Standoff chemical detection and identification techniques are necessary for ensuring safe exposure to dangerous substances. Molecular fingerprints of unknown chemicals can be measured using wavelength-tunable quantum cascade lasers operating in long-wavelength infrared. In this work, we present a method that can identify liquid chemicals on a reflective substrate via diffuse reflection spectra measurement from 50 cm away and multiple nonlinear regression analysis. Experimental measurements and numerical analyses were conducted for different chemical surface densities and angles of light incidence using diethyl phthalate (DEP) and dimethyl methylphosphonate (DMMP). Candidate substances can be classified using a deep learning model to reduce analysis time.