气象统计常用公式

Chapter2 气象资料及其表示方法

1. 方差

$$\begin{array}{rl}{s}_x& =\frac{1}{n}\sum _{t=1}^n(x_t-\overline{x}{)}^{2}t=1,2,3,\cdots ,n\\ & =\frac{1}{n}{x}^{{}^{\prime }}{x}^{{}^{\prime }T}{x}^{{}^{\prime }}表示距平向量\end{array}$$

2. 标准差

$$\begin{array}{r}{s}_x=\sqrt{\frac{1}{n}\sum _{t=1}^n(x_t-\overline{x}{)}^2}t=1,2,3,\cdots ,n\end{array}$$

3. 变率

• 绝对变率

$$\begin{array}{r}{V}_a=\frac{1}{n}\sum _{t=1}^n|x_t-\overline{x}|t=1,2,3,\cdots ,n\end{array}$$

• 相对变率

$$\begin{array}{r}{V}_r=\frac{V_a}{\overline{x}}\end{array}$$

4. 变差系数

$$\begin{array}{rl}{V}_p& =\frac{s_x}{\overline{x}}\\ & =\frac{1}{\overline{x}}\sqrt{\frac{1}{n}\sum _{t=1}^n(x_t-\overline{x}{)}^2}\end{array}$$

5. 标准化变量

$$\begin{array}{r}{x}^{\star }=\frac{x-\overline{x}}{s_x}\\ 易得：E(x^{\star })=0,Var(x^{\star })=1\end{array}$$

6. 数据矩阵

$$\begin{gathered} {}_m{X}_n=\left[\begin{array}{cccc}{x}_{11}& x_{12}& \cdots & x_{1n}\\ x_{21}& x_{22}& \cdots & x_{2n}\\ ⋮& ⋮& \ddots & ⋮\\ x_{m1}& x_{m2}& \cdots & x_{mn}\end{array}\right]=(x_1,x_2,\cdots ,x_n)=(x_t)t=1,2,3,\cdots ,n \\ 该矩阵表示n个样本的m个要素，即行表示一个要素(变量)，列表示一个观测(样本) \end{gathered}$$
$$x_t=(x_{1t},x_{2t},\cdots ,x_{mt}{)}^T第t个样本的m个要素向量$$

7. 均值向量

$$\begin{array}{rl}\overline{x}& =(\overline{x_1},\overline{x_2},\cdots ,\overline{x_m}{)}^T\\ \overline{x_i}& =\frac{1}{n}\sum _{t=1}^n{x}_{it}(t=1,2,3,\cdots ,n.i=1,2,3,\cdots ,m)\end{array}$$

8. 协方差

$$\begin{array}{rl}{s}_{ij}& =\frac{1}{n}\sum _{t=1}^n(x_{it}-\overline{x_i})(x_{jt}-\overline{x_j})\\ & =\frac{1}{n}\sum _{t=1}^n{x}_{it}{x}_{jt}-\overline{x_i}\overline{x_j}\\ & =\frac{1}{n}\sum _{t=1}^n{x}_{it}^{{}^{\prime }}{x}_{jt}^{{}^{\prime }}距平形式\\ & =\frac{1}{n}{x}_i{x}_j向量形式，距平向量的内积\end{array}$$

9. 协方差矩阵(离差积矩阵)

$$\begin{array}{rl}SS& =(s{s}_{ij})(i,j=1,2,\cdots ,m)\\ 其中:s{s}_{ij}& =\sum _{t=1}^n(x_{it}-\overline{x_i})(x_{jt}-\overline{x_j})\\ 总体协方差矩阵的无偏估计:S& =\frac{1}{n-1}SS\end{array}$$

10. 偏度系数

• 老师PPT里的公式

$$\begin{array}{r}{g}_1=\sqrt{\frac{1}{6n}}\sum _{i=1}^n{\left(\frac{x_i-\overline{x}}{s}\right)}^3\end{array}$$

• 概率论课本的公式

$$\begin{array}{r}{\beta }_s=\frac{{\nu }_3}{{\nu }_2^{3/2}}{\nu }_k表示样本k阶中心距\end{array}$$

11. 峰度系数

• 老师PPT里的公式

$$\begin{array}{r}{g}_2=\sqrt{\frac{1}{24n}}\left[\frac{1}{n}\sum _{i=1}^n{\left(\frac{x_i-\overline{x}}{s}\right)}^4-3\right]\end{array}$$

• 概率论课本的公式

$$\begin{array}{r}{\beta }_k=\frac{{\nu }_4}{{\nu }_2^2}-3\end{array}$$

Chapter3 选择最大信息的预报因子

1. 样本值对总体值的抽样误差

$$\begin{array}{rl}E(\overline{x}-m{)}^2& =\frac{E(x-m{)}^2}{n}\\ {\mu }_x^2=\frac{{\sigma }^2}{n}& \end{array}$$

2. 条件概率

$$\begin{array}{r}P(A|B)=\frac{P(AB)}{P(B)}\end{array}$$

• A与B相互独立
  $$P(AB)=P(A)P(B)$$

3. 二项分布

$$\begin{array}{rl}{P}_n(m)& =C_n^m{p}^m(1-p{)}^{m-n}\\ 其中C_n^m& =\frac{n!}{m!(n-m)!}\end{array}$$

4. 简单相关系数

$$\begin{array}{rl}{r}_{xy}& =\frac{s_{xy}}{s_x{s}_y}\\ & =\frac{\frac{1}{n}{\sum }_{i=1}^n(x_i-\overline{x})(y_i-\overline{y})}{\sqrt{\frac{1}{n}{\sum }_{i=1}^n(x_i-\overline{x}{)}^2}\sqrt{\frac{1}{n}{\sum }_{i=1}^n(y_i-\overline{y}{)}^2}}\\ & =\frac{{\sum }_{i=1}^n{x}_i{y}_i-n\overline{x}\cdot \overline{y}}{\sqrt{\frac{1}{n}{\sum }_{i=1}^n(x_i-\overline{x}{)}^2}\sqrt{\frac{1}{n}{\sum }_{i=1}^n(y_i-\overline{y}{)}^2}}\\ & =\frac{\frac{1}{n}{\sum }_{i=1}^n{x}_i^{{}^{\prime }}{y}_i^{{}^{\prime }}}{\sqrt{\frac{1}{n}{\sum }_{i=1}^n(x_i^{{}^{\prime }}{)}^2}\sqrt{\frac{1}{n}{\sum }_{i=1}^n(y_i^{{}^{\prime }}{)}^2}}\end{array}$$

5. 自相关系数

$$\begin{array}{rl}r(j)& =\frac{s(j)}{s^2}\\ & =\frac{1}{s^2}\cdot \frac{1}{n-j}\sum _{t=1}^{n-j}(x_t-\overline{x})(x_{t+j}-\overline{x})\\ & 其中s^2为该序列的方差,t=1,2,\cdots ,n-j\end{array}$$

6. 落后交叉相关系数

$$\begin{array}{rl}{r}_{xy}(j)& =\frac{s_{xy}(j)}{s_x{s}_y}\\ & =\frac{1}{s_x{s}_y}\cdot \frac{1}{n-j}\sum _{t=1}^{n-j}(x_t-\overline{x})(y_{t+j}-\overline{y})\\ & 其中s_x,s_y分别为序列x和y的标准差,t=1,2,\cdots ,n-j\end{array}$$

7. 偏相关系数

$$\begin{gathered} {}_3{R}_3=\left[\begin{array}{ccc}{r}_{DD}& r_{DJ}& r_{DF}\\ r_{JD}& r_{JJ}& r_{JF}\\ r_{FD}& r_{FJ}& r_{FF}\end{array} \\ \right] \\ 其中D为12月,J为1月,F为2月 \end{gathered}$$

$$\begin{array}{rl}{r}_{DJ\cdot F}& =-\frac{R_{DJ}^{\star }}{\sqrt{R_{DD}^{\star }{R}_{JJ}^{\star }}}\\ r_{DF\cdot J}& =-\frac{R_{DF}^{\star }}{\sqrt{R_{DD}^{\star }{R}_{FF}^{\star }}}\end{array}$$

Chapter4 回归分析

1. 一元线性回归

• 回归系数

$$\begin{array}{rl}b& =(n\overline{x}\cdot \overline{y}-\sum _{i=1}^n{x}_i{y}_i)/(n{\overline{x}}^2-\sum _{i=1}^n{x}_i^2)\\ & =s_{xy}/s_x^2\\ a& =\overline{y}-b\cdot \overline{x}\end{array}$$

• 反映回归效果的统计量

  • 离差平方和
    $$\begin{array}{r}{S}_{yy}=\sum _{t=1}^n(y_t-\overline{y}{)}^2\end{array}$$

  • 回归平方和
      也称为可解释平方和或者回归方差
    $$\begin{array}{r}U=\sum _{t=1}^n({\hat{y}}_t-\overline{y}{)}^2\end{array}$$

  • 残差平方和
      也称为不可解释平方和或者剩余方差
    $$\begin{array}{r}Q=\sum _{t=1}^n(y_t-{\hat{y}}_t{)}^2\end{array}$$

  • 判决系数
    $$\begin{array}{r}{R}^2=\frac{U}{S_{yy}}=r_{xy}^2\end{array}$$

• 回归系数与相关系数的关系
  $$\begin{array}{r}b=\frac{s_{xy}}{s_x^2}=\frac{s_y{s}_{xy}}{s_x{s}_x{s}_y}=\frac{s_y}{s_x}{r}_{xy}\end{array}$$

• 回归方程的显著性检验(F检验)
  $$\begin{array}{rl}F& =\frac{U/m}{Q/(n-m-1)}或\\ F& =\frac{r_{xy}^2/m}{(1-r_{xy}^2)/(n-m-1)}\\ 服& 从自由度为(m,n-m-1)的F分布\end{array}$$

2. 多元线性回归

• 回归系数

$$\begin{array}{r}b=(X^{T}X{)}^{-1}{X}^{T}Y\end{array}$$

• 复相关系数
  $$\begin{array}{r}R=\sqrt{1-\frac{Q}{S_{yy}}}=\sqrt{\frac{U}{S_{yy}}}此处Q、U、S_{yy}均与一元线性回归的定义一致\end{array}$$

• 多元线性回归方程的检验统计量
  $$\begin{array}{rl}F& =\frac{U/m}{Q/(n-m-1)}或\\ F& =\frac{R^2/m}{(1-R^2)/(n-m-1)}\\ 服& 从自由度为(m,n-m-1)的F分布\end{array}$$

• 方差贡献
  $$\begin{array}{r}{Q}_i=\frac{b_i^2}{c_{ii}}\end{array}$$

• 预报因子的显著性检验

$$\begin{array}{rl}{F}_i& =\frac{Q_i}{Q/(n-m-1)}=\frac{b_i^2/c_{ii}}{Q/(n-m-1)}\\ 服& 从自由度为(1,n-m-1)的F分布\end{array}$$

Chapter5 气候趋势分析

1. 滑动平均

$$\begin{array}{rl}\widehat{x_j}& =\frac{1}{k}\sum _{i=1}^k{x}_{i+j-1}(j=1,2,\cdots ,n-k+1)\\ & k为滑动长度，一般取奇数\end{array}$$

2. 累积距平

$$\begin{array}{r}\widehat{x_t}=\sum _{i=1}^t(x_i-\overline{x})(t=1,2,\cdots ,n)\end{array}$$

3. 五、七、九点二次平滑

• 五点平滑
  $$\begin{array}{r}{\hat{x}}_{i-2}=\frac{1}{35}(-3x_{i-2}+12x_{i-1}+17x_i+12x_{i+1}-3x_{i+2})\end{array}$$

• 七点平滑

$$\begin{array}{r}{\hat{x}}_{i-3}=\frac{1}{21}(-2x_{i-3}+3x_{i-2}+6x_{i-1}+7x_i+6x_{i+1}+3x_{i+2}-2x_{i+3})\end{array}$$

• 九点平滑

$$\begin{array}{r}{\hat{x}}_{i-4}=\frac{1}{231}(-21x_{i-4}+14x_{i-3}+39x_{i-2}+54x_{i-1}+59x_i+54x_{i+1}+39x_{i+2}+14x_{i+3}-21x_{i+4})\end{array}$$