Inverse source problems with reduced interior data for a coupled reaction-diffusion system

1. Problem Formulation

The paper addresses the Inverse Source Problem (ISP) for the Klausmeier-Gray-Scott model, a coupled semilinear reaction-diffusion system describing the interaction between water density ($u$) and vegetation biomass density ($v$) in a bounded domain $\Omega \subset \mathbb{R}^n$.

The system is governed by:
$$\begin{cases} \partial_t u = d_1 \Delta u - u v^2 - a_1(x, t) u + f(x), & (x, t) \in Q = \Omega \times (0, T) \\ \partial_t v = d_2 \Delta v + u v^2 - a_2(x, t) v, & (x, t) \in Q \\ u = g, \quad v = h & \text{on } \Sigma = \partial\Omega \times (0, T) \\ u(\cdot, 0) = u_0, \quad v(\cdot, 0) = v_0 & \text{in } \Omega \end{cases}$$
Key Features:

• Unknown Source: $f(x) \in L^2(\Omega)$ is an unknown spatially dependent source (e.g., external water supply) appearing only in the first equation.
• Nonlinearity: The coupling term $uv^2$ represents the facilitative feedback where vegetation enhances water uptake, and water converts to biomass.
• Goal: Determine $f(x)$ from limited interior data on $(u, v)$.
• Motivation: In ecological monitoring, measuring water content is often sparse compared to vegetation (remote sensing). The authors investigate whether source recovery is possible with reduced data (e.g., observing only $v$ or only $u$ in the interior).

2. Methodology

The primary mathematical tool employed is the Carleman estimate, a weighted integral inequality used to establish uniqueness and stability for ill-posed inverse problems. The authors utilize two distinct types of Carleman estimates:

A. Singular Weight Carleman Estimate (for Lipschitz Stability)

• Weight Function: Uses a weight $\phi(x,t) = e^{\lambda d(x)} / \theta(t)$ where $\theta(t)$ vanishes at the endpoints of the observation interval $I = (t_1, t_2)$.
• Mechanism: The singularity at the endpoints ensures that boundary terms vanish during integration by parts. This allows for the derivation of a Lipschitz stability estimate over the entire domain $\Omega$.
• Application: Used to prove Theorem 2.1, where full data (snapshots and time-resolved observations of both components) is available.

B. Regular Weight Carleman Estimate (for Hölder Stability)

• Weight Function: Uses a regular weight $\psi(x,t) = e^{\lambda(d(x) - \beta(t-t_0)^2)}$ which does not blow up at the boundaries.
• Mechanism: Since the weight is bounded, boundary and temporal-endpoint terms do not vanish. These terms are controlled using a priori bounds (assuming solutions belong to specific Sobolev spaces).
• Application: Used to prove Theorem 2.2, establishing Hölder stability on any compactly contained interior subdomain $\Omega_0 \subset \subset \Omega$ without requiring boundary data.

C. Data Reduction Strategy

The authors exploit the specific algebraic structure of the coupling term $uv^2$. By analyzing the difference system between two solutions, they show that under positivity conditions (guaranteed by the maximum principle if initial/boundary data are non-negative and non-trivial), one component can be algebraically recovered from the other in the observation region $\omega \times I$. This allows the reduction of observation data while maintaining uniqueness.

3. Key Contributions and Results

A. Stability Estimates

1. Lipschitz Stability (Theorem 2.1):

   • Data: Full snapshot $u(\cdot, t_0)$ on $\Omega$ and time-resolved observations of both $u$ and $v$ on a subdomain $\omega \times I$.
   • Result: $\|f - \tilde{f}\|_{L^2(\Omega)} \leq C (\text{data discrepancies})$.
   • Significance: Provides a strong stability guarantee over the whole domain, provided boundary data is known.

2. Hölder Stability (Theorem 2.2):

   • Data: Same as above, but no boundary data is assumed.
   • Result: $\|f - \tilde{f}\|_{L^2(\Omega_0)} \leq C (\text{data discrepancies})^\theta$ for any $\Omega_0 \subset \subset \Omega$, with $\theta \in (0, 1)$.
   • Significance: Addresses practical scenarios where boundary conditions are unknown, trading global Lipschitz stability for local Hölder stability.

B. Reduced Data Uniqueness (Corollaries 2.1–2.3)

The paper proves that the source $f$ can be uniquely determined even if the observation data is significantly reduced, provided the solution components remain positive:

• Case M2 (Observation of $v$ only): If $u(\cdot, t_0)$ is known and $v$ is observed in $\omega \times I$, $f$ is unique. (Relies on $v > 0$ to recover $u$ via $uv^2$).
• Case M3 (Observation of $u$ with $f|_\omega$ known): If $u$ is observed and $f$ is known on the subdomain, $f$ is unique globally.
• Case M4 (Two Snapshots): If snapshots of both $u$ and $v$ at $t_0$ are known, and $u$ is observed in $\omega \times I$, $f$ is unique.

C. Technical Innovation

• Integral Relation Handling: In the proof of Theorem 2.2, the authors differentiate the system to eliminate the source $F$ from the time-differentiated equations. They handle the remaining $u$ term (which contains the source) via the integral relation $y = \int_{t_0}^t y_1 d\eta + y_0$, allowing them to apply Carleman estimates to a reduced set of functions ($y_1, z_1, z$) rather than the full set.
• Positivity Arguments: The use of the Strong Maximum Principle to ensure $u, v > 0$ in the observation region is crucial for the algebraic recovery of missing data.

4. Significance and Impact

• First ISP for Klausmeier-Gray-Scott: This is the first work to establish rigorous stability and uniqueness results for the inverse source problem in this specific ecological model.
• Practical Data Reduction: The results are highly relevant for ecological monitoring. They demonstrate that expensive or difficult measurements (e.g., soil moisture $u$) can potentially be omitted if the other component ($v$) is observed, provided the system is in a non-degenerate state (positive biomass/water).
• Robustness to Boundary Uncertainty: The Hölder stability result (Theorem 2.2) is significant because it removes the unrealistic requirement of knowing boundary values on the entire domain, which is often impossible in field studies.
• Methodological Extension: The paper extends the Carleman estimate framework to coupled nonlinear systems with quadratic coupling, offering a template for analyzing other reaction-diffusion systems in ecology and chemistry.

5. Future Directions

The authors suggest several avenues for future research:

• Extending the results to cases where unknown sources appear in both equations.
• Incorporating advection terms (transport) to model water flow on hillslopes more accurately.
• Developing numerical algorithms for source reconstruction based on these stability estimates.
• Investigating whether the positivity conditions (non-degeneracy) can be relaxed.