Breakout Summary Report
ARM/ASR User and PI Meeting
13 - 17 March 2017
From Models to Virtual Observatories using Simulators
16 March 2017
1:45 PM - 3:45 PM
40
Pavlos Kollias and Mariko Oue
16 March 2017
1:45 PM - 3:45 PM
40
Pavlos Kollias and Mariko Oue
Breakout Description
The ARM atmospheric fixed and mobile observatories are designed to provide four-dimensional (4D) observations of clouds and precipitation; they are heavily instrumented with active and passive remote sensors that provide complementary views of the atmosphere. Thus, well known data products (e.g., ARSCL) and retrievals (e.g., multi-Doppler-based wind retrievals) are based on multi-sensor observations. Furthermore, domain-averaged cloud fraction estimates using scanning cloud radars or process studies using radar polarimetric or Doppler spectra observations are based on complex measurements that required an adequate understanding of the sensor characteristics. Simulators offer the possibility to compare measurements and model output at the observational space using the same parameters (sampling errors, sensitivity) inherent in the ARM instruments. The objective of the session is to survey existing or planned instrument simulators applicable to ARM observatories, including applications or instrument simulators for high-resolution modeling (e.g., LASSO) and process studies (e.g., using polarimetric and Doppler spectra simulators). In addition, we will discuss the instrument simulators needs for shallow and deep convection and the development of advanced synthetic products (e.g., virtual ARSCL) from multi-sensor simulators that best represent the capabilities of the ARM observatories.Main Discussion
This breakout session featured six presentations and ample time for discussion. First, Jeffrey Snyder (CIMMS/NSSL) presented forward polarimetric radar simulations using the University of Oklahoma advanced polarimetric radar forward operator and bin microphysics output from a deep convection simulation using the Hebrew University Cloud Model (HUCM). The analysis clearly illustrated the ability of the forward model to emulate Zdr and Kdp columns. Brenda Dolan provided a synopsis of another forward operator (POLARRIS) from the Colorado State University group and how it was used to evaluate hydrometeor ID schemes. Scott Collis presented a short summary of an internal Argonne National Laboratory (ANL) project that studies the island effects at the ARM Eastern North Atlantic (ENA) site using a very high-resolution large-eddy simulation (LES) and a forward model that can reproduce ARSCL. Mariko Oue presented the Brookhaven National Laboratory (BNL)/SBU CR-SIM and highlighted its ability to reproduce ARSCL, lidar (ceilometer and MPL) forward simulations, and MWR radiometer simulations and how these capabilities were used to produce ARM VAPs (ARSCL, 3DVAR winds in deep convection and LWP). Ann Fridlind (NASA/GISS) presented highlights of a recently submitted paper that shows clearly the value of forward radar Doppler spectra simulations in warm clouds for evaluation of cloud microphysical schemes. Finally, Andy Vogelmann discussed observational products useful to evaluate LES such as cloud fraction, entrainment, mass flux, liquid water path, etc. and indicated that simulators like CR-SIM play a key role in developing an objective methodology for comparing models and observations.Key Findings
Jeff Snyder, Scott Collis, and Mariko Oue presented uncertainties in observational products (polarimetry, cloud fraction, LWP, and wind retrieval) using their simulators. Sampling strategy significantly impacts underestimating of the products and larger variability.Jeff Snyder suggested that water fraction and water distribution of mixed-phase particles can significantly affect radar observables. Brenda Dolan proposed improvement of scattering calculations in their simulator (POLARRIS) to consider particle canting angle and axis ratio distributions. The different simulators also use different algorithms and scattering calculation methods for forward simulations and are somewhat dedicated to their interests. It is realized that intercomparison analysis is needed to reveal the uncertainty in simulators. Although it would be possible to estimate dispersion of these simulation results, it would be difficult to decide the best simulator.
Simulated radar/lidar observables strongly depend on large, major particles and have little dependency on small, minor particles. A question about how to deal with the minor particles for model evaluation came up. The Penn State/NASA GISS group is developing a new method to investigate uncertainty in model microphysics schemes using various moments from low to high orders (low-order moments depend more on smaller, high-number-concentration particles). Doppler spectra simulation is useful to evaluate the minor particles, as Ann Fridlind presented.
Needs
- Cloud fraction and LWP simulations have been done. Entrainment and mass flux simulations are additionally desired for LES evaluation.
- The simulators still need improvement in terms of scattering calculations and assumption of particle state (e.g., canting angle, axis ratio, and water fraction).
- Quantification of the uncertainties is needed. LES is useful for the analysis of observational uncertainty estimates. Uncertainties in model, simulator, and observation should be separately estimated.