The seislet transform uses a prediction operator which is connected to the local slope or frequency of seismic events. In this paper, we propose combining the 1D non-stationary seislet transform with empirical mode decomposition (EMD) in the $f-x$ domain. We use the EMD to decompose data into smoothly variable frequency components for the following 1D seislet transform. The resultant representation shows remarkable sparsity. We introduce the detailed algorithm and use a field example to demonstrate the application of the new seislet transform for sparsity-promoting seismic data processing.