Lightning Probability Tool

The Patrick Air Force Base 45th Contracting Squadron is seeking sources with the capability to fill the requirement for the 45th Weather Squadron (45 WS) to update the current lightning probability tool used by the 45 WS for their daily planning forecasts. There are two main goals to be fulfilled in the update of the lightning probability tool: 1) extend the tool to include Oct, and 2) determine if stratifying the predictive equations by variable lightning sub-seasons improves performance as compared to the monthly stratifications currently being used and implement the updated tool into 45 WS operations accordingly. Extend the period the tool runs to include 1 May-31 Oct (currently the tool only runs 1 May-30 Sep). Stratify the tool by lightning sub-season, rather than by month. The dates of the lightning sub-seasons vary year to year, which must be considered in developing the updated tool. If that doesn't out-perform the current monthly stratifications, update the tool to using monthly stratifications to include Oct (May, Jun, Jul, Aug, Sep, Oct). Extend the study to include the most recent lightning seasons. Optimize the tool's performance on dependent data and verify its performance on independent data. Document the performance of the tool. Implement operationally on the Meteorological Information Data Display System (MIDDS) used by 45 WS. Attached documents describe the current 45 WS lightning probability tool and the MIDDS system. The following steps are the 45 WS suggested approach to the achievement of the above stated goals, though an alternate contractor approach may be considered at such time if an RFP is issued: 1) First Step: Develop an algorithm to identify the lightning sub-season objectively. The five lightning sub-seasons are: 1) pre-season, 2) ramp-up, 3) mid-season, 4) ramp-down, and 5) post-season. There is also a sixth - out-of-season' sub-season from 1 Jan thru 30 Apr/1 Nov thru 31 Dec; however, this sub-season doesn't apply to this task, as the tool only runs 1 May thru 31 Oct. Refer to the attached description of the current tool for the lightning sub-seasons. Note: One approach for the algorithm to identify the lightning sub-seasons objectively for each year is to use the National Lightning Detection Network (NLDN) daily flash count across central Florida, correlate with known dates, and check if the NLDN's climatology agrees with the accepted dates. The sub-seasons must occur sequentially. Once a sub-season begins within a certain season, do not revert back to a previous sub-season, i.e. once the sub-season begins, subsequent days must either be that season, or the next season in sequence. For example, if the mid-season has begun in a certain summer, do not fall-back to the ramp-up season or pre-season. 2) Second Step: Develop logistic regression equations stratified by lightning sub-season to predict the probability of lightning for each day. Select optimum predictor variables based on sub-season, repeating the process used in developing the current tool. If the new equations do not outperform the current tool, update the tool using monthly stratification to include Oct. Confer with the 45 WS on the decision of which version of the update to implement. Use all years, including new seasons, since last version of the tool was developed. Optimize the coefficients of the logistic regression on training data. Test on random independent verification data set where the data are selected randomly across all days, i.e. entire seasons are not selected randomly. 3) Third Step: Document performance of the updated tool on independent data. Provide both absolute performance metrics to include probability of detection, probability of false alarm, true skill score, attribute/reliability diagram, sharpness diagrams, and briar skill score. Also document performance of the new tool relative to the current tool. Use same period of record for the comparison. This may have been done under the second goal. 4) Fourth Step: Implement updated tool in MIDDS. Assistance from Eastern Range personnel may be required. 5) Fifth Step: Provide training to 45 WS on how to use the tool on MIDDS and summarizing the development and performance of the new tool. 7) Sixth Step: Provide final report documenting the development of the updated tool and its performance. It is the Government's intention to set aside this requirement under a small business set-aside program. The North American Industry Classification Systems (NAICS) Code proposed for the requirement is 541990. The size standard for NAICS 541990 is $7M. The Government is interested in all qualified and experienced small businesses, including 8(a), Historically Underutilized Business Zone, or Service Disabled Veteran Owned Small Businesses that, in the Government's opinion, are capable of performing this requirement. In the event that no qualified and experienced small business responds, it will be the Government's intention to consider all other sources. The Government requests that interested parties submit a brief description of their company's business size (i.e. annual revenues and employee size), business status (i.e., 8(a), Historically Underutilized Business Zone, or Service Disabled Veteran Owned Small Businesses, small business), and a description of similar services offered to the Government and to commercial customers. The Government will use this information in determining its small business set-aside decision. Sources should submit a brief capabilities statement in response to this notice. The response should be no more than 3 typed pages and must address the following areas: 1) provide a description of the firm's background in meteorology and general statistics, 2) demonstrate past performance in creating lightning probability forecast tools, preferably for the CCAFS/KSC local area 3) demonstrate experience developing and implementing software tools for MIDDS 4) demonstrate experience using statistical analysis techniques working with lightning probability tools, particularly logistic regression, to include optimum selection of predictor variables, and experience using statistical analysis software to perform those statistical techniques, and 5) demonstrate knowledge of local weather influences for the CCAFS/KSC and Central Florida. Submit capabilities statement to: Attn: Leisa Zemba, Bldg 1704, Hangar Rd, Cape Canaveral Air Force Station, FL 32920-2200 or email to leisa.zemba@patrick.af.mil. Responses are due no later than 06 July 2010. Direct all questions to the attention of Leisa Zemba at 321 854-0974. This sources sought is NOT a Request For Proposal (RFP) and does NOT commit the Government to pay for information requested, issuance of a solicitation, or award of a contract. This effort is considered Commercial in nature and any resultant RFP/Contract will be issued/awarded IAW FAR Part 12, Acquisition of Commercial Items.