The widespread use of advanced electronic devices has led to the emergence of spatial crowdsourcing, a method that taps into collective efforts to perform real-world tasks like environmental monitoring and traffic surveillance. Our research focuses on a specific type of spatial crowdsourcing that involves ongoing, collaborative efforts for continuous spatial data acquisition. However, due to limited budgets and workforce availability, the collected data often lacks completeness, posing a data deficiency problem. To address this, we propose a reciprocal framework to optimize task assignments by leveraging the mutual benefits of spatiotemporal subtask execution. We introduce an entropy-based quality metric to capture the combined effects of incomplete data acquisition and interpolation imprecision. Building on this, we explore a quality-aware task assignment method, corresponding to spatiotemporal assignment strategies. Since the assignment problem is NP-hard, we develop a polynomial-time algorithm with the guaranteed approximation ratio. Novel indexing and pruning techniques are proposed to further enhance performance. Extensive experiments conducted on datasets validate the effectiveness of our methods.