Propose marking this IPTD as complete Enter new IPTD for work plan with CGI to do ICIS Air node update, data model and low changes

From John Smith: “One of the deliverables for this effort is developing an SOP, which is still ongoing. Also, the “case file” process still needs some refinement. Therefore we want to keep this project open”

as per John Smith Aug 7: We have successfully submitted batch files to ICIS Air production for Informal Enforcement Actions, Formal Enforcement Actions, and 1 variety of Compliance Monitoring (FCE – full compliance evaluations) for some FY 22 & FY 23 data. When Yannis came aboard we had her enter the Air data manually to learn the ICIS Air system & how to navigate NJEMS, which she got quite good at. We then introduced her to doing the batch upload for the ICIS NPDES to learn how to use the XML generators sheets & to do batch uploads. She has for the most part learned how to do the batch uploads for ICIS NPDES without much intervention from me. Starting next month (this month’s Air data has already been entered) we plan to have her use the XML generators for Air data mentioned earlier. If that all goes well, we can continue developing XML generators for the remaining data families

Meeting 3/14/23 with John Smith, Nicole Pellegrini and Larry Si

The current project is aiming to model on partial automation comparable to ICIS-NPDES enforcement updating. This involves reporting that can be scheduled with translation (kicked off manually) to XML for transfer via an EPA provided node. About half of the data can be done this way thanks to in-house modification of about half the translation routines previously built by a vendor for ICIS-NPDES.

The deliverables are:

1. Reports to extract the needed data (complete)

2. Routines to convert the extracted data to XML (50% complete)

3. SOP for all the human steps needed to get, translate and transfer the data on a batch/monthly basis.

It was noted that all manual steps are currently done by DOIT staff, including takeover from Fallow Freck (now with Pesticides) by new hire Yannis Zhang.

Pursuit of automaton beyond the current effort might reconsider the problem and potential for practical limitations covering at least these observations:

Problem: From the Problem worksheet and today’s discussion there is a clear long-term complaint from EPA about data delays. With completion of deliverables, these should be improved comparable to NDPES (which has not expressed need for further automation). John described having already batched 2023 Air data using the partially automated process. This is already within a calendar quarter and may be optimal timeliness knowing the typical offset from activity occurrence to data entry and locking.

Practical limits? John Smith raised the example of the only more-timely or “automated” process which is the daily exchange of similar information in RCRA, with its much closer alignment between Federal and State regulation. While “automated”, that process involves a steady amount of human intervention to manage the built in error reporting. This is almost certainly due in part to the designed flexibility and lack of validation controls on entry of most enforcement data, which can regularly introduce scenarios and resultant data that will not pass full automation. This challenge seems to correlate to the description provided about the problem which describes “Training takes years (3-5) for a new person to understand and have good knowledge of the data and know if the data make sense or not.” Its not clear that routines can be written to account for that much human understanding.