netCDF4 处理nc文件总结

创建nc文件

# 数据类型包括：f4(32位浮点数),f8(64位浮点数),i1(8位符号整型),i2(16位符号整型),i4(32位符号整型),i8(64位符号整型),
#             u1(8位无符号整型),u2(16位无符号整型),u4(32位无符号整型),u8(64位无符号整型),S1(单字符字符串)

with nc.Dataset(nc_name, 'w', format='NETCDF4') as fnc:
    fnc.createDimension('lat', 3541)  # 确定维度大小
    fnc.createDimension('lon', 6165)
    #                     变量名   数据类型  维度大小
    lat = fnc.createVariable('lat', 'f4', ('lat',))  # 创建变量
    lat.long_name = "latitude"  # 该变量的名称
    lat.units = "degrees_north"  # 该变量的单位

    lon = fnc.createVariable('lon', 'f4', ('lon',))
    lon.long_name = "longitude"
    lon.units = "degrees_east"
    #                                                  是否压缩数据    压缩级别（1-9级） 默认初始填充值
    alt = fnc.createVariable('alt', 'f4', ('lat', 'lon'), zlib=True, complevel=9, fill_value=-9999)
    alt.long_name = 'height above sea level'
    alt.coordinates = 'lat lon alt'
    alt.units = 'meters'
    alt.positive = 'up'  # 高度的方向  up为向上

    PRS = fnc.createVariable('PRS', 'short', ('lat', 'lon'), zlib=True, complevel=9, fill_value=-9999)  # 气压
    PRS.long_name = 'station pressure'
    PRS.units = 'hPa'
    PRS.missing_value = -9999  # 默认缺测填充
    PRS.coordinates = 'lat lon alt'
    PRS.add_offset = np.float32(0.0)  # 增加偏移
    PRS.scale_factor = np.float32(0.1)  # 比例偏移  实际值存入自动扩大10倍（除以0.1）,取出自动缩小10倍（乘以0.1）
    PRS.valid_min = 5200  # 限制存入最小值  小于最小为缺测
    PRS.valid_max = 10850  # 限制存入最大值  大于最大为缺测

    # 给变量赋值               赋值的数据维度--必须和创建变量时指定的维度大小一致
    fnc.variables['lat'][:] = np.arange(53.56, 18.16, -0.01)
    fnc.variables['lon'][:] = np.arange(73.44, 135.09, 0.01)
    fnc.variables['alt'][:] = np.array(data['alt'], dtype='float').reshape(3541, 6165)
    fnc.variables['PRS'][:] = np.array(data['PRS'], dtype='float').reshape(3541, 6165)

    fnc.author = 'ecoweather'  # 创建文件作者
    fnc.createdate = '%s' % datetime.datetime.now().strftime('%Y-%m-%d %H:%M') # 创建文件时间

创建有时间变量的nc文件

with nc.Dataset(nc_name, 'w', format='NETCDF4') as fnc:
    fnc.createDimension('time', 24)  # 确定维度大小
    fnc.createDimension('size', 10447)

    tt = fnc.createVariable('time', 'i4', 'time')
    tt.long_name = 'time'
    # 起始小时时间   它会自动根据你赋值的数据，从起始小时累加
    tt.units = 'hours since %s' % datetime.datetime.strptime(start_time, '%Y%m%d%H%M%S').strftime('%Y-%m-%d %H:%M:%S')
    # 日历（公历）
    tt.calendar = "gregorian"

    sta = fnc.createVariable('sta', 'i4', ('size',))  # 创建变量
    sta.coordinates = 'lat lon alt'
    sta.long_name = 'station number'  # 站号

    lat = fnc.createVariable('lat', 'f4', ('size',))
    lat.long_name = "latitude"
    lat.units = "degrees_north"

    lon = fnc.createVariable('lon', 'f4', ('size',))
    lon.long_name = "longitude"
    lon.units = "degrees_east"

    PRE_1m = fnc.createVariable('PRE_1m', 'f4', ('time', 'size'), zlib=True, complevel=9, fill_value=-9999)  # 一分钟降水
    PRE_1m.long_name = 'precipitation of 1 min'
    PRE_1m.units = 'mm'
    PRE_1m.missing_value = -9999
    PRE_1m._CoordinateAxes = 'lat lon'

    fnc.variables['time'][:] = np.array([i for i in range(1, 25)])
    fnc.variables['sta'][:] = np.array(dictknn.index.tolist())
    fnc.variables['lat'][:] = np.array(dictknn['lat'].tolist())
    fnc.variables['lon'][:] = np.array(dictknn['lon'].tolist())
    fnc.variables['PRE_1m'][:] = np.array(data)

读取nc文件

def read_nc(ncpath: str):
    print('读取 nc 文件 %s' % ncpath)
    f = nc.Dataset(ncpath, 'r')
    # 获取nc文件中的变量名
    var_key = list(f.variables.keys())
    var_len = f.variables['lat'][:].shape
    # print(var_key, var_len)
    # 创建DataFrame存放数据，一列一类数据
    nc_data = pd.DataFrame(index=range(var_len[0]), columns=var_key[1:], dtype='short')
    for e in var_key:
        # 获取变量数据
        vardata = f.variables[e][:]
        size = vardata.shape
        # print('{}-{}'.format(e, size))  # (3541, 6165)   3541行，每行6165个

        if len(size) == 1:
            if e == 'sta':
                # nc_data[e] = vardata
                nc_data.index = vardata
            else:
                nc_data[e] = list(map(lambda s: float('%s' % s), vardata))
        else:
            nc_data[e] = list(vardata)

    # print(f.variables)
    f.close()
    print(nc_data)

    return nc_data

 创建群组

    with nc.Dataset(nc_name, 'w', format='NETCDF4') as fnc:

        # 创建一个群组
        group1 = fnc.createGroup('LatLon_1441X2880')
        group1.createDimension('lat', 1441)  # 确定维度大小
        group1.createDimension('lon', 2880)  # 确定维度大小

        lat1 = group1.createVariable('lat', 'f4', ('lat',), zlib=True, complevel=9, fill_value=-9999)
        lat1.units = "degrees_north"
        lon1 = group1.createVariable('lon', 'f4', ('lon',), zlib=True, complevel=9, fill_value=-9999)
        lon1.units = "degrees_east"
        group1.variables['lat'][:] = data[1441]['lat']
        group1.variables['lon'][:] = data[1441]['lon']
        for elem in data[1441]['value']:
            d = group1.createVariable(elem['name'], 'short', ('lat', 'lon'), zlib=True, complevel=9, fill_value=-9999)
            d.units = elem['units']
            d.missing_value = -9999
            d.add_offset = np.float32(0.0)
            d.scale_factor = np.float32(0.1)
            group1.variables[elem['name']][:] = elem['value']

        group1.close  # 关闭群组， 注意，这里没有括号

        # 创建一个群组
        group2 = fnc.createGroup('LatLon_281X361')
        group2.createDimension('lat1', 281)  # 确定维度大小
        group2.createDimension('lon1', 361)  # 确定维度大小

        lat2 = group2.createVariable('lat', 'f4', ('lat1',), zlib=True, complevel=9, fill_value=-9999)
        lat2.units = "degrees_north"
        lon2 = group2.createVariable('lon', 'f4', ('lon1',), zlib=True, complevel=9, fill_value=-9999)
        lon2.units = "degrees_east"
        group2.variables['lat'][:] = data[281]['lat']
        group2.variables['lon'][:] = data[281]['lon']
        for elem in data[281]['value']:
            d = group2.createVariable(elem['name'], 'short', ('lat1', 'lon1'), zlib=True, complevel=9, fill_value=-9999)
            d.units = elem['units']
            d.missing_value = -9999
            d.add_offset = np.float32(0.0)
            d.scale_factor = np.float32(0.1)
            group2.variables[elem['name']][:] = elem['value']

        group2.close

        fnc.author = 'ecoweather'
        fnc.createdate = '%s' % datetime.datetime.now().strftime('%Y-%m-%d %H:%M')