用python gdal裁剪栅格数据提取添加xy经纬度和栅格值

用python gdal裁剪栅格数据提取添加xy经纬度和栅格值

问题：把遥感影像转为一张表。

现有一全球经济作物数据alfalfa的产量。

alfalfa是一种亚洲西南部多年生草本植物，是重要的经济作物。在图中也可以看到，主要分布在热带和南美洲。

我们想把影像转表，即经纬度、栅格值（苜蓿产量）

上述功能在ArcGIS中是这样实现的。

对于我上述全球影像来说，栅格转点需要6分钟。添加字段和计算几何都需要花费更多的时间。

采用python的gdal方法，首先进行影像裁剪。直接上代码：

dataset = gdal.Open("D:/work/0318/Suitability Raster files/Suitability Raster files/High input/high_banana_plaintain.tif")
output_raster=r'D:/work/0318/Suitability Raster files/Suitability Raster files/High_mask/high_banana_plaintain_mask.tif' 
input_shape = r'D:/work/0318/shp/Africa.shp'

# 开始裁剪
ds = gdal.Warp(output_raster,
              dataset,
              format = 'GTiff',
              cutlineDSName = input_shape,      
              cutlineWhere="FIELD = 'whatever'",
              dstNodata = -90)   

这里我设置nodata为负值，是我本来影像的nodata值，可以在GIS查看

然后再进行统计：

import time
from osgeo import gdal
import numpy as np
import pandas as pd
import os


def rasterToPoints(rasterfile, nodata=None, v_name=None):
    """
    :param rasterfile: 待执行栅格转点的栅格文件
    :param nodata:栅格中的无数据值,默认取栅格的最小值
    :param v_name:导出表格中栅格值所在列的名称，默认为栅格的文件名
    :return:x、y、value
    """
    # numpy禁用科学计数法,pandas中存储浮点型时只保留四位小数
    np.set_printoptions(suppress=True)
    pd.set_option('display.float_format', lambda x: '%.4f' % x)

    rds = gdal.Open(rasterfile)  # type:gdal.Dataset
    if rds.RasterCount != 1:
        print("Warning, RasterCount > 1")

    cols = rds.RasterXSize
    rows = rds.RasterYSize
    band = rds.GetRasterBand(1)  # type:gdal.Band
    transform = rds.GetGeoTransform()
    print(transform)
    x_origin = transform[0]
    y_origin = transform[3]
    pixel_width = transform[1]
    pixel_height = transform[5]
    if (pixel_height + pixel_width) != 0:
        print("Warning, pixelWidth != pixelHeight")
    # 读取栅格
    values = np.array(band.ReadAsArray())
    x = np.arange(x_origin + pixel_width * 0.5, x_origin + (cols + 0.5) * pixel_width, pixel_width)
    y = np.arange(y_origin + pixel_height*0.5, y_origin + (rows+0.5) * pixel_height, pixel_height)
    px, py = np.meshgrid(x, y)
    if v_name is None:
        v_name = os.path.splitext(os.path.split(rasterfile)[1])[0]
    dataset = {"x": px.ravel(),
               "y": py.ravel(),
               v_name: values.ravel()}
    df_temp = pd.DataFrame(dataset, dtype="float32")

    # 删除缺失值
    if nodata is None:
        nodata = df_temp[v_name].min()
        df_temp = df_temp[df_temp[v_name] != nodata]
    else:
        df_temp = df_temp[df_temp[v_name] != nodata]

    df_temp.index = range(len(df_temp))
    return df_temp


if __name__ == "__main__":
    # 禁用科学计数法
    np.set_printoptions(suppress=True)
    pd.set_option('display.float_format', lambda x: '%.4f' % x)
    # 执行栅格转点，并计时
    s = time.time()
    # in_tif是输入栅格，刚才裁剪的结果
    in_tif = r"D:/work/0318/Suitability Raster files/Suitability Raster files/High_mask/high_banana_plaintain_mask.tif" 
    outfile = rasterToPoints(in_tif, v_name="high_banana_plaintain") # v_name是你自己定义的栅格字段列名称
    outfile.to_csv("high_banana_plaintain.csv") # 导出csv文件
    e = time.time()
    print("time used {0}s".format(e-s))

成功了。

看看统计结果

cs = pd.read_csv('high_banana_plaintain.csv')
cs

Unnamed: 0	x	y	high_banana_plaintain
0	0	9.3750	37.2917	0.0000
1	1	9.4583	37.2917	0.0000
2	2	9.5417	37.2917	0.0000
3	3	9.6250	37.2917	0.0000
4	4	9.7083	37.2917	0.0000
…	…	…	…	…
360807	360807	19.7083	-34.7917	0.0000
360808	360808	19.7917	-34.7917	0.0000
360809	360809	19.8750	-34.7917	0.0000
360810	360810	19.9583	-34.7917	0.0000
360811	360811	20.0417	-34.7917	0.0000

360812 rows × 4 columns

很完美。