The extent of road slippery is an important indicator of road traffic safety. The road surface meteorological sensing technologies can accurately detect the conditions of water accumulation and ice formation on the road, and thus can quantitatively analyze the slippery parameters such as critical hydroplaning velocity and road adhesion coefficient. Using these technologies, we can effectively prevent traffic accidents by adjusting vehicle speed or even closing roads in real time. The non-contact road surface meteorological sensing method is currently a hot research topic because of its high accuracy, no damage to roadbeds and flexible installation. However, its key technology is still monopolized abroad and the equipment is very expensive. Based on the characteristics of infrared reflection spectra of stagnant water and ice on road surfaces, we use the reflected laser intensities at 1310 nm and 1550 nm to study a set of mathematical models and detection methods for detecting water film thickness and distinguishing water/ice state. Based on it, a road surface meteorological sensing system is established and the optical structure of this system is optimized by optical simulation. The experimental results show that the detection distance of the system reaches 3-5 m and the measurement error of water film thickness is less than 0.1 mm in the range of 9 mm. The system can accurately distinguish the meteorological states of drying, water accumulation and ice formation on road surfaces, which has good practical value.