An enhanced dual-layer wide area wireless sensor network (WSN) system was developed and applied in fresh food tracking application. The sensor nodes of WSN provide a set of useful measurements about the environmental conditions during the fresh food transportation. Since the 3-axis acceleration data in such application account for more than 99% out of all collected source data, we propose a novel acceleration data compression scheme for this specific application, to overcome the resource and cost limitation imposed on embedded sensing nodes and wireless communication links in WSN. According to general fresh food transport behavior pattern, the acceleration data can be extracted into three components (D waveform, A waveform and P waveform). Each component could be compressed separately according to its distinct characteristic and then packetized as an encoded frame. In this paper we examine the experimental acceleration dataset from a field test (fresh melon fruit transport from Brazil to Sweden in 20 days) as a test case. Experimental results show that a high compression ratio (around 10:1) with the acceptable distortion could be achieved, confirming the effectiveness of the compression scheme proposed in this work.