A04
Precipitation processes
PD. Dr. Silke Trömel
University of Bonn | +49-228-73779 |
Summary
Atmospheric models still do not adequately represent precipitation generating processes and previous studies identify cloud and precipitation processes as one of the largest uncertainties in current weather and climate prediction models. In this project, this incomplete representation of precipitation generating processes is assumed to be at least partly responsible for their inability to reproduce observed regional patterns of drying and wetting. We will quantify these deficiencies by comparing model simulations with reanalyses of terrestrial water storage (TWS) and polarimetric weather radar observations, which carry information on precipitation generating processes. We will focus on Germany -with its only weak TWS trends- and Türkiye -where pronounced negative TWS trends were observed-. Both regions are monitored by modern polarimetric radar networks. With the model simulations provided by other projects, we will explore if deficiencies in modelled precipitation generation explain the differing TWS trends compared to the reanalyses, if land-atmosphere feedbacks are well represented in the model and what is the impact of greenhouse gas forcing and regional anthropogenic interventions on precipitation generation.
Contribution to CRC
The project will contribute to the CRC’s key objectives by investigating deficiencies in precipitation generating processes and feedback mechanisms in the modelling platform, and the impact of anthropogenic modifications on precipitation generating processes and microphysical characteristics. Overall, we will work on the model simulation scenarios and reanalyses provided by other projects and produce a detailed analysis of precipitation characteristics.
Approach
Six years of radar data (2015-2020) for three radar sites in Germany (two in the northeast and one in the southwest) and three radar sites in Türkiye (two in the southeast and one in the northwest) will be obtained. The project will use the services provided by Z03 in Forschungszentrum Jülich for storage and processing of the data. The polarimetric radar forward operator EMVORADO will be used to analyze the model simulations in radar space (i.e., to generate synthetic radar observations from the model simulations). The project will discuss with Z04 a potential implementation of EMVORADO in the modelling platform. Depending on the availability of the model runs, we will either exploit simulations with the current TerrSysMP framework using the COSMO model and CLM3.5 at 12 km resolution for the atmosphere or use the simulations with the Integrated Modelling System (IMS) (ICON-CLM5-ParFlow) at 3 km resolution for the atmosphere scheduled for the end of the first project year, if available.
We will compare model simulations, the IMERG satellite-gauge product and polarimetric radar observations concerning precipitation trends, micro-physical quantities, and precipitation generating process signatures in both Germany and Türkiye and focus on the historic climate scenario simulation with real-world history (greenhouse gases and anthropogenic water use) provided for 1980-2020 by D02. Simulated/observed trends in precipitation will be related to trends in reanalyses of TWS and IMS-simulated TWS; at much higher spatial resolution than the raw GRACE-product data (a satellite-based dataset of TWS). Comparisons between observed and modelled precipitation will elucidate whether insufficiently reproduced precipitation trends in the simulations explain the different trends between the reanalysis of TWS and IMS model output. Microphysical retrievals, e.g. for mean volume diameter, number concentrations, liquid water content (LWC) and ice water content (IWC), will be applied to 3D multivariate polarimetric radar data and compared with simulated counterparts of the IMS to identify biases (in the period 2015-2020 for which we will have radar data). Radar process signatures for precipitation generation will be inspected in both observed and synthetic radar variables to identify potential differences (with the use of EMVORADO).
Finally, will analyze the potential impact of anthropogenic modifications on the hydrometeors and microphysical processes aloft by investigating and comparing additional climate historic model scenarios provided by D02. Statistical differences between simulated rain rates, LWC, IWC, hydrometeor types and sizes for different scenarios will be calculated. Comparisons through contoured frequency or altitude diagrams (CFADs) for different regions and different scenarios will be performed. The comparison of retrievals will again be complemented with comparisons in synthetic radar observation space, providing another perspective and potentially guide further investigations. By comparing the statistics of rain rates, hydrometeor types and sizes, concentration, IWC, etc., between observed and simulated precipitation events with and without pronounced anthropogenic forcing, we can improve our understanding of potential shortcomings of the model to reproduce observed anthropogenic patterns of drying and wetting.
Main Results in 2022 and 2023
The total annual precipitation over Germany and Türkiye is compared in the figure below. Since the new model simulations are still in production, we made preliminary comparisons with a previous TSMP-COSMO-ERA5 Euro-CORDEX 11 km simulation already available. The results compare TSMP, IMERG, the ERA5 reanalysis and, in the case of Germany, RADKLIM (a gauge adjusted radar-derived precipitation product). We can see in Germany that TSMP has some agreement with the other products, although there is an overall overestimation compared to RADKLIM and a slight downward trend. The interannual variability is also higher in TSMP. In Türkiye, TSMP overestimates quite substantially the total precipitation, compared to IMERG and ERA5.
The radar data for three German sites and for five Turkish sites was received and preprocessed (quality check, re-formatting, calibration, etc.). The forward operator EMVORADO was used to derive synthetic radar data for the site in Prötzel in northeastern Germany from the TSMP simulation.
A test case for 12/07/2017 is shown below, comparing the observed radar reflectivity for 1.5 degrees of elevation with the synthetic reflectivity (reflectivity can be directly related to rain intensity). The results show that an 11 km run is not well suited for detailed comparison of precipitation processes since the fine spatial details are lost. We rely on the future 3 km runs to expect better results.
A different way to analyze precipitation events is with quasi-vertical profiles (QVPs), i.e., azimuthally averaged height-time plots of the radar moments at elevation angles between 10-20 degrees. A QVP of the same test case is shown below, for the reflectivity (DBZH) and differential reflectivity (ZDR, related to heavy rain or oriented ice crystals). We see from this plot the evolution of the rain event over time. Again, we see that the coarse spatial resolution together with the coarse temporal (3 h) resolution of the TSMP run is not sufficient for a fair comparison.
Statistical analyses can be derived by summarizing the QVP data for several dates or during a period. Below is an example of DBZH for a 12 degree elevation QVP for 25/07/2017 in Prötzel. Algorithms for detecting the melting layer (the layer below 0 C where ice crystals melt into liquid rain) and to classify the event areas corresponding to stratiform precipitation.
By adding vertical temperature data and aggregating all stratiform events in the 2016-2020 period we can derive statistics of the different polarimetric variables, like shown below. This plot gives us the most frequent values of each radar variable for each 1°C (color shading). The red lines show the median and 10-90 percentiles, while the blue line shows the count of samples. For example, we can see that the melting layer between 0-5°C is typically characterized by high reflectivity and differential reflectivity values and low cross-correlation coefficient values. This kind of plot can also be generated for synthetic radar data and then the average behavior of precipitation can be evaluated.
