At first, we give the following bootstrap assumption:
\begin{split}
&\text{\it when $0\leq t\leq t_0$ and integer $N_0\ge 6$,
$|\partial^\kappa\Omega^k\phi|\leq\delta^{\nu_0-|\kappa|}$ holds for $|\kappa|+k\leq N_0$},\\
&\text{\it where $\nu_0\in(1, \nu)$ is any fixed constant.}
\end{split}
For (?) and n\in\mathbb N_0, define the energy
M_n(t)=\sum_{|\kappa|+k\leq n}\delta^{2|\kappa|}\|\partial\partial^\kappa\Omega^k\phi(t,\cdot)\|_{L^2(\mathbb R^2)}^2.
Set w=\delta^{|\kappa|}\partial^\kappa\Omega^k\phi with |\kappa|+k\leq 2N_0-2.
It follows from (?) and integration by parts that
\begin{split}
&\int_0^t\int_{\Sigma_\tau}\partial_t w\,g^{\alpha\beta}\partial_{\alpha\beta}^2 w\,dyd\tau\\
=&\int_{\Sigma_{t}}\big(g^{0\alpha}\partial_\alpha w\,\partial_t w
-\frac12g^{\alpha\beta}\partial_\alpha w\,\partial_\beta w\big)dy
-\int_{\Sigma_{0}}\big(g^{0\alpha}\partial_\alpha w\,\partial_t w
-\frac12g^{\alpha\beta}\partial_\alpha w\,\partial_\beta w\big)dy\\
&+\int_0^t\int_{\Sigma_\tau}\big(-(\partial_\alpha g^{\alpha\beta})\partial_\beta w\,\partial_t w
+\frac12(\partial_t g^{\alpha\beta})\partial_\alpha w\,\partial_\beta w\big)dyd\tau,
\end{split}
where
\begin{split}
|g^{\alpha\beta}&\partial_{\alpha\beta}^2w|
\lesssim\delta^{|\kappa|}\sum_{\substack{
|\kappa_1|+|\kappa_2|\leq|\kappa|,\ k_1+k_2\leq k,\\
|\kappa_2|+k_2+2<|\kappa|+k}}
|\partial^{\kappa_1}\Omega^{k_1}g^{\alpha\beta}|
\cdot|\partial^{\kappa_2}\partial^2\Omega^{k_2}\phi|.
\end{split}
Combining (Proof)–(Proof) with Gronwall’s inequality yields that for 0\leq t\leq t_0,
M_{2N_0-2}(t)\lesssim M_{2N_0-2}(0)\lesssim\delta^{2\nu-1}.
Applying the Sobolev embedding on the circle \mathbb S_r (centered at the origin with radius r), we arrive at
|w(t,y)|\lesssim \frac{1}{\sqrt{r}}\|\Omega^{\le 1}w\|_{L^2(\mathbb S_r)}.
Due to r\sim 1 for t\in[0,t_0] and (t,y)\in\mathrm{supp}\,w, then one has that for |\kappa|+k\leq N_0,
|\partial^\kappa\Omega^k\phi(t,y)|
\lesssim \|\Omega^{\le 1}\partial^\kappa\Omega^k\phi\|_{L^2(\mathbb S_r)}
\lesssim\delta^{1/2}\|\partial\Omega^{\le 1}\partial^\kappa\Omega^k\phi\|_{L^2(\Sigma_{t})}
\lesssim\delta^{\nu-|\kappa|}.
Therefore, for small \delta>0 and \nu_0<\nu, the bootstrap assumption (Proof) can be closed by the continuity argument.
Based on (Proof), we now improve the L^\infty estimate of \phi on the domain D_1=\{(t,y): 0\le t \le t_0,
\frac{1}{\mathcal A} - t\le r \le t+\frac{ 1}{\mathcal A}\}.
The equation in (?) can be rewritten as
L\underline{L}\phi
= \frac{1}{2r}L\phi
- \frac{1}{2r}\underline{L}\phi
+ \frac{1}{r^2}\Omega^2\phi
+ 2g^{0i}\partial_{0i}^2\phi
+ (g^{ij}-m^{ij})\partial_{ij}^2\phi,
where m^{ii} = 1 for i=1, 2 and m^{ij} = 0 for i\neq j.
It follows from (Proof) and (Proof) that
|\underline{L}L\partial^q\Omega^k\phi(t,y)|
\le \delta^{2\nu-3-|q|},
\quad 0 \le t \le t_0.
Note that \phi vanishes on the surface \{(t,y) : r- t = \frac{1}{\mathcal A}\}, then integrating (Proof) along
the integral curves of \underline{L} yields
|L\partial^q\Omega^k\phi(t,y)|
\le \delta^{2\nu-2-|q|},
\quad (t,y) \in D_1.
By applying L repeatedly to both sides of (Proof) and using an induction argument,
we conclude that for m\in\Bbb N,
|L^m\partial^q\Omega^k\phi(t,y)|
\le \delta^{\nu-m(2-\nu)-|q|},
\quad (t,y) \in D_1,
which yields (2.1).
Similarly, for any (t,y) \in D_2=\{(t,y): 0\le t \le t_0,
\frac{1-\delta}{\mathcal A} - t \le r \le t+\frac{ 1-\delta}{\mathcal A}\},
|\underline L^m\partial^q\Omega^k\phi(t,y)|
\le \delta^{\nu-m(2-\nu)-|q|},
and hence (2.1) is proved.
For any fixed \omega\in\Bbb S, since both (0,(r_0-t_0)\omega) and (t_0, r_0\omega) lie on the line \{(t,r\omega): r-t=r_0-t_0\},
then it follows from (Proof) that
\begin{split}
&\underline{L}\partial_{0}\phi(t_0,r_0\omega)
=\underline{L}\partial_{0}\phi(0,(r_0-t_0)\omega)+\int_{(0,(r_0-t_0)\omega)}^{(t_0,r_0\omega)}L\underline{L}\partial_{0}\phi\\
=&\delta^{\nu-2}\mathcal{A}\big(\mathcal{A}\partial_s^2\Phi_0\big(\frac{\mathcal{A}(r_0-t_0)-1}{\delta}, \omega\big)-\partial_s\Phi_1\big(\frac{\mathcal{A}(r_0-t_0)-1}{\delta}, \omega\big)\big)+O(\delta^{2\nu-3}),
\end{split}
which yields (2.1). Analogously, (2.1) can be obtained
by integrating (Proof) along the integral curves of \underline L.