Electromagnetic probes such as direct photons and dileptons are crucial to study the properties of a Quark-Gluon Plasma (QGP) created in heavy ion collisions. Based on the (3+1)-dimensional event-by-event viscous hydrodynamic model EPOS 3.102, we calculated the anisotropic emission of thermal photons and dileptons in Au+Au collisions at the Relativistic Heavy Ion Collider (RHIC) energy √ sNN = 200 GeV. Thermal emissions from both QGP phase and hadronic gas phase are considered, with AMY rate for photons and Lattice QCD based rates for dileptons in QGP phase.For emission from hadron gas phase, rates based vector meson dominant model are used for both photons and dileptons. Non-thermal contribution to direct photons is calculated with next to leading order QCD. STAR cocktail data are directly used for non-thermal contribution to dileptons. With the same space-time evolution of the collision systems, the two penetrating probes, photons and dileptonsshow some consistency, ie, the emission of both thermal photons and dileptons are underestimated, compared with the transverse momentum spectra of direct photons measured by PHENIX collaboration and the invariant mass spectra of dileptons measured by STAR collaboration, for all centrality classes. With a good constraint of anisotropy of the plasma via the elliptic flow and triangular flow of charged hadrons, the resulted elliptic flow and triangular flow of direct photons agree with PHENIX measurements reasonably well. Thus we made predictions to the flows of thermal dileptons. The elliptic flow of thermal dileptons is predicted larger than the available results from other models, and comparable to the STAR measurement referring to all dileptons (thermal + cocktail) for minimal bias collisions.