Register now After registration you will be able to apply for this opportunity online.
Implementing a dynamic noise model to improve drone-based GNSS RTK tropospheric estimation
A dynamic tropospheric process noise model will be implemented into GNSS real time kinematic (RTK) algorithms to improve the estimation of drone-based GNSS zenith total delays (ZTDs).
Keywords: GNSS, drone, tropospheric process noise, atmospheric monitoring
Global Navigation Satellite Systems (GNSS) provide positioning services and help retrieve tropospheric information by estimating zenith total delays (ZTDs). The benefit of ZTD values is that they can be assimilated into numerical weather models and improve weather forecasts. Typically, data from static GNSS ground stations are used for this purpose. In this project, we want to explore the feasibility of using GNSS data collected on a drone (UAV) to determine ZTD. For this purpose, we will use Real Time Kinematic (RTK), a relative positioning method, and improve its stochastic model. Unlike static ground stations, the height difference between the reference station and the vertically moving drone changes significantly, making the traditional constant tropospheric process noise not optimal. To better reflect and model real ZTD changes, a dynamic tropospheric process noise model should be implemented, which we expect to improve ZTD estimation accuracy. However, the GNSS community has not yet investigated this possibility, offering an opportunity for us to improve upon the current state-of-the-art.
**Figure Description**
Figure 1 Illustration of drone-based GNSS RTK. (https://drotek.gitbook.io/rtk-f9p-positioning-solutions/what-is-rtk)
Global Navigation Satellite Systems (GNSS) provide positioning services and help retrieve tropospheric information by estimating zenith total delays (ZTDs). The benefit of ZTD values is that they can be assimilated into numerical weather models and improve weather forecasts. Typically, data from static GNSS ground stations are used for this purpose. In this project, we want to explore the feasibility of using GNSS data collected on a drone (UAV) to determine ZTD. For this purpose, we will use Real Time Kinematic (RTK), a relative positioning method, and improve its stochastic model. Unlike static ground stations, the height difference between the reference station and the vertically moving drone changes significantly, making the traditional constant tropospheric process noise not optimal. To better reflect and model real ZTD changes, a dynamic tropospheric process noise model should be implemented, which we expect to improve ZTD estimation accuracy. However, the GNSS community has not yet investigated this possibility, offering an opportunity for us to improve upon the current state-of-the-art.
**Figure Description**
Figure 1 Illustration of drone-based GNSS RTK. (https://drotek.gitbook.io/rtk-f9p-positioning-solutions/what-is-rtk)
In this study, the student will implement a dynamic tropospheric process noise model based on the existing research (Zhang et al., 2022) in real-time mode using weather forecast products and post-processing mode using atmospheric reanalysis products. After that, the model will be implemented into a GNSS RTK processing software such as the CamaliotGNSS software (KÅ‚opotek et al., 2024), for which the student would ideally program in C language (with the support of the supervisors). To evaluate the model, a drone will be used to collect GNSS data with significant height changes, with meteorological observations onboard the drone as references. Compared to the traditional constant noise model, the new model is promising to estimate GNSS ZTDs with higher accuracy and stability. Upon successful completion of this study, the improved drone-based GNSS ZTDs could be assimilated into numerical weather forecasting, producing better weather forecasts and benefiting the field GNSS meteorology.
In this study, the student will implement a dynamic tropospheric process noise model based on the existing research (Zhang et al., 2022) in real-time mode using weather forecast products and post-processing mode using atmospheric reanalysis products. After that, the model will be implemented into a GNSS RTK processing software such as the CamaliotGNSS software (KÅ‚opotek et al., 2024), for which the student would ideally program in C language (with the support of the supervisors). To evaluate the model, a drone will be used to collect GNSS data with significant height changes, with meteorological observations onboard the drone as references. Compared to the traditional constant noise model, the new model is promising to estimate GNSS ZTDs with higher accuracy and stability. Upon successful completion of this study, the improved drone-based GNSS ZTDs could be assimilated into numerical weather forecasting, producing better weather forecasts and benefiting the field GNSS meteorology.
Zhenyi Zhang (zhenyzhang@ethz.ch)
Yuanxin Pan (yxpan@ethz.ch)
Prof. Benedikt Soja (soja@ethz.ch)
Zhenyi Zhang (zhenyzhang@ethz.ch) Yuanxin Pan (yxpan@ethz.ch) Prof. Benedikt Soja (soja@ethz.ch)