contourf, hatching not working on the last plot in a panel plot

Question

Hatching don't work on the last plot in a panel plot. The first and second plots are ploted fine, but the last plot doesn't work. The "DJF_sig" variable is fine, I have attached a previously failed plot where the statistically significant areas are colored yellow. But I don't know why it doesn't work. If anyone knows why, please let me know. enter image description here

import netCDF4 as nc
import numpy as np
import matplotlib.pyplot as plt
import sys
from mpl_toolkits.basemap import Basemap
from matplotlib.patches import Polygon

file_path = './Diff_COBALT_DAILY_DJF_JJA.nc'
a_path='./signi_COBALT_DAILY_DJF_JJA.nc'
ds = nc.Dataset(file_path)
ds_a=nc.Dataset(a_path)
annual_sig=ds_a.variables['annual_sig2'][:]
DJF_sig=ds_a.variables['DJF_sig2'][:]
JJA_sig=ds_a.variables['JJA_sig2'][:]

annual = ds.variables['tos_diff_annual'][:]
DJF = ds.variables['tos_diff_DJF'][:]
JJA = ds.variables['tos_diff_JJA'][:]
yh=ds.variables['yh'][:]
xh=ds.variables['xh'][:]

an_sig_grid = np.where(annual_sig == 2)
an_sig_lon = xh[an_sig_grid[1]]
an_sig_lat = yh[an_sig_grid[0]]

DJF_sig_grid = np.where(DJF_sig == 2)
DJF_sig_lon = xh[DJF_sig_grid[1]]
DJF_sig_lat = yh[DJF_sig_grid[0]]

JJA_sig_grid = np.where(JJA_sig == 2)
JJA_sig_lon = xh[JJA_sig_grid[1]]
JJA_sig_lat = yh[JJA_sig_grid[0]]
xh, yh = np.meshgrid(xh, yh)

fig, axes = plt.subplots(nrows=3, ncols=1, figsize=(12, 12), sharex=True, sharey=True)

def plot_with_hatching(ax, m, sst_data, sig_data):
    x, y = m(xh, yh)
    cs = m.pcolormesh(x, y, sst_data, shading='auto', cmap='coolwarm', vmin=-0.5, vmax=0.5)
    sig_mask = np.ma.masked_where(sig_data != 2, sig_data)
    m.contourf(x, y, sig_mask, colors='none', hatches=['////'],extend='both', alpha=0)

    m.drawcoastlines()
    m.drawcountries()
    m.fillcontinents(color='black')
    m.drawparallels(np.arange(-90., 91., 30.), labels=[1,0,0,0])
    m.drawmeridians(np.arange(-180., 181., 60.), labels=[0,0,0,1])
    return cs
ax1 = axes[0]
m1 = Basemap(projection='cyl', lon_0=180, resolution='c', ax=ax1)
cs1=plot_with_hatching(ax1, m1, annual, annual_sig)
ax1.set_title('Annual')

ax2 = axes[1]
m2 = Basemap(projection='cyl', lon_0=180, resolution='c', ax=ax2)
cs2=plot_with_hatching(ax2, m2, JJA, JJA_sig)
ax2.set_title('JJA')

ax3 = axes[2]
m3 = Basemap(projection='cyl', lon_0=180, resolution='c', ax=ax3)
cs3=plot_with_hatching(ax3, m3, DJF, DJF_sig)
ax3.set_title('DJF')

cbar_ax = fig.add_axes([0.75, 0.15, 0.02, 0.7])
cbar = fig.colorbar(m3.pcolormesh(xh, yh, DJF, shading='auto', cmap='coolwarm', vmin=-0.5, vmax=0.5), cax=cbar_ax, ticks=np.arange(-0.5, 0.6, 0.1))

plt.subplots_adjust(hspace=0.4)

plt.show()

enter image description here

If the "*_sig" variables have "2", these are statistically significant areas.