MetPy 1.6+散度计算逻辑变更咨询：原导数求和方法失效

• 发现MetPy 1.6及以上版本中散度计算存在潜在问题，需明确其计算逻辑以排查问题。2019年使用时，mpcalc.divergence(u,v)的结果与mpcalc.first_derivative(u, axis=1) + mpcalc.first_derivative(v, axis=0)一致，但当前两者结果存在明显差异。
• 官方文档仅给出散度公式：$$\nabla \cdot U = \frac{\partial u}{\partial x} + \frac{\partial v}{\partial y}$$，作为非软件工程师，难以梳理冗长的源码。
• 本次测试使用from metpy.cbook import example_data获取的示例数据，基于官方文档示例代码进行验证，并通过Cartopy绘制了误差对比图，测试代码如下：

import numpy as np
import cartopy.crs as ccrs
import cartopy.feature as cfeature
import shapely.geometry as sgeom
import cartopy.io.shapereader as shpreader
import matplotlib.pyplot as plt
plt.rcParams['text.usetex'] = True
import matplotlib.ticker as mticker

import metpy.calc as mpcalc
from metpy.cbook import example_data

# load example data
ds = example_data()
u = ds.uwind;v = ds.vwind;
lon = ds.lon.values; lat = ds.lat.values
# Calculate the total deformation of the flow
div = mpcalc.divergence(ds.uwind, ds.vwind).values

div3 = mpcalc.first_derivative(u,axis = 1).values+mpcalc.first_derivative(v,axis = 0).values
error = (div3-div)/div*100
ax3 = plt.axes(projection=ccrs.Orthographic(265-360,35))
shapename = 'admin_1_states_provinces_lakes'
states_shp = shpreader.natural_earth(resolution='110m',
                                         category='cultural', name=shapename)
cf = ax3.contourf(lon, lat, error,range(-100, 100, 10), cmap=plt.cm.bwr_r,transform=ccrs.PlateCarree())
plt.colorbar(cf, pad=.15, aspect=50, label = '%error', orientation = 'vertical')

ax3.set_title('Error between Divergence and Computed Divergence')
ax3.add_feature(cfeature.STATES, edgecolor='black', linewidth=2)
ax3.coastlines()
gl = ax3.gridlines(crs = ccrs.PlateCarree(),draw_labels = True)
gl.xlocator = mticker.FixedLocator([270-360,265-360,260-360])
gl.ylocator = mticker.FixedLocator([30,32.5,35,37.5])

plt.show()

• 需求：获取mpcalc库中散度计算的具体实现代码；此前参考的导数计算相关文章目前无法访问，需替代信息。

内容的提问来源于stack exchange，提问作者Researcher R