Statistic And Probabilities In Hydrology Probability and Statistics in Hydrology treats probability theory and mathematical statistics as applied to hydrology. Probability theory is presented in a summarized form with emphasis on its use in hydrology. Statistically, the emphasis is on inferential rather than descriptive statistics of classical hydrologic applications. Since hydrologic processes in nature are governed by the laws of chance, the use of probability theory and mathematical statistics is unavoidable in the extraction of information from hydrologic data and for the best mathematical description of these processes. This book is not a conventional statistical treatment of the subject. Certain concepts are defined; mathematical expressions are written in a format that is ready to use; techniques are explained, as well as their applications to hydrology (with examples) are presented. This book is aimed at practicing hydrologists and engineers; graduate students; post-graduate students; and specialist interested in probability theory and mathematical statistics as applied to hydrology. It is now part of WRP’s Classic Resource Edition.
Trang 1授課教師 : 張斐章
Trang 3 2.5 機率紙圖解法 Graphic Method Using Probability Paper
2.6 信賴極限 Confidence Limits
2.7 迴歸與相關分析 Regression and Correlation
2.8多變量線性迴歸與相關 Multivariate Linear Regression and
Trang 52.1 序論 Introduction
水文統計的目的 :
替自然界的水文現象用數理模式來 建立關係
提供工程設計參考 預測未來的水文現象
Trang 62.1 序論 Introduction
模式方法 :
定率歷程
自然界的水文現象遵守物理定律而非機率定律, 降水與逕流間的 關係確定不變.
機率歷程
自然界的水文現象是按照機率定律而發生, 水文事件的發生與時 間前後順序無關, 且遵循著機率分佈.
序率歷程
自然界的水文現象是依照機率定律而發生,水文事件的發生隨時
Trang 92.2 統計參數 Statistical Parameters
Example :
Station A B C D E F G H I Rainfall(cm) 51.8 32.0 28.7 43.4 38.6 50.5 59.6 31.5 31.7 Area(sq km) 31.0 58.0 31.5 31.0 86.0 71.0 27.0 43.5 66.0
X X
Trang 10w X w X w X X
Trang 112.2 統計參數 Statistical Parameters
樣本中最中間的值, 代表集中趨勢 但代表性不如平均數, 運用在無母數統計中較多.
樣本中發生次數最多的值, 代表集中趨勢 但代表性不如平 均數集中位數, 並且不一定存在有時不只一個.
Trang 122.2 統計參數 Statistical Parameters
代表每一個數與樣本平均數之平均距離, 可以看出母體之 平均距離情形.
( )2 1
1
n
i i
x x n
x N
Trang 14.
C S µ
σ
=
Trang 16.
C K µ
σ
=
Trang 195 1 4 1
n = p = q =
Trang 202.3.1.1 二項分佈 Biominal Distribution
Example
某水庫的建造時間須花費十年, 在施工期間可能遭遇到5年重現期距的洪 水而造成損失的機率.
Trang 22e p
Trang 23σ
µ π
σ
x x
Trang 242
z z
Trang 262.4 頻率分析 Frequency Analysis
主要目的為根據歷史資料,以統計方法推求大於(小於) 或等於某一水文量,在未來一定期間內可能發生之機率, 以作為規劃設計的依據。
4.水質研究 5.水波研究
1.洪水及河川流量頻率 2.降雨量頻率
3.乾旱或低流量頻率
Trang 27 常用的頻率分析方法
(i) 頻率因子(Frequency factor)
周文德博士(Dr Ven Te Chow, 1951)提出 極端事件(洪水與乾旱)頻率分析方法
(ii)定點法(Plotting position)
將水文事件點繪於於適當的機率紙上,再求出最佳適合之 頻率曲線《 2.5節》
Trang 28x 是一個值 (如:200cms)
Trang 31X x x
µ σ
迴歸週期為 時之水文量
、 水文資料的平均值
Δ 均值偏差 :標準差
Trang 352.4.2 Pearson Type-III Distribution
Trang 372.4.3 Log-Pearson Type-III Distribution
37
Trang 402.4.3 Log Normal Distribution
y y y
x
f x
x
µ σ πσ
Trang 42(mm) 133 94.5 76 87.5 92.7 71.3 77.3 85.1 122.8 69.4Year 1960 1961 1962 1963 1964 1965 1966 1967 1968 1969
Year 1970 1971 1972 1973 1974
Runoff
Runoff
(mm) 81 94.5 86.3 68.6 82.5 90.7 99.8 74.4 66.6 65
Trang 43N C
Trang 440.124358
25 1 0.072
0.0372 1.9327
Trang 49T T T
e T
Trang 51Graphical Method Using Probability Paper
2.5 機率紙圖解法
定點法(Plotting position)
水文資料在某種指定的機率紙上,呈現近似一直線 的現象
Trang 535 找出迴歸週期T年之水文量
T= 1/P = (N+1)/m
5 1 5
135 1955
150 1951
285 1950
445 1954
排序後流量x m
排序後年代
5 4 3 2 1
Trang 552.5.1 Construction of Probability Paper
Trang 56海生機率紙
Trang 57甘保機率紙
Trang 582.5.2 Selection of Type of Distribution
甘保分佈(Extreme value distribution)
Extreme value Type-III distribution
Log normal distribution
Exponential distribution
選擇機率分佈
Trang 592.6 信賴區間 Confidence Limits
真實,使得當迴歸週期較大時,只能以外差方式來推 估水文量
Trang 622.7.1 Graphical Method
將所有資料點點繪於座標上,以目視方法約略畫一條迴歸線
Trang 67−
=
1/ 2 2
Trang 682.7.5 Standard Form of Bivariate Equations
其他類型雙變數間的迴歸式
Linear:
Exponential:
Parabola:
Higher order equation:
Other forms of equations:
Trang 692 2
2 2
Trang 70(1 )
r
r S
S S S S
Trang 71Example 2.13
Assume a non-linear relation: y=ax b
兩邊取log => log(y) = log(a)+blog(x)
Trang 73 樣本之發生機率與所假設分佈之理論機率的 值可表 示為
Trang 74 上式中v為自由度;O i為紀錄資料內之實際觀測數量;
Pei為所假設機率分佈之期望發生數量。若上式所得之 計算值大於理論值,則表示所假設的機率分佈並不適 合此水文紀錄資料。
Trang 75σ
Trang 76The chi-square test statistics is calculated in col (9)
For example, 0.141 in row(4) is
For degrees of freedom of (8-2-1) = 5,
2
0.141 0.224
2