climate, 2018. 24: p. 982-993.
3.Cifuentes, J., et al., Air Temperature Forecasting Using Machine Learning Techniques: A Review. Energies, 2020. 13(16): p. 4215.
4.Salcedo-Sanz, S., et al., Monthly prediction of air temperature in Australia and New Zealand with machine learning algorithms. Theoretical and applied climatology, 2016. 125(1): p. 13-25.
5.Radhika, Y. and M. Shashi, Atmospheric temperature prediction using support vector machines. International journal of computer theory and engineering, 2009. 1(1): p. 55.
6.Wang, P., Global monsoon in a geological perspective. Chinese Science Bulletin, 2009. 54(7): p. 1113-1136.
7.Wang, B., et al., Recent change of the global monsoon precipitation (1979–2008). Climate Dynamics, 2012. 39(5): p. 1123-1135.
8.Lin, R., T. Zhou, and Y. Qian, Evaluation of Global Monsoon Precipitation Changes based on Five Reanalysis Datasets. Journal of Climate, 2014. 27(3): p. 1271-1289.
9.Webster, P.J. and S. Yang, Monsoon and ENSO: Selectively interactive systems. Quarterly Journal of the Royal Meteorological Society, 1992. 118(507): p. 877-926.
10.Wang, B. and Z. Fan, Choice of South Asian Summer Monsoon Indices. Bulletin of the American Meteorological Society, 1999. 80(4): p. 629-638.
11.Wu, R. and B. Wang, Interannual Variability of Summer Monsoon Onset over the Western North Pacific and the Underlying Processes. Journal of Climate, 2000. 13(14): p. 2483-2501.
12.Wang, B., Reply. 2000. p. 822-824.
13.Chen, J.-M., T. Li, and C.-F. Shih, Tropical Cyclone– and Monsoon-Induced Rainfall Variability in Taiwan. Journal of Climate, 2010. 23(15): p. 4107-4120.
14.Chen, J.-M., H.-S. Chen, and J.-S. Liu, Coherent Interdecadal Variability of Tropical Cyclone Rainfall and Seasonal Rainfall in Taiwan during October. Journal of Climate, 2013. 26(1): p. 308-321.
15.Gray, W.M., Tropical cyclone genesis. 1975, Colorado State University. Libraries.
16.Frank, W., Tropical cyclone formation. A global view of tropical cyclones. US Office of Naval Research Publication, 1987.
17.Chu, P., ENSO and tropical cyclone activity. Hurricanes and Typhoons: Past, Present, and Potential, RJ Murnane and K.-B. Liu, Eds. Columbia University Press, 2004. 297: p. 332.
18.Chu, P.-S., et al., Climate prediction of tropical cyclone activity in the vicinity of Taiwan using the multivariate least absolute deviation regression method. Terrestrial Atmospheric and Oceanic Sciences, 2007. 18(4): p. 805.
19.Wang, B., Rainy season of the Asian–Pacific summer monsoon. Journal of Climate, 2002. 15(4): p. 386-398.
20.Hsu, H., East Asian and western North Pacific summer monsoon region. Intraseasonal Variability in the Atmosphere-Ocean System, 2005. 65: p. 98.
21.Chen, C.-S. and Y.-L. Chen, The rainfall characteristics of Taiwan. Monthly Weather Review, 2003. 131(7): p. 1323-1341.
22.Wang, B., et al., How to measure the strength of the East Asian summer monsoon. Journal of Climate, 2008. 21(17): p. 4449-4463.
23.Wang, B., R. Wu, and K.-M. Lau, Interannual Variability of the Asian Summer Monsoon: Contrasts between the Indian and the Western North Pacific–East Asian Monsoons. Journal of Climate, 2001. 14(20): p. 4073-4090.
24.Weng, Q., D. Lu, and J. Schubring, Estimation of land surface temperature–vegetation abundance relationship for urban heat island studies. Remote Sensing of Environment, 2004. 89(4): p. 467-483.
25.Chen, X.-L., et al., Remote sensing image-based analysis of the relationship between urban heat island and land use/cover changes. Remote sensing of environment, 2006. 104(2): p. 133-146.
26.Deng, C. and C. Wu, Estimating very high resolution urban surface temperature using a spectral unmixing and thermal mixing approach. International Journal of Applied Earth Observation and Geoinformation, 2013. 23: p. 155-164.
27.Song, Y. and C. Wu, Examining the impact of urban biophysical composition and neighboring environment on surface urban heat island effect. Advances in Space Research, 2016. 57(1): p. 96-109.
28.Shiu, C.-J., S.C. Liu, and J.-P. Chen, Diurnally Asymmetric Trends of Temperature, Humidity, and Precipitation in Taiwan. Journal of Climate, 2009. 22(21): p. 5635-5649.
29.Weng, S.-P., Changes of diurnal temperature range in Taiwan and their large-scale associations: univariate and multivariate trend analyses. Journal of the Meteorological Society of Japan. Ser. II, 2010. 88(2): p. 203-226.
30.Chen, Y.-C., et al., The application of a high-density street-level air temperature observation network (HiSAN): Dynamic variation characteristics of urban heat island in Tainan, Taiwan. Science of The Total Environment, 2018. 626: p. 555-566.
31.Zhou, D., et al., Surface urban heat island in China's 32 major cities: Spatial patterns and drivers. Remote Sensing of Environment, 2014. 152: p. 51-61.
32.Li, L., Y. Zha, and R. Wang, Relationship of surface urban heat island with air temperature and precipitation in global large cities. Ecological Indicators, 2020. 117: p. 106683.
33.Li, D., S. Malyshev, and E. Shevliakova, Exploring historical and future urban climate in the Earth System Modeling framework: 2. Impact of urban land use over the Continental United States. Journal of Advances in Modeling Earth Systems, 2016. 8(2): p. 936-953.
34.Zhou, D., et al., Climate–vegetation control on the diurnal and seasonal variations of surface urban heat islands in China. Environmental Research Letters, 2016. 11(7): p. 074009.
35.Zhao, L., et al., Strong contributions of local background climate to urban heat islands. Nature, 2014. 511(7508): p. 216-219.
36.Arnfield, A.J., Two decades of urban climate research: a review of turbulence, exchanges of energy and water, and the urban heat island. International Journal of Climatology: a Journal of the Royal Meteorological Society, 2003. 23(1): p. 1-26.
37.中央氣象局, 全球與臺灣溫度趨勢分析報告.
38.Bai, Y., J.-Y. Juang, and A. Kondoh, Urban Warming and Urban Heat Islands in Taipei, Taiwan, in Groundwater and Subsurface Environments: Human Impacts in Asian Coastal Cities, M. Taniguchi, Editor. 2011, Springer Japan: Tokyo. p. 231-246.
39.Hwang, R.-L., C.-Y. Lin, and K.-T. Huang, Spatial and temporal analysis of urban heat island and global warming on residential thermal comfort and cooling energy in Taiwan. Energy and Buildings, 2017. 152: p. 804-812.
40.Sung, T.I., et al., Relationship between heat index and mortality of 6 major cities in Taiwan. Sci Total Environ, 2013. 442: p. 275-81.
41.Wang, Y.C., et al., Associating emergency room visits with first and prolonged extreme temperature event in Taiwan: A population-based cohort study. Sci Total Environ, 2012. 416: p. 97-104.
42.Smargiassi, A., et al., Variation of daily warm season mortality as a function of micro-urban heat islands. Journal of Epidemiology and Community Health, 2009. 63(8): p. 659.
43.Gabriel, K.M.A. and W.R. Endlicher, Urban and rural mortality rates during heat waves in Berlin and Brandenburg, Germany. Environmental Pollution, 2011. 159(8): p. 2044-2050.
44.Vandentorren, S., et al., August 2003 Heat Wave in France: Risk Factors for Death of Elderly People Living at Home. European Journal of Public Health, 2006. 16(6): p. 583-591.
45.Bobb, J.F., et al., Cause-specific risk of hospital admission related to extreme heat in older adults. Jama, 2014. 312(24): p. 2659-67.
46.Brunetti, N.D., et al., Hot spot: impact of July 2011 heat wave in southern Italy (Apulia) on cardiovascular disease assessed by emergency medical service and telemedicine support. Telemed J E Health, 2014. 20(3): p. 272-81.
47.Chung, Y., et al., Mortality related to extreme temperature for 15 cities in northeast Asia. Epidemiology, 2015. 26(2): p. 255-62.
48.Huynen, M., et al., The Impact of Heat Waves and Cold Spells on Mortality Rates in the Dutch Population. Environmental health perspectives, 2001. 109: p. 463-70.
49.Kim, J., et al., Comprehensive approach to understand the association between diurnal temperature range and mortality in East Asia. Sci Total Environ, 2016. 539: p. 313-321.
50.Diffenbaugh, N.S., et al., Fine-scale processes regulate the response of extreme events to global climate change. Proc Natl Acad Sci U S A, 2005. 102(44): p. 15774-8.
51.Hance, T., et al., Impact of extreme temperatures on parasitoids in a climate change perspective. Annu Rev Entomol, 2007. 52: p. 107-26.
52.McMichael, A.J. and B.A. Wilcox, Climate Change, Human Health, and Integrative Research: A Transformative Imperative. EcoHealth, 2009. 6(2): p. 163-164.
53.McMichael, A.J., R.E. Woodruff, and S. Hales, Climate change and human health: present and future risks. The Lancet, 2006. 367(9513): p. 859-869.
54.Pearson, K., Principal components analysis. The London, Edinburgh, and Dublin Philosophical Magazine and Journal of Science, 1901. 6(2): p. 559.
55.Wold, S., K. Esbensen, and P. Geladi, Principal component analysis. Chemometrics and Intelligent Laboratory Systems, 1987. 2(1): p. 37-52.
56.Chen, T. and C. Guestrin. Xgboost: A scalable tree boosting system. in Proceedings of the 22nd acm sigkdd international conference on knowledge discovery and data mining. 2016.
57.Tianqi Chen, U.o.W., et al. XGBoost Documentation. 2021; Available from: https://xgboost.readthedocs.io/en/stable/index.html.
58.Ma, X., C. Fang, and J. Ji. Prediction of outdoor air temperature and humidity using Xgboost. in IOP Conference Series: Earth and Environmental Science. 2020. IOP Publishing.
59.Zhang, Y., et al. Prediction of daily smoking behavior based on decision tree machine learning algorithm. in 2019 IEEE 9th International Conference on Electronics Information and Emergency Communication (ICEIEC). 2019. IEEE.
60.Ahmad, M.W., M. Mourshed, and Y. Rezgui, Trees vs Neurons: Comparison between random forest and ANN for high-resolution prediction of building energy consumption. Energy and buildings, 2017. 147: p. 77-89.
61.Ho, T.K. Random decision forests. in Proceedings of 3rd international conference on document analysis and recognition. 1995. IEEE.
62.Breiman, L., Random forests. Machine learning, 2001. 45(1): p. 5-32.


電子全文 電子全文(網際網路公開日期:20270811)
QRCODE
 
 
 
 
 
                                                                                                                                                                                                                                                                                                                                                                                                               
第一頁 上一頁 下一頁 最後一頁 top
無相關期刊