This study investigates the spatial and temporal distribution of lightning channels in relation to radar-derived 3D kinematic and microphysical storm structures toward the goal of better understanding the production and convective transport of LNOx. Four cases are analyzed from the Deep Convective Clouds and Chemistry (DC3) field campaign (May-June 2012), two each in northern Colorado and Alabama. Lightning sources are combined into 3D flash channels using the Colorado and Northern Alabama lightning mapping arrays and attributed to each storm. Dual-Doppler syntheses are performed using the CSU-CHILL and Pawnee radars (Colorado) and Armor and MAX radars (Alabama) to retrieve the 3D wind components for each storm. Dominant hydrometeor types are calculated using a HID algorithm with dual-polarization radar measurements serving as requisite inputs. All radar variables are gridded to match the 3D flash channel resolution. Flash initiation locations and respective channels are spatially analyzed with respect to storm regions such as convective updrafts and downdrafts, reflectivity, and dominant hydrometeor type. Preliminary results suggest a large percentage of flash channels reside not just within the 20 dBZ reflectivity volume but also in regions of low- and high-density graupel above the melting level. LNOx transport efficiency is investigated by first producing LNOx along the observed flash channels and then advecting the concentrations forward according to the dual-Doppler derived winds.