Material migration is a crucial issue in thermonuclear fusion devices. To study carbon migration, 13CH4 has been injected through a polished graphite roof-like test limiter in the TEXTOR scrape-off layer. The interpretation of the experimental 13C deposition patterns on the roof limiter surface has been done with the ERO impurity transport code. To reproduce the very low experimental 13C deposition efficiencies with ERO, an enhanced re-erosion mechanism for re-deposited carbon had to be assumed in previous studies. However, erosion by hydrogenic species produced during dissociation of injected 13CH4 was not taken into account by ERO in these studies. This additional erosion could maybe explain the very low experimental 13C deposition efficiencies. Therefore, it is now taken into account in ERO. Also more realistic physical sputtering yields and hydrocarbon reflection probabilities have been implemented in ERO. The simulations with these improvements included clearly confirm the need for enhanced re-erosion of re-deposited carbon.