Data management is an important issue in wireless sensor networks. However, the data gathering process may be unsuccessful in disconnected or random networks: some data gathered cannot be delivered to the sink, coverage holes may occur and result in missing data. To cope with this problem, we propose two distributed redeployment algorithms (DVFA and ADVFA), based on virtual forces, that provide a uniform deployment of sensor nodes. Both ensure a full area coverage and network connectivity. Hence, an accurate data gathering can be done. For that purpose, we define a three-tier architecture where after the execution of a redeployment algorithm, sensor nodes are uniformly deployed to monitor a temporary worksite. A uniform deployment provides a better balancing of the routing tree (used to report the data gathered) leading to smaller data gathering delays. We also show how to save energy during the data gathering phase. We compute the optimal number of sensors needed to cover an area. Using this result, we parameterize DVFA and evaluate its performances. We then propose ADVFA to cope with DVFA drawbacks (e.g. node oscillations greedy in energy). ADVFA adapts the target distance between two neighbors to the number of operational nodes discovered. ADVFA outperforms DVFA, considerably reducing the distance traveled by nodes and then maximizing network lifetime by saving energy.