WATER VAPOR ESTIMATION USING GPS

Abstract

Water vapor shares only up to 4% volume of the atmosphere (Navarra, 1979), but is the largest contributor towards weather phenomena (global & regional), climate, hydrologic cycle and global warming. Direct measurement of precipitable water vapor is not possible, owing to it's highly variable and asymmetric distribution with both season and location. Therefore, it is obtained from indirect measurements of radiosonde, water vapor radiometer and satellite based sensors, but due to their own drawbacks and advantages of GPS over them, later is gradually replacing the former. GPS gives promise of all weather operation, economy and high temporal resolution with accuracy. Atmospheric layers ionosphere and neutral atmosphere complicates the GPS signal propagation by causing delay of GPS signal and thereby deteriorating the performance. GPS aided remote sensing of atmospheric constituents is thus based on the GPS signal propagation delay modeling. With the assumption of wet delay being caused only due to water vapor, precipitable water vapor amount can be related to zenith wet delay by proportionality constant. Performance of models varies with location and season both, therefore need of development of regional models has been growing rapidly. Additionally, estimation of PWV from GPS is highly influenced by any variations in processing methodology of GPS observations. Focus of present thesis is on application of ground based dual frequency receivers with network based post-processing mode in estimating precipitable water vapor. Objective is focused on methodology of precipitable water vapor estimation using GPS and validating it on, experimental data from both external source and field observations. Precipitable water vapor has been estimated for Bangalore, from raw observations and precise orbits obtained through IGS archive. PWV estimation has been performed for 140 days between October, 2006 and May, 2007. Precise station coordinates have been obtained from IGS. Additional IGS stations are included in the network processing. Comparison of results with radiosonde derived PWV shows an agreement of 7.68 mm between the two. Additionally, assessment of various zenith hydrostatic delay models and mean temperature models has been performed in order to get optimum results. Results show insignificant difference between performances of different models. Further, adopting similar methodology tested above, hourly estimates of PWV have been computed for IIT Roorkee, from observations taken' for four days using a dual frequency geodetic quality GPS receiver. Results have been confirmed by comparing them with surface relative humidity values. In the absence of validation data, results at IIT Roorkee have not been checked for accuracy. By including more local GPS stations, spatial variations in PWV can be mapped with high temporal resolution in order to assist in applications of local weather forecasting, landslide hazard monitoring & irrigation. iv