we study a cooperative coding scheme for densely deployed wireless sensor networks (WSNs) where a number of sensors transmit data to a single destination with the help of a relay. The latter applies algebraic network coding to the source codewords and forwards only the additional redundancy to the destination that observes a product code matrix built based on source codewords and relay-generated redundancy. However, for such an application two types of correlation can be found between the different sensors' observations. The first type is due to the high density of the WSN that results in a correlation between observations delivered by neighbor sensors, and we call it spatial correlation. The nature of the measured physical phenomena induces also some correlation between successive observations of the same sensor, and this type is called temporal correlation. Spatial and temporal correlations represent extra source information that was neglected in previous contributions dealing with cooperative communications. In this contribution, we investigate a joint source channel decoding scheme that exploits the source memory structure to improve the product code iterative decoding performance. Significant performance improvements are demonstrated depending on the spatial and temporal correlation level. A performance gain achieving 0.8 dB for the additive white Gaussian noise (AWGN) channel, and 1.5 dB for the fast Rayleigh fading channel are demonstrated.