读取月-日表，变为文本序列

↓

代码如下：

1.读取源数据，添加到空列表，重点对空值的处理，即28，29，30的月份

首先12个月所在列依次读取，然后从31~28依次尝试读取，如果不能转化为float，说明是空的，该月天数没这么多

path_runoff = ''
os.chdir(path_runoff)
inflow_ = pd.read_csv(.csv',encoding='GBK',skiprows=5, usecols=range(2, 14))
inflow_1 = pd.read_csv('.csv',skiprows=4, usecols=range(4, 16))

b = []
for column_name in inflow_.columns[:12]:
    column_data = inflow_[column_name].values

    for num_rows in [31,30, 29, 28]:
        column_data_need = column_data[:num_rows]

        try:
            column_data_need = column_data_need.astype(float)
            break  # 如果成功转换为浮点数，退出循环
        except ValueError:
            continue  # 如果转换失败，继续减少行数再次尝试
    b = np.concatenate((b,column_data_need))
for column_name in inflow_1.columns[:12]:
    column_data = inflow_1[column_name].values

    for num_rows in [31,30, 29, 28]:
        column_data_need = column_data[:num_rows]

        try:
            column_data_need = column_data_need.astype(float)
            break  # 如果成功转换为浮点数，退出循环
        except ValueError:
            continue  # 如果转换失败，继续减少行数再次尝试
    b = np.concatenate((b,column_data_need))

2.把得到的数据添加到指定时间范围的空数据集之中

# 要添加的新时间段
new_start_date = "2019-01-01"
new_end_date = "2020-12-31"
new_date_rng = pd.date_range(start=new_start_date, end=new_end_date)
# 创建一个新的DataFrame，仅有索引部分，用None填充
new_data = {
    1: b
}
new_forcing = pd.DataFrame(new_data, index=new_date_rng)

3.转化为月尺度并保存

outflow_month = outflow_new_1.resample('M').mean()
outflow_month.index = outflow_month.index.to_period('M')
inflow_month.to_csv('.inp',header=None,sep=' ')