Activity and Change Log

This page records changes in forecast approaches and other activities that relate to the forecasting effort at UW. Routine activities, such as forecast production or various interactions, are not listed. The most recent are listed first.

Oct 2007

• Streamflow forecasts for Arkansas-Red, Guld and Upper Mississippi Basin installed on cluster

Sep 2007

• ESP soil moisture forecasting for Upper and Lower Mississippi installed on cluster

Aug 2007

• ESP soil moisture forecasting for Gulf and Arkansas-Red installed on cluster
• Runoff forecasting added to SW Monitor

July 2007

• Web page for streamflow forecast point information installed on web

Jun 2007

• Additional streamflow forecast points for Missouri basin

May 2007

• Soil moisture nowcasting for Lower and Upper Mississippi installed on cluster

Apr 2007

• Soil moisture nowcasting for Gulf of Mexico (South Central) installed on cluster
• Soil moisture nowcasting for Rio Grande expanded to lower Rio Grande

Feb 2007

• Soil moisture nowcasting for Arkansas-Red installed on cluster

Jan 2007

• streamflow forecasts for the Missouri and upper Rio Grande installed on cluster

Dec 2006

• streamflow forecasts for the colorado and california basin installed on cluster.
• streamflow routing for the missouri and riogrande tested on cluster, but not
• california soil file reverted to previous soil file: all retrospective climatological run data regenerated and processed.
• ancillary runs to support snotel assimilation completed for all basins

Nov 2006

• ESP forecasts installed on cluster, with streamflow forecasts in the PNW, using daily nowcast as initial state.
• C-based routing model debugged, and implemented in place of old FORTRAN routing model. modifications include direct use of spinup fluxes.
• run locations on cluster re-configured to improve run times
• various system status monitoring scripts and webpages put in place
• automated data offload for SARP project installed

Oct 2006

• EMC's Y. Xia visits UW to continue transfer of several west-wide system related techniques, including CPC based outlooks, to EMC systems
• CPC seasonal forecast approach revised as part of this effort to improve monthly disaggregation. Approach still considered to be evolving.
• SW monitor website remodeled.

Sep 2006

• Forecasting migrating to cluster; will use daily-updating 1/8 nowcast instead of original system for Oct 1 forecast.
• SW Monitor realtime implementation switched from ponic.hydro to phobic.hydro to enable FreeBSD-to-Linux conversion of PCs in LSHRG lab.

Aug 2006

• Missouri R. basin 1/8 degree nowcasts added to daily-updating system.
• Multi-objective streamflow calibration approach implemented for specific basins to facilitate hindcast ESP comparisons with NWS & NRCS (A. Wood, K. Werner, T. Pagano collaboration).

Jul 2006

• Mexico 1/8 degree nowcasts added to daily-updating system.

Jun 2006

• BAMS paper on west-wide forecasting system accepted.
• "VIC Model" project formalized within the USDA Collaborative Software Development Laboratory (COLAB), and a test-model (for Trinity R. basin) was isolated, reorganized and transferred.
• the provisional 1/8 degree daily updating nowcasts for the westwide domain appear to be stable, and links to the WSHFS website were established (see left-navigation bar)). In the future, these fields will be used for model initialization.
• Retrospective simulations for Mexico completed; now real-time simulations nearing readiness
• A. Wood visits NCEP EMC to initiate transfer of several west-wide system related techniques to EMC systems

May 2006

• contact was made with the USDA/ARS Object Modeling System project to explore the use of the OMS framework for porting westwide-system based forecasts to external users
• provisional 1/8 degree daily updating nowcasts for the PNW were completed (see temporary page here, & scroll down for test 2-wk animations). Real-time EDAS temperature data feeds & processing were established to support similar nowcasting in Mexico.

Apr 2006

• Added two forecast points in the Upper Klamath R. basin
• provisional 1/8 degree daily updating nowcasts for California, Great Basin, Colorado and the upper Rio Grande now implemented for viewing on temporary page. Scroll down for test 2-wk animations. The PNW, Missouri R. basin and Mexico are close to being finished.

Mar 2006

• Spin-up condition basin-averaged plot generation added to system, and plots added to website, initially in the streamflow pop-up windows.
• 1-month streamflow forecast changes are now calculated and summarized in new tables on the website. During the forecast season, the changes in conditions from month to month are important to operational forecasters.
• CPC forecast generation altered once again, to constrain the mean of the random ensemble to within 5% of the generating statistical mean derived from the CPC forecasts. This decision allows the use of only 30 ensemble members, an ensemble size which otherwise would have too large of a sampling error. The higher order moments are not similarly constrained.

Feb 2006

• The ENSO-conditioned ESP forecast was changed to ENSO Cold (La Nina) from Neutral. Unfortunately, this causes the ENSO/PDO subset to be too small (5 members) to warrant much attention.
• Spatial current conditions 1 month change plots for SWE and soil moisture were added to the "spatial" page.
• This activity/change log was implemented.

Jan 2006

• *Daily-updating* observed SWE plots were added as a new major section of the website - these are based on the re-implementation of prior monthly update scripts on the "flood" Linux cluster. Change plots (1 wk and 2 wk), and percentile, anomaly, and SWE-scaled views, were also added. This datastream will form the foundation of future snow assimilation approaches.
• Efforts to migrate the hydrologic nowcast to the cluster, and move it from a monthly to a daily update, with expanded domain, also began.
• The CPC forecast method approach was changed to a purely distribution-based ensemble generation, and the ensemble size changed from 13 to 30 members (and may increase further). The CPC ensembles have a known problem of underestimated variance, stemming from their derivation from 3-month average forecasts. We are working on a solution.

Dec 2005

• SWE assimilation was started again for the season. Typically SWE assimilation is not implememented from July to November. In addition, ESP forecasts *without SWE assimilation* were also added to the forecast set. These are run after the main (assimilating) ESP forecasts, and are not shown on the website, but will be used for analysis as the season progresses.