- Forecast Confidence
- National Weather Service Products
- Forecast Models
- Special Wildland Fire Guidance Tools
- Long Range Forecast Drivers
The typical Confidence Horizon for an application requiring forecast information with "Very High" confidence/accuracy is about three days. What is critical is that there are frequent shifts in this relationship that both expand and contract the amount of time that "Very High" skill is available, thus changing the Confidence Horizon.
This is especially helpful when choosing the appropriate analysis tool in WFDSS (Near-term, Short-term or FS Pro) or applying the weather forecast within them. The concept is also helpful when placing explanations in the analysis or incident notes section of WFDSS.
The atmosphere is chaotic but sometimes there is rhythm in the chaos. Climate outlook scientists study coupled oceanic and atmospheric forcing mechanisms like the El Nino Southern Oscillation (ENSO) cycle and Madden Julian Oscillation (MJO). These forcing mechanisms are tied to several teleconnection or long range-long term correlated weather patterns that determine meteorological impacts across North America.
Forecast skill typically goes up (i.e., rhythm in the chaos) in the mid and long term when certain forcing mechanisms and teleconnection patterns align, thus adjusting the normal Confidence Horizon. Under these circumstances, a monthly or seasonal outlook can have Medium to sometimes High confidence for certain weather patterns such as warmer and drier conditions and sometimes frequent wind events. This is a shift from Low to Very Low confidence under normal situations for longer-range forecasts.
Other teleconnection alignments can lead to mixed atmospheric forcing signals, thus leading to unusually low confidence periods (i.e., conflicting rhythm).
Best Practices for Determining Weather Confidence
Review text discussions routinely issued by NWS, Predictive Services (7 Day Significant Fire Potential), Climate Prediction Center (CPC) and Storm Prediction Center (SPC).
NWS Area Forecast Discussion - This NWS product is intended to provide a well-reasoned discussion of the meteorological thinking which went into the preparation of the gridded forecast. The forecaster will try to focus on the most particular challenges of the forecast. The text will be written in plain language or in proper contractions and confidence can often times be discerned by the verbiage used. At the end of the discussion, there will be a list of all advisories, non-convective watches, and non-convective warnings. The term non-convective refers to weather that is not caused by thunderstorms. An intermediate Area Forecast Discussion will be issued when either significant forecast updates are being made or if interesting weather is expected to occur.
PS 7 Day Significant Fire Potential: Expression of confidence will sometimes be written in the weather discussion portion of the product.
SPC - CONUS Discussions are provided twice daily for the Day 1 and 2 outlooks during the early to late morning hours. The Day 3-8 outlook and discussion is updated once per day during the afternoon.
CPC 6-10 Day, 8-14 Day, and 3-4 Week Discussions: 6-10 and 8-14 day discussions are updated Mon to Fri but not updated by a forecaster during the weekend. 3-4 week discussions are updated on Fri. Confidence is oftentimes mentioned or alluded to in these discussions.
When the weather information gets fuzzy and you are not sure where it lies on the Confidence Horizon it is best to directly consult Predictive Service or NWS forecasters and Incident Meteorologists (IMETs). Don’t shy away from talking to all three sources so you hopefully can gain a clearer picture of the Confidence Horizon.
There are several types of questions you should consider asking.
- Does your office have access to a high-resolution model and use it for grid editing?
- How often do you update your discussions, is there a routine fire weather discussion during the fire season?
- Do you have a long-range forecast expert in the office?
- Does your office send out special email notices or produce webinar or webcasts routinely during the fire season or during significant events?
- Inquire about the spread or consistency in recent modeling?
- Are there any localized critical fire weather growth patterns for the area?
National Weather Service Products
NWS Fire Weather and Enhanced Data Display - National Weather Service’s (NWS) Experimental Enhanced Data Display (EDD) fills a void that currently exists in the NWS. It provides our partners and customers a single interface to access all things GIS related in the NWS. EDD is an extremely powerful and flexible GIS web application. Before the development of EDD, users had to navigate to countless web pages to get at the information they desired. EDD puts this information in one place making it very easy to display and manipulate this data. EDD is hosted on the National Internet Dissemination System (NIDS) and was developed by the Weather Ready Nation Pilot Project in Charleston, WV.
Short-Term Forecast Products (1-7 Days)
NWS Routine Fire Weather Forecasts - Generated twice daily during the area’s main fire season and updated when conditions warrant, these narrative forecasts include a discussion of expected weather events, general weather parameters for the forecast zone over the next 48 hours, and an outlook summary for days 3-5.
NWS Watches and Warnings - General site provides access point to different products with map emphasizing different watches and warnings. Issuance of a warning or watch implies stronger confidence levels in growth conditions between 12-96 hours ahead of an event.
NWS Spot Weather Forecasts - This interface is intended to be used solely for the relay of forecast information to the National Weather Service. Submissions sent through this online form are intended for internal agency use.
NWS Weather Prediction Center (WPC) - The WPC provides forecast products including deterministic and probabilistic quantitative precipitation forecasts and discussions, short-term and medium-range forecasts, and surface analysis.
National Digital Forecast Database (NDFD) and Wildand Fire Assessment System NDFD Point Forecasts (CONUS Only) - The NWS's NDFD graphic products are derived from a prescribed set of data contained within the NDFD. These graphics are representations of the official NWS digital forecast. Forecast information for the Canadian portion of the Great Lakes is for informational purposes only and does not constitute an official forecast. In CONUS only, WFAS provides a tool that allows NDFD forecasts to be queried for individual latitude-longitude locations.
NWS Storm Prediction Center - This Center provides convective and fire weather outlooks out to 8 days across the CONUS. They also provide special fire, convective and precipitation weather interrogation tools such as the High Resolution Ensemble Forecast (HREF) and Short Range Ensemble Forecast (SREF) viewers.
Predictive Services 7 Day Significant Fire Potential – Product that highlights the likelihood that a wildland fire event will require mobilization of additional resources from outside the area in which the fire situation originates. Weather, fuels, resource availability and probabilities of ignition are assessed each day to create the probability levels out 7 days. Forecasts of Energy Release Component, Burning Index, 10-100 and 1000 hr dead fuel moistures can also be reviewed using the main webpage.
- High Risk = Days in this category represent a probability of 20% or higher for a significant fire.
Extended Forecast Products (1 Week to Several Months)
NWS Climate Prediction Center Extended Outlooks - The Climate Prediction Center (CPC) is responsible for issuing seasonal climate outlook maps for 1-13 months in the future. In addition, the CPC issues extended range outlook maps for 6-10 and 8-14 days as well as several special outlooks, such as degree day, drought and soil moisture, and a forecast for daily ultraviolet (UV) radiation index. Many of the outlook maps have an accompanying technical discussion.
Predictive Services Seasonal Outlooks – At the beginning of each month, seasonal outlooks are posted. These outlooks serve as assessments of the current and projected fuels, weather and fire potential out four months.
NWS Climate Prediction Center Expert Assessments - Includes Hazards Assessment, ENSO update, and others. Climate Prediction Center (CPC) meteorologists and oceanographers review climate and weather observations and data along with model results and assess their meaning, significance, current status, and likely future climate impacts. Their findings are issued as assessments, advisories, special outlook discussions, and bulletins.
Drought Outlooks and Seasonal Climate Forecasts - These forecasts show predicted trends for areas experiencing drought depicted in the U.S. Drought Monitor, as well as indicating areas where new droughts may develop.
Meteorologists produce weather forecasts based largely on interpreting and validating weather models. Several models are available with varying update cycles, output time intervals, temporal range and spatial resolution. Models are semi-routinely adjusted based on new research, new observing techniques and increased computer processing power.
Deterministic and Ensemble models have different classifications such as global, regional and convective allowing (CAMS). High resolution models (≤ 5 km) are generally convective allowing and have 15 min to 1-hour time-steps. They can resolve terrain induced and altered wind regimes, small scale boundary passages and banded and cellular precipitation features. Due to the differing scales, physics and initial conditions CAMs models can produce widely differing solutions when compared to each other thus making it tricky for forecasters to choose the right solution.
Deterministic, or operational, models use the best estimate of the current state of the weather (initial conditions) to provide a singular solution out in time. Outputs generally contain higher spatial resolution compared to ensemble outputs.
|Deterministic Models||Updates per day||Output Timestep (hrs)||Range (hrs)||Resolution|
|Global Forecast Model (GFS)||4||6||384||13 km|
|European Computer Forecast Model (ECMWF)||2||12||240||9 km|
|North American Mesoscale Model (NAM)||4||1 to 3||60 to 84||3, 12, and 22 km|
|Global Environmental Multi-scale Model Canada (GDPS/GEM/CMC)||2 to 4||1 to 6||54 to 240||varies|
|Unified Model United Kingdom (UKMET)||4||6||144||10 km|
|Navy Global Environmental Model (NAVGEM)||4||3 to 6||144 to 180||13 to 31 km|
|High Resolution Rapid Refresh (HRRR)||24||15 min to 1 hr||18 to 36||3 km|
|Weather Research & Forecasting (WRF/different variants)||varies||1||varies but generally not beyond 72 hrs||1.33 to 5 km|
Probabilistic or ensembles are a collection of numerical model outputs that show slightly different possible outcomes. The outcomes are slightly different because each member of the ensemble has been perturbed or initialized with slightly different starting atmospheric conditions. Forecast skill can increase the first 1-2 weeks of a forecast period using these ensemble modeling systems.
|Probabilistic Models||Updates per day||Output Timestep (hrs)||Range (hrs)||Resolution||Members & Control|
|Global Ensemble Forecast System (GFES/GFS ensemble variant)||4||6||384||33 km||21|
|Ensemble Prediction System (EPS/ECMWF ensemble variant)||2||24||360||18 km||52|
|Global Ensemble Prediction System (GEPS/GEM ensemble variant)||2||6||384||66 km||21|
|North American Ensemble Forecast System (NAEFS/blend of GEPS & GFES)||2||6||336||111 km||42|
|Navy Global Environmental Model Ensemble (NAVGEM)||2||6||384||111 km||16|
|High Resolution Ensemble Forecast (HREFv2)||4||1||48||~3 km||8|
|Short Range Ensemble Forecast (SREF)||4||1 to 3||87||16 to 40 km||22 to 26|
|North American Multi-Model Ensemble (NMME)||1||6 to 24||6 months||111 km||7|
The amount of spread in an ensemble model should directly be related to the uncertainty of the forecast because the actual observed weather solution should fall somewhere in between the individual members solutions. Tighter the spread, more confidence in the forecast. Wider the spread, less confidence.
|Ensemble Terms to be familiar with||Definition and Uses|
|Members||Individual model run in an ensemble suite perturbed by slightly different initial conditions. Forecasters look at spread of the members.|
|Mean||Average value of the individual ensemble model members. Smooths out “chaos.”|
|Control||Mimics the operational model run because it uses the same initial conditions but with a courser spatial (vertical and horizontal) resolution. Forecasters compare it to the operational run, if output different then differing spatial resolution has effect on the output.|
|Spread||Usually represented by standard deviation. Larger the deviation, more uncertainty in the forecast.|
|Normalized Spread||Puts standard deviation in context with the general ensemble model behavior. For example, how does the current day 10 forecast standard deviation value compare to the past 30 days of day10 model forecasts? Helps determine proper confidence level.|
There are several types of useful outputs that can result from ensemble weather modeling. For example, NAEFS provides outputs based on reviewing a 3-week period centered on the valid day using a 30 year climate record.
- Standard anomaly: How different is forecast from climatological mean?
- Percentile: Where would the forecast fall with respect to climatology? Max and Min values indicate the forecast falls out of the 30 year climate record thus a potentially significant event.
- Return interval: How often do these forecasts show up in the climatology?
- Probabilities: How many of the ensemble members are producing extreme values?
Special Wildland Fire Guidance Tools
These tools will help identify unusual or important fire weather and danger time periods and give higher confidence to a critical fire growth or slowing pattern. Publications explaining the tools or indexes are provided in the Weather References section of this guide.
Global Wildfire Information System (GWIS) which provides 8 km ECMWF based forecasts of FWI out 10 days across the Globe.
Hot Dry Windy (HDW): Designed to help identify days which are more likely to have adverse weather conditions. Represents the multiplication of maximum wind speed and maximum vapor pressure deficit (VPD).
Severe Fire Danger Index (SFDI): 5 separate classifications (Low-Moderate-High-Very High and Severe) represent the normalized percentile values of combining ERC and BI. Between 2003 to 2016, 75% of all MODIS active large fire pixels and 81% of all fire radiative power occurred when the index was Very High or Severe. NWS NDFD weather forecasts feed this index thus forecasted wind speeds may not adequately represent a locally significant terrain altered wind event especially beyond 3 days.
Fire Potential Index (FPI): Is a linearly scaled from 0 to 100 with 0 occurring when vegetation is near or fully green and 100 hr dead fuels are too moist to burn. FPI is 100 when vegetation is fully cured and 10 hr dead fuel moisture is 2%.
Southern California Fire Buster: Routine 5 km forecasts updated twice per day out 72 hours for southwest portions of CONUS including CA and NV. 1-km forecasts can be requested. FFWI and Large Fire Potential are one of many outputs found in Fire Buster.
Alaska Fire & Fuels (AKFF): CFFDRS Fire Weather Index (FWI) forecasts
Great Lakes Fire & Fuels (GLFF): CFFDRS Fire Weather Index (FWI) forecasts
Long Range Forecast Drivers
It’s the meteorologist’s job to understand coupled atmospheric-oceanic teleconnections but the analyst should be aware of these teleconnections because they are generally referred to in longer range forecast discussions. The analyst will be introduced to a few important teleconnections and/or long-range outlook terms in the following table. It is important to point out that some teleconnection patterns can work together or work against each other. This is called constructive and destructive interference. Research continues in this arena and how climate change impacts traditional teleconnection impacts.
Teleconnections and Climate Forcing Mechanisms
|Coupled oceanic- atmospheric forcing mechanisms||Usefulness and Definition||Confidence Impact|
|Spring Predictability Barrier||Refers to the period of time when the statistical and dynamic climate models have less skill to predict ENSO trends. April-May-June represents this period for the northern hemisphere.||Lower confidence in long range climate outputs when the models are initialized during the spring period.|
|El Nino Southern Oscillation (ENSO)||Irregular periodic variation in winds and sea surface temperature over tropical eastern Pacific. Warming phase is El Nino and cooling phase is La Nina. There is also an ENSO neutral condition. Deep thunderstorm development over the eastern equatorial Pacific will help alter the jet stream and make more moisture available for Pacific storm systems that impact the USA.||Warm or cool phases combined with other climate drivers such as certain phases of the PDO can lead to higher confidence long term outlooks.|
|Pacific Decadal Oscillation (PDO)||A longer lived northern Pacific climate variability that has a negative and positive phase. The jet stream will tend to dip further south over the western United States under a positive phase and move further north under a negative phase.||Positive or negative phases to the PDO combined with other climate drivers such as ENSO conditions can lead to higher confidence long term outlooks.|
|Pacific Meridional Model (PMM)||Describes the interaction between ENSO and PDO teleconnections and has a negative and positive phase.||Determining positive or negative phases of the PMM may help resolve the Spring Barrier issue and ENSO prediction. This should help lead to higher confidence outlooks originating during months of April through June.|
|Modoki||Alternation in normal La Nina and El Nino SST patterns across the tropical Pacific thus disrupting typical teleconnection patterns. “This isn’t your grandfather’s El Nino”.||Confidence becomes varied and can be less depending on other coupled ocean-atmosphere phenomena.|
|Madden Julian Oscillation (MJO)||MJO is a tropical disturbance that propagates eastward around the global tropics with a cycle on the order of 30 to 60 days. They typically are most active during late fall, winter and early spring period. MJO can influence ENSO tendencies.||Active MJO’s can increase forecast confidence for certain weather anomalies across USA.|
|Aleutian Low||Strength of the Aleutian low combined with PMM modes can accentuate or dampen MJO impacts on the jet stream position and storm track across western USA.||Confidence of an active storm track across western US increases when the Aleutian low is stronger than average and combines with a positive PMM and a MJO is present.|
|Arctic Oscillation (AO)||Represent a pattern of differences in air pressure between the Arctic and mid-latitudes. AO’s influence winter weather in the northern hemisphere. Positive phase AO’s are representative of strong pressure systems where low pressure is found over the Arctic and high pressure over mid-latitudes. Negative phase AO’s represent a weaker pressure gradient due to less low pressure across Arctic and less high pressure in the mid-latitudes. AO phases can change in a matter of weeks.||During negative phase AO’s there is higher confidence for cooler air, generally locked up in the Arctic, to eventually spill southward into CONUS while warmer air moves to the north.|
|North Atlantic Oscillation (NAO)||Represents pressure difference between Iceland and the Azores. Azores typically have high pressure and Iceland low pressure. The magnitude of change dictates whether the NAO is in a positive phase or negative phase. Positive phases represent a large change in pressure, while negative phases are a small change. NAOs influence northern hemispheric weather by influencing the jet stream, especially during the winter months.||During positive phase, jet stream generally less undulating with muted, more progressive troughs and ridges. During negative phase, jet stream is represented by high amplitude blocking troughs and ridges with a trough more over eastern portion of CONUS.|
|Atlantic Multi-Decadal Oscillation (AMO)||Represents a coherent mode of natural variability occurring in the North Atlantic. Measurement is based upon average SST anomalies (0 to 80N). Cooler SST’s represent a negative phase. Warmer SST’s represent a positive phase.||Has been used to correlate temperature and precipitation anomalies across northern hemisphere. Positive AMO’s have been linked to long-sustained droughts in Midwest and Southwest with increased Hurricane activity in N. Atlantic.|