Close Window

Penn State University: "A method for predicting heavy rainfall events in the MARFC domain using mesoscale model QPF biases and current climatology"

Final Report


1.1 Introduction

The purpose of this project was threefold, to expand the climatological base for 6 and 24 hourly precipitation in the MARFC sub-basins to at least 10 years, to examine CPS (convective parameterization schemes) for heavy rainfall cases in the MARFC domain to determine which schemes work best in certain environments and to quantitatively estimate and verify QPF from a couple of mesoscale models using multi-sensor estimates as ground truth.

This partnership was a joint effort of MARFC Senior HAS forecaster Patricia Wnek along with CTP lead forecaster Michael Evans and State Climatologist and Penn State instructor Paul Knight. The contributions were MAP data from the early and late 1990's and climo-hydrologic charts derived from this expanded database from MARFC. The NWS created graphs and tables to summarize the new climatology. Installation of a workstation version of the (meso)ETA along with MM5 along with initial conditions for case studies were provided by the NWS at CTP. Model simulations and casebook summations along with the design of the online project were the task of Penn State and its meteorology students.

1.2 Description of Accomplishments

The new mean areal precipitation database was imported into an Informix data management program (the same format as stored at the MARFC) and then sorted for amounts and extremes. This summation data was then transferred to the MARFC where statistical representations of the expanded database were produced for all the basins. These charts were placed on the project web page during the early summer (

The case studies were selected from 3 different convective occurrences over the MARFC domain. They took place in very different environments and therefore afforded the opportunity to study the effects of antecedent atmospheric conditions on the success of the convective parameterization schemes in the mesoscale models. These cases were also part of a special 1 credit seminar class (Meteo 497C) held in conjunction with this project. The cases selected were; August 19,1999 flash floods in central and western Pennsylvania, August 9, 2000 derecho in the Ohio Valley and Middle Atlantic region and the August 12, 2000 flash flood in northern New Jersey.

The third part of the project was the development of a verification scheme to compare the mesoscale models qpf for a heavy rainfall event in the MARFC domain with the new multi-sensor precipitation estimates. Due to the novelty of these fields, this element will be on-going and results will be included on the web page during the months ahead.


There were several important exchanges that took place during this project. The most significant exchange was the development of a course based on the objectives of this project. Meteorology 497C - Mesoscale Quantitative Precipitation - was offered during the Fall 2000 semester. Ten students enrolled in this class (2 graduate and 8 undergraduates). The instructors were the Pi's along with the CTP-SOO, Richard Grumm. There were ten instructional classes followed by four sessions with student presentations.

An exam was administered which contained questions from each of the instructors. The instructors (Pi's) evaluated the student projects and were evaluated by the students concerning the content and presentation of the class material.

A number of people involved in the class (including a PI) learned how to run a workstation version of the meso-eta model as a result of this collaboration.

The NWS provided considerable assistance with the installation and use of the database program Informix for designing the MAP climatology. Penn State provided the climatic statistics for the NWS to develop charts and graphs for the web page.

Plans are underway to offer teletraining to Eastern Region offices on the results of this project.


A paper, written by Michael Evans and Paul Knight, was submitted and accepted for the poster portion of the Extreme Precipitation subconference held at the AMS annual meeting in Albuquerque in January 2001. Patti Wnek, Paul Knight and an undergraduate intern, Ray Willard presented the poster. A mini-version is included with the hard copy of this report.

ABC News interviewed one of the PI's about this project and Penn State's public information sent a news release about the paper and the project. Copies of these publications are included in the mailed report.


4.1 Project benefits to the NWS and Penn State resulting from this collaboration:

Encouraged work on the development of the workstation Eta model, which will aid in future research projects on mesoscale modeling, and aid in operational forecasting.

Class lectures gave forecasters insight on how convective parameterization schemes work to release instability in models, preventing unrealistically large areas of extreme instability to build up over time, and therefore preventing unrealistically large grid-scale precipitation forecasts.

Students in the QPF course became familiarized with the NWS River Forecast Operations and the terms of QPE. While the NWS became more knowledgeable of Excel, Penn State became familiar with Informix, the database of the MARK office.

Case studies gave forecasters insight on the limitations of the QPF forecasts from mesoscale models when significant convection is forecast to occur. In particular, an emphasis on looking at diagnostic fields other than QPF was emphasized.

Case studies and lectures also gave forecasters insight on the importance of examining a range of model solutions, with the idea that outlying quantitative precipitation forecasts will be lower confidence forecasts than forecasts that are closer to the model consensus.

The categorization of the anomaly conditions associated with all previous flood events in the MARK domain became the seedling for a new Comet proposal on developing an early alert system for flood potential.


Thankfully, problems were quite limited. Penn State did have considerable challenges in learning and using Informix as it was installed on both and NT and Linux OS in the State Climate Office. The NWS had difficulty in finding a useable format of the multi-sensor precipitation estimates. In addition, a series of spring and summer dry spells made it difficult to select case studies for this portion of the project. As a result, the Pi's will pursue this element during the months ahead.