Python遥感开发之数据趋势分析Sen+mk

前言：主要使用Python完成遥感时间序列数据趋势分析Sen+mk，得到slope、trend、p、s、tau、z指标。

---

1 方法介绍

1.1 Theil-Sen Median方法

又被称为 Sen 斜率估计，是一种稳健的非参数统计的趋势计算方法。该方法计算效率高，对于测量误差和离群数据不敏感，常被用于长时间序列数据的趋势分析中。对于后续代码计算结果中的slope.tif解读,当slope大于0表示随时间序列呈现上升趋势；slope小于0表示随时间序列呈现下降趋势。

1.2 Mann-Kendall方法

Mann-Kendall是一种非参数统计检验方法，最初由Mann在1945年提出，后由Kendall和Sneyers进一步完善，其优点是不需要测量值服从正态分布，也不要求趋势是线性的，并且不受缺失值和异常值的影响，在长时间序列数据的趋势显著检验中得到了十分广泛的应用。对于后续代码计算结果中的z.tif,当|Z|大于1.65、1.96和2.58时，表示趋势分别通过了置信度为90%、95%和99%的显著性检验。

2 Python实现Sen+mk

# coding:utf-8
import numpy as np
import pymannkendall as mk
import os
import rasterio as ras

# 写影像
def writeImage(image_save_path, height1, width1, para_array, bandDes, transform1):
    with ras.open(
            image_save_path,
            'w',
            driver='GTiff',
            height=height1,
            width=width1,
            count=1,
            dtype=para_array.dtype,
            crs='+proj=latlong',
            transform=transform1,
    ) as dst:
        dst.write_band(1, para_array)
        dst.set_band_description(1, bandDes)
    del dst

def sen_mk_test(image_path, outputPath):
    # image_path:影像的存储路径
    # outputPath:结果输出路径
    global path1
    filepaths = []
    for file in os.listdir(path1):
        filepath1 = os.path.join(path1, file)
        filepaths.append(filepath1)

    # 获取影像数量
    num_images = len(filepaths)
    # 读取影像数据
    img1 = ras.open(filepaths[0])
    # 获取影像的投影，高度和宽度
    transform1 = img1.transform
    height1 = img1.height
    width1 = img1.width
    array1 = img1.read()
    img1.close()
    # 读取所有影像
    for path1 in filepaths[1:]:
        if path1[-3:] == 'tif':
            print(path1)
            img2 = ras.open(path1)
            array2 = img2.read()
            array1 = np.vstack((array1, array2))
            img2.close()
    nums, width, height = array1.shape
    # 输出矩阵，无值区用-9999填充
    slope_array = np.full([width, height], np.nan)
    z_array = np.full([width, height], np.nan)
    Trend_array = np.full([width, height], np.nan)
    Tau_array = np.full([width, height], np.nan)
    s_array = np.full([width, height], np.nan)
    p_array = np.full([width, height], np.nan)
    # 只有有值的区域才进行mk检验
    c1 = np.isnan(array1)
    sum_array1 = np.sum(c1, axis=0)
    nan_positions = np.where(sum_array1 == num_images)
    positions = np.where(sum_array1 != num_images)
    # 输出总像元数量
    print("all the pixel counts are {0}".format(len(positions[0])))
    # mk test
    for i in range(len(positions[0])):
        print(i)
        x = positions[0][i]
        y = positions[1][i]
        mk_list1 = array1[:, x, y]
        trend, h, p, z, Tau, s, var_s, slope, intercept = mk.original_test(mk_list1)
        '''        
        trend: tells the trend (increasing, decreasing or no trend)
                h: True (if trend is present) or False (if trend is absence)
                p: p-value of the significance test
                z: normalized test statistics
                Tau: Kendall Tau
                s: Mann-Kendal's score
                var_s: Variance S
                slope: Theil-Sen estimator/slope
                intercept: intercept of Kendall-Theil Robust Line
        '''
        if trend == "decreasing":
            trend_value = -1
        elif trend == "increasing":
            trend_value = 1
        else:
            trend_value = 0
        slope_array[x, y] = slope  # senslope
        s_array[x, y] = s
        z_array[x, y] = z
        Trend_array[x, y] = trend_value
        p_array[x, y] = p
        Tau_array[x, y] = Tau
    all_array = [slope_array, Trend_array, p_array, s_array, Tau_array, z_array]
    slope_save_path = os.path.join(result_path, "slope.tif")
    Trend_save_path = os.path.join(result_path, "Trend.tif")
    p_save_path = os.path.join(result_path, "p.tif")
    s_save_path = os.path.join(result_path, "s.tif")
    tau_save_path = os.path.join(result_path, "tau.tif")
    z_save_path = os.path.join(result_path, "z.tif")
    image_save_paths = [slope_save_path, Trend_save_path, p_save_path, s_save_path, tau_save_path, z_save_path]
    band_Des = ['slope', 'trend', 'p_value', 'score', 'tau', 'z_value']
    for i in range(len(all_array)):
        writeImage(image_save_paths[i], height1, width1, all_array[i], band_Des[i], transform1)

if __name__ == '__main__':
    # 调用
    path1 = r"E:\AAWORK\work\研究方向\rsei\data\年\rsei_year-25km"
    result_path = r"E:\AAWORK\work\研究方向\rsei\data\年\out"
    sen_mk_test(path1, result_path)

• slope_array：存储每个像元的Sen’s Slope值，表示该像元的趋势斜率。
• trend_array：存储每个像元的趋势方向，1表示增加，-1表示减少，0表示无趋势。
• p_array：存储每个像元的p值，用于判断趋势的显著性。
• s_array：存储每个像元的Mann-Kendall S统计量。
• tau_array：存储每个像元的Kendall Tau值，表示相关性强度和方向。
• z_array：存储每个像元的标准化检验统计量Z，用于计算p值并判断趋势显著性。

3 最终结果

3.1 对slope.tif以及z.tif进行重分类

• slope>0赋值1表示增加
• slope=0赋值0表示不变
• slope<0赋值-1表示减少
• |z|>1.96赋值2表示显著
• |z|<=1.96赋值1表示不显著

3.2 对分类结果相乘

• -2:显著减少
• -1:不显著减少
• 0:稳定不变
• 1:不显著增加
• 2:显著增加

3.3 最终结果Python代码实现

import numpy as np
import rasterio as ras

def classify_rasters(slope_path, z_path, output_path):
    with ras.open(slope_path) as slope_src, ras.open(z_path) as z_src:
        slope = slope_src.read(1)
        z = z_src.read(1)
        profile = slope_src.profile

    # 重分类
    classified_slope = np.full(slope.shape, 0, dtype=np.int8)
    classified_slope[slope > 0] = 1
    classified_slope[slope < 0] = -1

    classified_z = np.full(z.shape, 1, dtype=np.int8)
    classified_z[np.abs(z) > 1.96] = 2
    classified_z[np.abs(z) <= 1.96] = 1

    # 相乘得到最终分类结果
    final_class = classified_slope * classified_z

    # 保存结果
    with ras.open(
        output_path, 'w',
        driver='GTiff',
        height=final_class.shape[0],
        width=final_class.shape[1],
        count=1,
        dtype=final_class.dtype,
        crs=profile['crs'],
        transform=profile['transform']
    ) as dst:
        dst.write(final_class, 1)

if __name__ == '__main__':
    # 调用
    slope_path = r"E:\AAWORK\work\研究方向\rsei\data\年\out\slope.tif"
    z_path = r"E:\AAWORK\work\研究方向\rsei\data\年\out\z.tif"
    output_path = r"E:\AAWORK\work\研究方向\rsei\data\年\out\final_classification.tif"
    classify_rasters(slope_path, z_path, output_path)