OHD/HL/CHPS - Implementation Strategy

CHPS Implementation Strategy

BACKGROUND

The desire to modernize the software infrastructure for NOAA NWS River Forecast Center (RFC) forecasting operations has been discussed within the OHD for more than a decade. Prior to CHPS, the OHD had considered migration of its primary river forecasting tool, the NWSRFS, to a Common Object Request Broker Architecture (CORBA), but found CORBA to be too complex and too proprietary in nature, potentially resulting in unacceptably high costs. It was also unclear how the migration of the NWSRFS to a CORBA-based solution could be accomplished without seriously impacting operations.

Beginning in 2002, the OHD contracted with Apex Digital Services in Silver Spring, MD in an effort to find an alternative for the software infrastructure. In their final report "Software Architecture Engagement Summary", dated January 14 2004, Apex recommended that the OHD adopt a Service Oriented Architecture (SOA) approach, using open source (i.e., free) software components where possible.

At the time the expected approach was to incrementally upgrade ("modernize") the existing computational infrastructure of NWSRFS. Some proof-of-concept projects were conducted to show how modernization could be accomplished by "wrapping" existing software modules. The OHD subsequently initiated several contract tasks to begin the process of modernizing existing components of the NWSRFS. That effort, however, proved to be of limited success; it was evident that the task would take many years with little or no benefit to operational forecasters.

As an alternative strategy, the OHD began to look to ready-made solutions. OHD staff evaluated a small number of service oriented forecasting frameworks that might serve to accelerate the deployment of CHPS. One was the Object Modeling System (OMS) developed by the US Department of Agriculture (USDA). Another system referred to as GISRS, was developed by the NOAA NWS National Operational Hydrologic Remote Sensing Center (NOHRSC). Both of these software systems were found to be lacking the maturity needed to serve as a reliable, robust system for NWS RFC operations, and were not extensible enough to support the needs of CHPS.

In October 2005, at the invitation of the OHD Senior Scientist, representatives from the Dutch hydraulics engineering company WL | Delft traveled to NWS headquarters and provided OHD staff with a demonstration of Delft's open source software framework called the Delft Flood Early Warning System (Delft-FEWS). Not only did the system appear to be a manifestation of the solution described by Apex, but the system was also specifically designed for use in hydrologic forecasting operations and was actively being used in Europe. The OHD interviewed Delft's primary user of the system, the U.K. Environment Agency, to learn more about the quality of the product and customer satisfaction; both were found to be quite acceptable. The Delft-FEWS solution is non-proprietary but requires a licensing agreement between NOAA and Delft.

[Back To Top]

CURRENT APPROACH

1. CHPS FEWS Pilot

A CHPS Acceleration Team (CAT) was formed in early 2006, consisting of the following RFC representatives and the OHD Senior Scientist:

Rob Hartman, Hydrologist In Charge, CNRFC
John Halquist, Development & Operations Hydrologist, NCRFC
Harold Opitz, Hydrologist In Charge, NWRFC
Billy Olsen, Hydrologist In Charge, ABRFC
Pedro Restrepo, OHD Senior Scientist

The CAT members recognized that modernizing or replacing the NWSRFS would be a painful task. The OHD engaged Apex Digital Systems under contract to conduct a "discovery engagement" to determine the strengths and weaknesses of the current system, and to solicit a minimal set of high level requirements for a modernized NWSRFS. Apex led a series of meetings during July through October 2006, the results of which were delivered in the report "OHD Software Development Roadmap Definition", dated December 4 2006 (available here...).

During these meetings the CAT members expressed a strong desire to replace, not modernize, the NWSRFS.

Thus in September 2006, through its contract with Riverside Technology, inc (RTi), the OHD requested that Delft develop a system which would demonstrate the Delft-FEWS in an NWS RFC environment. The name of this demonstration system was the "CHPS FEWS Pilot".

The NWS worked closely with RTi, Delft, and a third contractor RS Information Systems, Inc (RSIS) to define and develop a minimally capable prototype CHPS system running at the NWRFC and the NCRFC. Basic functionality included the following:

• Unit Hydrograph model (Delft’s existing Unit Hydrograph model)
• Snow model (Delft’s existing SNOWMELT model)
• Sac-SMA with Heat Transfer model (SacSma-HT application developed by OHD/RSIS)
• Channel Routing model (Delft’s existing Muskingham model)
• Reservoir routing model (Delft’s existing reservoir model)
• Workflows that mimic existing HCL (Delft’s existing XML-based control flow)
• Basic time series transformations (Delft’s existing time-series transformation modules)
• Ability to conduct ‘what-if’ scenarios (Delft’s existing what-if capability)
• Estimation of missing data and computation of basin area averages from point values (Delft’s existing modules)
• Export of point data from the operational AWIPS hydrologic database to FEWS (Export Application developed by RTi)

After an intense 4 month period, the system was completed and a demonstration of the CHPS FEWS Pilot system took place at the Northwest RFC (NWRFC) in Portland, Oregon on April 17-18 2007. The demonstration was successful and generated a great deal of interest in pursuing the Delft-FEWS.

2. Expansion of the CHPS FEWS Pilot

One outcome of the demonstration was the recognition that certain key capabilities, currently in widespread use via NWSRFS, are essential if Delft-FEWS is to be a viable solution for CHPS. These included:

• Provision of a MODS-like capability (to supplement what-if scenarios)
• Inclusion of the SNOW-17 model (to replace SNOWMELT)
• Implementation of a distributed architecture (currently standalone)
• Additional training

Furthermore, the ABRFC indicated a desire to install and evaluate the CHPS FEWS Pilot. OHD initiated new tasking through existing contracts with RTi and RSIS to implement those enhancements; work was completed in December 2007 and demonstrated to the CAT in Chanhassen, MN at NCRFC on December 16-17, 2007.

3. Evaluation of the CHPS FEWS Pilot

OHD engaged Apex Digital Systems to determine success (evaluation) criteria for the CHPS FEWS Pilot. A draft evaluation criteria document was delivered in July 2007 and put on hold until the Pilot enhancements were complete.

In January 2008 the CAT used this document as the basis for their recommendation to Gary Carter, Director of OHD that NOAA's Hydrology Program should proceed with implementation of FEWS as the infrastructure component of CHPS. A copy of the recommendation report is available here.

4. OSIP

CHPS is managed under the NOAA NWS Operations & Services Improvement Process (OSIP). It is identified as project number 07-017, entitled "CHPS Infrastructure". (Access to the OSIP web site is password protected.)

The purpose of the CHPS Infrastructure project is to develop and deploy a minimally capable replacement software system at the RFCs to help the NOAA’s Hydrologic Program meet its future goals. It is currently expected that the Infrastructure system will be based on the Delft-FEWS software. The CHPS Infrastructure project entered OSIP Stage 3 (Applied Research & Analysis) on May 15 2007. Development and evaluation of the Pilot system are OSIP Stage 3 activities.

At the end of OSIP Stage 3, the OSIP Integrated Work Team (IWT) will put together an Applied Research Evaluation, a Business Case, a Technical Requirements document, and a Plan for OSIP Stage 4. OSIP Gate 3 is expected in late 2008.

OHD leads the OSIP IWT in all stages.

5. Future projects

With a new Infrastructure in place, other essential components of CHPS will be introduced incrementally. For example:

• the existing Calibration and Verification capabilities of the NWSRFS must be added to the CHPS Infrastructure;
• new science that is currently in prototype status, such as short-to-long term Hydrologic Ensembles, must be integrated into a CHPS Infrastructure rather than into the aging and less flexible NWSRFS;
• HEC-RAS will be adapted for use in CHPS as the preferred hydraulic model (FldWav.is currently used in the NWSRFS environment; there are no plans to provide access to FldWav within CHPS)

Other potential projects will continue to be identified. New projects will be initiated under the OSIP Water Resources umbrella project (number 05-014) once the CHPS Infrastructure project is complete.

[Back To Top]

[Back to main CHPS web page]