Data Assimilation & Prediction

Evaluation of Small-Satellite Architectures to Address the Future Needs of the NOAA Enterprise and its Stakeholders

Project Period: September 1, 2018 to August 30, 2019
Principal Investigator(s): Chris Kummerow (CSU/CIRA) and Sid Boukabara (NOAA/NESDIS/StAR)

Investigator(s): Heather Cronk, Milija Zupanski, Ting-Chi Wu, Anton Kliewer, Phil Partain, Wes Berg, Steve Miller, Lewis Grasso, and Haidao Lin

Sponsor(s): National Oceanic and Atmospheric Administration (NOAA)


The goal of this research is to explore quick and agile methodologies to entrain small-satellites that have limited lifetimes into the NOAA processing stream.   The goal is to develop workflows that would allow NOAA, once it has identified an upcoming mission, to work with partners to ingest, calibrate, validate, and exploit these data in a minimum amount of time. The objective is twofold in that it would allow NOAA to better exploit upcoming constellations of very small satellites that have a very limited lifetime, as well as help NOAA assess the full utility of some small constellations of satellites being considered by the NOAA Satellite Observing System Architecture (NSOSA) study team.  In order to exercise these methodologies, the project proposes to use a constellation of hyperspectral mid-wave IR sounders, a microwave humidity sounder from the TEMPEST-Demonstration radiometer to be launched in June 2018, and wind profiles from the European Space Agency’s ADM Aeolus sensor, to be launched in fall 2018. These three diverse examples of non-operational, short duration missions, are typical of what may become available and thus useful to define the workflows for future missions.

Enabling Cloud Condensate Cycling for All-Sky Radiance Assimilation in HWRF

Project Period: September 1, 2018 to August 31, 2020
Principal Investigator(s): Ting-Chi Wu
Co-Principal Investigator(s): Milija Zupanski and Lewis Grasso
Sponsor(s): National Oceanic and Atmospheric Administration (NOAA)


NOAA’s operational HWRF excludes assimilation of cloudy and precipitation affected satellite radiances. Instead, clear-sky radiances are assimilated. Since clouds are part of the dynamics of hurricane processes, subsequent neglect of clouds during a data assimilation process may have adverse consequences to the prediction of hurricane track and intensity. One main goal of the proposed research aims to improve hurricane forecasts by creating an improved initial state via enabling cloud condensate cycling in the HWRF system in order to facilitate the assimilation of all-sky satellite radiances.


To achieve the proposed goal, modifications to the following components of HWRF are required: 1) Vortex Improvement (VI; also known as Vortex Initialization or Vortex Relocation), 2) data assimilation, which uses the hybrid Gridpoint Statistical Interpolation (GSI), and 3) Merge (MG), a procedure that interpolates and merges resulting analyses to all domains in an HWRF forecast. As of this writing, cloud condensate variables are excluded during both the VI and MG steps. Specifically, values of cloud condensate are intentionally set to zero during both the VI and MG procedures. Although hybrid GSI has the general capability to include cloud condensate updates, exclusion of condensate variables in both the VI and MG procedures prevents condensate updates in an analysis field within an HWRF cycle. As a result, new development is necessary in order to assimilate all-sky satellite radiances in HWRF.


In the proposed work, the following modifications to the HWRF components that include 1) the VI, 2) the hybrid GSI, and 3) the MG are essential. Modifications to VI will add the inclusion of non-zero values of cloud condensate, which will be from a 6-h HWRF forecast of a previous cycle. Subsequently, a resulting background field will contain information of cloud condensate. Modifications to the hybrid GSI will allow the inclusion of cloud condensate variables to an existing set of control variables. Therefore, cloud information will be updated via assimilation of observed satellite radiances. Modifications to MG will include the interpolation and merging of cloud condensate information from the resulting analyses to all HWRF forecast domains. As a result, an initial state with updated cloud condensate information will be used for an HWRF forecast. One priority of the proposed work will be to use all-sky radiances from the Advanced Technology Microwave Sounder (ATMS) in order to align with similar efforts at Environmental Modeling Center (EMC), which pertain to global all-sky assimilation. Finally, a tropical cyclone case will be selected to conduct experiments using the modified HWRF system with cloud condensate cycling via the assimilation of ATMS radiances. Results from the experiments will be evaluated using standard tropical cyclone forecast metrics and data assimilation techniques.




Establishing Links between Atmospheric Dynamics and Non-Gaussian Distributions and Quantifying Their Effects on Numerical Weather Prediction

Project Period: November 1, 2017 to October 31, 2020
Principal Investigator(s): Steve Fletcher
Co-Principal Investigator(s): Anton Kliewer, Andrew Jones, John Forsythe
Postdoc(s) and Student(s): Michael Goodliff
Sponsor(s): National Science Foundation (NSF)


This project is a continuation of a previous NSF-supported research project, “Analyzing the Impacts of Non-Gaussian Errors in Gaussian Data Assimilation Systems.” The goal of this project is to address several different aspects about how non-Gaussian distributed errors affect data assimilation and retrieval systems. Previously we have shown that it is possible to obtain better fits for the moisture and temperature channels from the Advanced Microwave Sounding Unit A and B by using a mixed Gaussian-lognormal based 1D variational microwave retrieval system for temperature and mixing ratio respectively. We also observed that it is possible for the flow to change probability density function by season and as such detection of these PDF changes is important to ensure that the DA system is consistent with the PDF at that time. To continue this previous research, two of the major goals of this project include linking the atmospheric dynamics to their respective conditional PDFs as well as developing an online kernel detection system.

Advancing littoral zone aerosol prediction via holistic studies in regime-dependent flows

Project Period: July 1, 2015 to June 30, 2020
Principal Investigator(s): Steve D. Miller (CIRA/CSU)
Co-Principal Investigator(s): Susan van den Heever, Sonia Kreidenweis, Milija Zupanski, Robert Holz, Jianglong Zhang, and Jun Wang

Other Investigators: Jeremy Solbrig, Renate Brummer, Anton Kliewer, Karina Apodaca-Martinez, Lewis Grasso, Matthew Rogers, Kevin Micke, Adele Igel, Qijing (Emily) Bian, Steve Saleeby, Min Oo, Steve Albers, and Ting-Chi Wu

Postdoc(s) and Student(s): Jungmin Park, Jennie Bukowski, Yi Wang, and Sam Atwood

Sponsors: Office of Naval Research (ONR)


This project takes a multi-faceted, integrated approach to understanding and addressing significant forecasting challenges of atmospheric aerosol in the coastal zone impacting the characteristics of electro/optical (E/O) propagation. The research aims to improve our basic understanding of the relative roles and synergistic interactions of key environmental factors that influence aerosol distributions and thereby E/O propagation, advance our ability to characterize littoral zone aerosol distribution and properties via next-generation satellite observations and algorithms, and examine how state of the art data assimilation approaches can exploit this information to provide representative high-resolution three dimensional (3D) analysis of the littoral zone aerosol. Click here to find out more about this Office of Naval Research $7.5 million sponsored project.

Data assimilation of GLM observations in HWRF/GSI system

Project Period: July 1, 2017 to June 30, 2020
Principal Investigator(s): Milija Zupanski
Co-Principal Investigator(s): Ting-Chi Wu, Karina Apodaca
Sponsors: National Oceanic and Atmospheric Administration (NOAA)


In this research we plan to exploit the unique capabilities of the Geostationary Lightning Mapper (GLM) instrument through data assimilation with the NOAA Hurricane WRF operational system. We propose to add the GLM assimilation capability to the HWRF data assimilation system, which is based on the hybrid Gridpoint Statistical Interpolation (GSI) algorithm. The new HWRF/GSI system with the GLM assimilation capability will be evaluated in detail. Given high spatiotemporal coverage of GLM over open oceans, this research represents a great opportunity for potential improvement of the NOAA operational HWRF forecasts, in particular the prediction of rapid intensification of tropical cyclones.


The assimilation of GLM lightning data requires a transformation from HWRF model forecast to lightning flash rate, referred to as the lightning observation operator, which will be adopted/developed for tropical cyclone applications in this research. Note that the same transformation can be used for post-processing of the HWRF forecast, effectively producing a GLM lightning flash rate forecast. This is similar to lightning threat forecasts for severe weather, however now applied to tropical cyclones and customized for such application. The GLM lightning forecast for hurricanes will be developed and evaluated in collaboration with the National Hurricane Center (NHC). Although NHC does forecast lightning, the comparison of the GLM forecasts to real GLM data for the early part of the forecast in real time, and for the entire forecast post-storm, will provide information on the realism of the HWRF model. Feedback from the NHC evaluation will also be provided to the EMC HWRF team to improve the lightning observation operator.


Collaboration with EMC HWRF team will assure that our research is aligned with EMC/HWRF operational plans, with clear path to operations. The proposed research will be the first application of assimilation of real GLM observations with NOAA HWRF/GSI assimilation/prediction system.

(COMPLETED) R2O Transition of the GOES-R GLM Assimilation Capability in GSI for Use in the NCEP GDAS

Project Period: May 1, 2017 to April 30, 2018
Principal Investigator(s): Karina Apodaca

Co-Principal Investigator(s): Milija Zupanski

Sponsor(s): Developmental Testbed Center (DTC)



Goal: Preparing the GOES-R GLM lightning assimilation development in the Gridpoint Statistical Interpolation system for using satellite lightning measurements in NOAA/NWS/NCEP Global Forecasting System.


Final Report

(COMPLETED) Advancing coupled land-atmosphere modeling with NASA-Unified WRF via process studies and satellite-scale data assimilation

Project Period: July 28, 2013 to July 29, 2017
Principal Investigator(s): Christa Peters-Lidard (NASA Goddard Space Flight Center)
Co-Principal Investigator(s): A. Hou and T. Matsui (NASA/GSFC)
Graduate Students, Postdoctoral and Other Investigators: J. Case (ENSCO Inc.), M. Chin (NASA/GSFC), J. Geiger (NASA/GSFC), Y. Liu (NASA/GSFC/ESSIC), J. Santanello (NASA/GSFC), J.J. Shi (NASA/GSFC), Q. Tan (NASA/GSFC/USRA), Wei-Kuo Tao (NASA/GSFC), Z. Tao (NASA/GSFC/USRA), B. Zaitchik (Johns Hopkins University), B. Zavodsky (NASA/MSFC), Sara Q. Zhang (NASA/GSFC), Milija Zupanski

Sponsor(s): National Aeronautics and Space Administration (NASA)



This project builds on the successful development and application of the NASA Unified Weather Research and Forecasting (NU-WRF) modeling system, with the goal of integrating and enhancing existing land and atmospheric data assimilation capabilities to advance regional-scale coupled land-atmosphere modeling for process studies. The NU-WRF currently combines the capabilities of the WRF with the Land Information System (LIS), the Goddard Chemistry Aerosol Radiation and Transport (GOCART) model, advanced microphysics and coupling between clouds and aerosols to better represent cloud-aerosol-precipitation-land surface processes. Outputs can be directly compared with satellite L1B data via the Goddard Satellite Data Simulator Unit (G-SDSU). Further, the NU-WRF connects with global-scale modeling efforts, including the GEOS-5 and the MERRA, which can be used as atmospheric boundary and initial conditions. This project will focus on advanced component couplings and integration of existing land (LIS-DA) and atmospheric (WRF-EDA) data assimilation in NU-WRF.



This proposal represents a continuation of the core development for NU-WRF, as described in the call. Specifically, we will (1) Characterize and/or reduce uncertainties in models and products (by evaluating the case studies with and without data assimilation, as well as after physics and chemistry refinements); (2) Extend the range of model or product validity by using new components (by incorporating and generalizing the LIS-DA and the WRF-EDA); and (3) Enable independent community validation and characterization (by distributing the NU-WRF code to partners). The proposed data assimilation and physics enhancements to NU-WRF further the core MAP program interests in regional observation-driven modeling and data assimilation.



Recent results have indicated the critical role that incorporating land surface and atmospheric observations can play in helping further advance our ability to represent coupled land-atmosphere processes. As part of this work, we will first enhance the couplings between key NU-WRF components, such as wet deposition, dust and biogenic emissions, aerosol-cloud- precipitation interactions, and mesoscale circulations. Next, we will fully incorporate and advance LIS-DA in the NU-WRF to assimilate satellite-based land surface products such as soil moisture, snow cover and depth, and skin temperature, while also updating land parameters such as real-time green vegetation fraction, albedo, and irrigated area. We will also fully incorporate and advance the WRF-EDA in the NU-WRF to assimilate cloud-precipitation-affected microwave radiances to advance the representation of clouds and precipitation in the NU-WRF. Finally, we will conduct case studies demonstrating the impact of the LIS-DA and the WRF- EDA on land-atmosphere processes for high-impact weather-to-climate scenarios such as tropical cyclones, droughts, floods, heat waves and extreme storms; as well as impacts on atmospheric chemical constituents including severe air quality degrading events.

(COMPLETED) Evaluation of the Newly Developed Observation Operators for Assimilating Satellite Cloud and Precipitation Observations in GSI within the HWRF system

Project Period: May 1, 2017 to April 30, 2018
Principal Investigator(s): Ting-Chi Wu
Co-Principal Investigator(s): Milija Zupanski
Sponsor(s): Developmental Testbed Center (DTC)


Data assimilation has contributed greatly to the improvement of numerical weather prediction (NWP) by optimally combining information from a forecast model with available observations. Among the available observations from various platforms that are assimilated in operational NWP centers, satellite measurements of cloud and precipitation information are currently not being fully exploited to their maximum capacity. Since the high-impact weather events such as hurricanes are often associated with clouds and precipitation, assimilating the satellite measured cloud and precipitation information may further improve the hurricane forecasting.


Proposed Work: Although the HWRF/GSI capabilities to assimilate satellite retrieved water contents have already been developed, the assimilation impact has only been preliminarily tested (e.g., Wu et al. 2016). In this project, however, we plan to perform a detailed evaluation of the impact of the new observation operators by employing an extensive set of selected cases (i.e., larger sample size) and assessing their value in a systematic manner before they are ready for general/public use. The very first step is to merge these capabilities into the GSI and HWRF trunk codes, where all the new developments, code updates and maintenances are retained. We will then conduct a set of tests that include experiments with and without the activation of the newly added capabilities to confirm the stability of the system with the added features. To offer a robust performance statistics, we will apply the added capabilities to the pre-selected set of cases to evaluate their effectiveness. Once the proposed testing and evaluation are completed, these new capabilities based our recent research efforts (Wu et al. 2016, Wu and Zupanski, 2017) will be available for operational considerations. With that, we anticipate a rather smooth research to operation (R2O) transition.


Final Report

(COMPLETED) Ensemble Data Assimilation Research for Hurricane Forecasting

Project Period: April 1, 2013 to March 31, 2017
Principal Investigator(s): Milija Zupanski
Co-Principal Investigator(s): Sara Q. Zhang (NASA Goddard Space Flight Center)
Sponsor(s): National Aeronautics and Space Administration (NASA)



In this research we will further develop the NASA WRF-EDAS data assimilation system and increase its robustness and efficiency, in particular by addressing the insufficient number of degrees of freedom in ensemble error covariances and the precipitation-affected radiance bias correction. The focus of the proposed continued development of WRF-EDAS is methodology improvement that will eventually lead to maximizing the impact of GPM measurements and improved estimates of rainfall. Therefore, the main goal of the proposed research is to continue development of WRF-EDAS and produce an operation-ready assimilation/downscaling system for assimilation of GPM data. This will be achieved by:

  • Developing hybrid variational-ensemble capability, and
  • Developing bias correction scheme for precipitation-affected microwave radiances.

In parallel with the proposed research development, the WRF-EDAS components (e.g., WRF model, microphysics, Goddard-SDSU, GSI, and MLEF) will be periodically updated with the latest versions available. The WRF-EDAS analysis and forecast will initially be produced at 9 km and 3 km resolutions, eventually reaching 1 km resolution relevant for hydrological applications.


In addition to NOAA operational observations available through the GSI interface, we will assimilate currently available TRMM TMI, SSM/IS, MHS data, as well as GPM data when they become available after the launch in 2014. Targeted regions of our experiments will be Hydro-Meteorological Testbeds (HMTs) due to their relevance for ground validation of GPM data and eventual utilization of GPM products in NWP operations. We will also evaluate the system for tropical cyclone and flood events. Following a suggestion from a reviewer, we will validate downscaling of TRMM against other higher resolution products such as a ground-based radar.


We will also make steps towards utilization/validation of the downscaled precipitation analyses and forecasts by hydrological communities. The database produced by the proposed research will reside on NASA supercomputers. It will be made available to other GPM team members. It will be also posted to this research project webpage and to the GPM webpage.

(COMPLETED) Analyzing the Impacts of Non-Gaussian Errors in Gaussian Data Assimilation Systems

Project Period: September 1, 2012 to August 31, 2016
Principal Investigator(s): Steve Fletcher
Co-Principal Investigator(s): Andrew Jones (CSU/CIRA)
Graduate Students, Postdoctoral and Other Investigators: John Forsythe (CSU/CIRA) and Anton Kliewer
Sponsor(s): National Science Foundation (NSF)


This NSF funded project investigated the impacts of non-Gaussian errors associated with a temperature and humidity 1-dimensional variational retrieval system called the CIRA 1-Dimensional Optimal Estimator, or C1DOE for short. As well as testing a mixed lognormal –Gaussian distribution formulation of a 1DVAR cost function against the standard Gaussian fits all and the logarithmic transform formulation, the PI and Co-PI developed the needed theory to extend the mixed distribution approach from the full field 3DVAR and 4DVAR formulation to incremental versions through the introduction of a geometric tangent linear approach. Given all these new formulations of non-Gaussian systems, we needed to address the question of how could we know in advance that we require a lognormal based data assimilation system? The way we addressed the question just presented was to develop a set of statistical tests that all needed to be satisfied at a high confidence level to ensure that we 1) did not make a false positive statement but 2) to overcome the auto-correlation in the data sample.


Another important aspect of the work on this project was associated with determine a new set of quality control measures to enable observations with lognormally distributed errors would not be rejected by a Gaussian based QC measure. We were able to develop a new linearization that enabled the interchanging of the logarithm and the observation operator such that the theory from the original buddy check system followed through to the lognormal case.


The final areas of research that was undertaken on this grant came about through a peculiarity in the performance of the idea case for the lognormal retrieval system. The first feature that was discovered about the lognormal based 1DVAR case for a 1 variable situation is that the three descriptive statistics: mode (maximum likelihood state), the median (unbiased state) and the mean (minimum variance state) had regions of optimality. The regions of optimality for each descriptive statistic was based upon certain values of the parameters in the retrieval, i.e. background state, background error covariance, observational error covariance as well as the measurement and representative errors. We found that if there was no observational errors, but a fixed error covariance and a fixed background error covariance, then there exists values for the background state such that there existed optimal regions for the values of the background state, such that each descriptive statistic was optimal.


For more information, please check the project summary PDF.


Project Summery