The Wind Forecast Improvement Project (WFIP) is a two-year effort funded by the Department of Energy (DOE) with collaborative support by the National Oceanic Atmospheric Administration (NOAA). The project goals include developing and demonstrating methods to improve the skill of 0- to 6-hour wind power forecasts in order to facilitate integration of larger amounts of wind power in the electrical grid systems of the United States. The WFIP consists of two parallel 2-year subprojects: one in the Northern Plains and Texas. The Texas component of the WFIP is a collaboration among the DOE, NOAA and a group of private sector partners, including MESO, Inc., that is being managed by AWS True power, LLC. During the 1-year field experiment from July 2011to July 2012, an array of sensors will be deployed at targeted locations by NOAA, DOE and the private sector team members to collect data within the experimental area. The data from those sensors will be assimilated into a set of advanced prediction systems to generate real-time 0- to 6-hour ahead wind power forecasts for wind farms on the electrical grid system operated by the Electric Reliability Council of Texas (ERCOT). An ensemble of high-resolution deterministic numerical weather prediction (NWP) forecasts is being used to assimilate observations from sodars, wind profilers, and meteorological towers, which have been deployed in targeted locations for this project. The output from this ensemble of NWP forecasts is used as input for a suite of statistical algorithms that generate deterministic and probabilistic forecasts of (1) average wind power production over 15-minute and 1-hour intervals for the 6-hour look-ahead period and (2) the occurrence of large power production change events (known as “ramps”) during the 6-hour forecast window. The NWP ensemble consists of 12 members, which are collectively executed by members of the project team. One member is the High Resolution Rapid Refresh (HRRR) executed hourly on a 3 km grid by NOAA/ESRL. A second member is a high resolution ARPS model executed on a 6-hour cycle on a 2-km grid by the University of Oklahoma. The other 10 members are executed by the MESO, Inc. - AWS Truepower team and consist of runs of the WRF, ARPS and MASS models on a 5 km grid every two hours using different combinations of initialization datasets, data assimilation procedures, assimilated data types, and model physics. The data from additional sensors deployed for this project as well as a set of participating wind farms within Texas are assimilated into most of the ensemble members. However, these data are withheld from some ensemble members to gauge their impact on the forecasts. The output from the NWP ensemble is used as input into a Model Output Statistics (MOS) procedure. The MOS is designed is to correct systematic errors in the NWP prediction of the relevant meteorological variables at the forecast sites (i.e. the wind farms). The output from the MOS procedures for the individual models is then used as input to the Optimized Ensemble Model, which creates a composite forecast by statistically weighting the MOS-adjusted forecasts from each ensemble member according to the relative historical performance of each member. Several different ensemble weighting schemes are being tested. One is based on the relative performance over a rolling sample of the last 30 days. A second approach is based on a customized “analog” training sample that is constructed by matching key weather parameters for the current forecast period with those for cases in a historical archive. The objective of this approach is to weight the forecasts according their performance in weather patterns that are similar to the one expected during the forecast period. An extensive analysis of the forecast ensembles is being performed on an ongoing basis through the experimental period. A fundamental objective is to diagnose the amount of improvement in forecasts of both average power production and ramp events that was obtained using the advanced forecast system and expanded observational dataset. However, the performance analysis procedure has also been designed to diagnose the relative impact of the individual components of the forecast system (e.g. additional input data, specific NWP systems, data assimilation procedures, etc.) on forecast performance for the purpose of identifying configurations that offer the most cost effective benefits for application beyond WFIP. A forecast performance analysis for the first few months of the WFIP experimental period will be presented at the conference. This analysis will include an assessment of the overall ensemble composite forecast performance for wind speed and direction, average wind power production, and wind ramp events as well as the relative performance of all of the raw and MOS-adjusted individual NWP ensemble members for these forecast elements.