Forecast Precipitation Data
Factors Influencing FMAP Calculations

Objectives

This section covers the evaluation of forecast precipitation (QPF) to be used in the computation of forecast MAP (FMAP). After completing this section, the reader should know:

• The three general types of QPF produced by the National Centers for Environmental Prediction (NCEP).
• How the RFC HAS function evaluates and edits NCEP products and produces QPF for input to the hydrologic modeling process, and the effects and benefits of these HAS operations during various meteorological situations.
• Why, when evaluating QPFs produced by the HAS function, it is important to consider possible relationships between the type of weather system being forecast and spatial distribution of precipitation, precipitation type, the likely influence of terrain, and forecast reliability with increasing time horizon.

At some point in the process of reviewing modeling system inputs, the forecaster must evaluate quantitative precipitation forecast (QPF) input and its potential impacts on hydrologic forecasts. As previously discussed in Topic 1 (RFC Preprocessors), QPFs are provided to the RFC by the NCEP. The forecaster should be aware of the three general types of QPF that may be input to the hydrologic modeling system:

• Short-term QPF. This QPF generally covers one or more days into the future and originates from products developed by HPC forecasters. As described under Topic 1, this QPF takes point or areal form depending on which part of the U.S. is being covered. The HPC QPFs are reviewed and adjusted by HAS forecasters as necessary.
• Model-based QPF. This QPF is direct output from NCEP atmospheric models and may or may not be currently used in your RFC's hydrologic modeling operations.
• Medium- and long-term QPF. This QPF is derived from atmospheric models and/or long-term outlook products from the Climate Prediction Center (CPC). It is used in ensemble hydrologic modeling operations for more extented time frames.

This section focuses primarily on the first type of QPF, since the near-term product has the strongest influence on river forecasts for the next few days. A forecaster's overriding concern about near-term QPF should be its potential impact on modeling system output and hydrologic forecasts. The main question to ask is essentially this - "given the area covered by the QPF and the depths involved, what is the likely impact on the hydrologic modeling system?" There have been instances where QPFs have caused an RFC's hydrologic modeling system to simulate record floods, when in fact the areal extent of significant forecast precipitation was far too large. Improved QPFs and better understanding of their hydrologic impacts by those who provide QPFs have reduced the occurance of this problem. Still, developing a "hydrometeorological sense" of the potential impacts of forecast precipitation through analysis of past events is highly useful.

It is the HAS function's role to provide the best possible QPF input to the hydrologic forecaster. However, the hydrologic forecaster should be aware of several factors which influence QPF accuracy, including:

• Model and forecaster bias. The HPC forecaster developing the original QPF product uses atmospheric model output as a starting point and then applies extensive forecasting expertise. It is well documented that certain atmospheric models have "tendencies." For example, one model may tend to put precipitation slightly north of the frontal line in certain situations and another may tend to underpredict convective precipitation. Also, multiple versions exist for some models used in the QPF process - i.e., operational and "experimental" versions. Models are frequently modified and the forecaster should understand the potential for this. In addition to model tendencies, there is also potential for biases directly related to the abilities, experience, and confidence of the forecaster producing the QPF - both at the HPC and RFC (HAS) level.
• Smoothness of QPF lines. The QPF isoheyts in manual QPF products have a tendency to be smoother and more symetrical than the typical precipitation pattern as observed in a post-event mode. This is particularly true in convective events or situations where locally heavy areas of precipitation are embedded within a larger area of stratiform precipitation.
• Effects of HAS editing. If the HPC products have been edited by a forecaster working the HAS function, the internal consistency between short-interval and daily QPF should be ensured. For example, considering a 1-day product consisting of four 6-hour QPFs, precipitation at any point as computed by summing up the four 6-hour periods should be equal to the 24-hour QPF.
• Storm type and terrain. Storm type and terrain are major issues in the accuracy of QPF. Obviously convective events lead to more uncertainty in timing, amount, and location of precipitation when compared to events that are more stratiform in nature. Terrain influences (such as orographic effects) also can have a significant impact on forecast of both convective and stratiform events.
• Forecast time period. In general, the longer the forecast horizon, the more uncertainty there will be in QPF. However, this uncertainty varies with season, type of precipitation event, and the area of the U.S. being considered. For example, a forecaster in the western U.S. can be reasonably certain a day or more in advance of a winter event that the heaviest precipitation will occur in certain high-elevation basins. It is considerably more difficult for a forecaster in the central U.S. to be certain about which individual basins will receive the heaviest precipitation from a pending summer frontal event.

A good way to improve one's ability to evaluate the validity of QPF is to periodically review verification statistics maintained on the Web page of the National Precipitation Verification Unit (NPVU).