This research focuses on developing optical sensing systems for 2D and 3D spatial monitoring of temperature and concentration distribution profiles of complex or reacting gas flows. Non-invasive, species specific and sensitive nature of monitoring allows spatial information to be extracted from harsh environments with poor physical access, allowing validation of computational models or process monitoring. This is suitable for processes like combustion engines or sealed atmospheric cloud chambers.A novel line-of-sight (LOS) Tunable Diode Laser Absorption Spectroscopy(TDLAS) system using a preselected laser diode centred at 7212.88 cm-1 was first designed to monitor the change of relative humidity (water vapour concentration) during an expansion process within the Manchester Ice Cloud Chamber (MICC), operating from atmospheric pressure, down to 0.7 atm. The experimental results were validated with an Aerosol Cloud Precipitation Interaction Model (ACPIM) simulation, feasible for tomography applications.The MICC shares similar combustion monitoring challenges such as minimal optical access or reactive gas flows. The TDLAS system developed for the MICC was then used as a foundation design for a TDLAS tomography setup capable of conducting temporal two-dimensional (2D) and three-dimensional (3D) concentration and temperature imaging. This system uses the principle of two-line thermometry, centred within the near infrared (NIR) region of 7181.93cm-1 and 7179.8 cm-1. The laser was divided into 4 simultaneous parallel beams using a 1 × 4 fiber coupler (4 LOS). Using a motorised platform, the beams were projected at 0.5° interval, from 0° to 179° angle within 3.6 s, around the exhaust of two asymmetrical shaped flame burners. A total of 360 projection slices comprised of 1440 integrated absorbance data were used per tomogram reconstruction. By solving for the spatial distribution of temperature first, the concentration distribution of water vapour could be then calculated.Reconstruction algorithms (Filtered Back Projection, Fourier Slice Reconstruction and Direct Fourier Reconstruction (DFR)) were compared using a range of criteria. The DFR method was selected as the best method at 700 zero padding, with a spatial in-plane resolution of 1-2 lp/cm, pixel resolution of 128 by 128, thermocouple temperature validations of ±5°C and a relative mean error performance of 8.12%. The concentration could not be validated due to the lack of a mass spectrometer.3D volumetric monitoring results took 36 seconds to complete, and was constructed using 10 interpolated parallel, 1 cm height interval spaced tomograms. Independent vertical slices along the x-axis and y-axis could also be extracted. The temporal results were also successfully conducted and consisted of a quick succession of 16 experiments at a temporal resolution of 0.28 frames per second.A tomographic system that performs 3D and 2D temporal sensing was successfully developed and validated. Although 3D work was conducted using planar imaging or hyperspectral tomography, no work has been conducted so far using NIR TDLAS systems to date.