The Observations and Modeling of the Green Ocean 2014-2015 (GoAmazon2014/5) field campaign is an international collaborated experiment conducted near Manaus, Brazil from January 2014 through December 2015.  This experiment is designed to enable the study of aerosols, tropical clouds, convections and their interactions.  To support modeling studies of these processes with data collected from the GoAmazon2014/5 campaign, we have developed a large-scale forcing data (e.g., vertical velocities and advective tendencies) during the second intensive operational period (IOP) of GoAmazon2014/5 from 1 Sep to 10 Oct, 2014.  The method used in this study is the constrained variational analysis method in which the large-scale state fields are constrained by the surface and top-of-atmosphere observations (e.g. surface precipitation and outgoing longwave radiation) to conserve column-integrated mass, moisture and dry static energy.  To address potential uncertainties in the derived forcing data due to uncertainties in surface precipitation, two sets of large-scale forcing data are developed based on the ECMWF analysis constrained by the two precipitation products respectively from SIPAM radar and TRMM 3B42 products.  Our initial analysis shows large differences in these two precipitation products, which causes considerable differences in the derived large-scale forcing data.  Potential uncertainties in the large-scale forcing data to other surface constraints such as surface latent and sensible fluxes will be explored. The characteristics of the large-scale forcing structures for selected cases will be discussed.