The Subtle Yet Fundamental Differences Between Deterministic and Probabilistic Weather Forecasting (credit: National Weather Service and Tropical Tidbits)
DISCUSSION: For many years, weather forecasting has proven challenging for forecasters and researchers. Marked by consistent improvement yet continual obstacles, the nature of forecasting any type of weather event from benign showers to a full-scale severe weather outbreak is loaded with stochasticity. But at the core of weather forecasting, two schools of thought dominate the practice: deterministic and probabilistic forecasting. Each one of these is subtly different at the surface, but fundamentally they have their characteristic differences.
Deterministic forecasts are based specifically on a given value or range for an area at a given time (e.g., temperature at morning rush hour or evening commute). This is the kind of product we are used to seeing on forecast bulletins and on local news media. Examples of a deterministic forecast include the first tweet above with a range of values for potential ice accumulation over central Oklahoma. A precise value or time is important for the general public as it gives people a frame of reference for what to expect.
Probabilistic forecasts take on a different approach and instead focus on the likelihood that a parameter of any weather event is likely to exceed or occur in a given area. There are indeed various tools that facilitate the growth and understanding of improving forecast accuracy through probabilistic forecasting methods. Mentioned in a previous article with more detail, ensembles are different iterations with parameters tuned slightly differently to reflect differing outcomes. This approach sacrifices a specific (or range) number in exchange for a probability of occurrence beyond a certain threshold (ex: probability of rainfall total greater than 0.01 inches suggested by the second tweet above). Yes, it may seem tricky given that it’s different than the accustomed way, but it is meant to illustrate the difference and carries an emphasis of its own regard.
Agencies like NOAA’s National Severe Storms Laboratory are leading projects such as the Warn-on-Forecast experiment which utilizes sophisticated and refined modeling and data collection techniques to generate weather forecasts based greatly on probability of occurrence. What is the ultimate goal? Forecasters could utilize the added information and produce more accurate forecasts and respond quicker to developing hazards. This in turn could lead to a greater chance of saving life and property in the event of hazardous weather phenomena like tornadoes, flash floods, and large hail by providing ample watch/warning times to the public.
It’s safe to say that as forecasting techniques become more refined with time, these two schools of thought will continue to branch out in technicality and carry a bigger impact in their own regard. What do you think about the differences between the two? Would you rather prefer a deterministic forecast with say a total range of rainfall in a given day, or a probabilistic forecast with a message of likelihood that it will rain/storm on a given day or time? Let us know in the comments!
Image credit: Tropical Tidbits
To learn more about other topics in weather research, be sure to click here!
© 2019 Meteorologist Brian Matilla
DISCUSSION: According to NASA and NOAA the year 2018 was the fourth warmest on record, coming in at around 0.8 degrees C above the 20th century average. This year also began with La Niña conditions in place over the eastern Tropical Pacific Ocean. These conditions held well into the spring and early summer, then during the fall weak El Niño conditions developed. In spite of the warm year globally, much of the central USA experienced cool conditions in March and April 2018, followed by a warm May and June. Also, a strong cold spell dominated much of the central USA from mid-October through November.
The phenomenon described last year, the pool of warm oceanic water in the Northeast Pacific known colloquially as “the blob” (e.g. Bond et al. 2015, Pinhero et al. 2019), was present again this year. But, the “blob” did not make much news in 2018, see last year’s publication for more on this event.
Here, we perform an overview the blocking occurrences in 2018 using the University of Missouri blocking event archive (http://weather.missouri.edu/gcc). We will examine the blocking occurrences for each region of the Northern Hemisphere (NH) and Southern Hemisphere (SH) separately, and discuss a few recent trends in blocking activity.
a.The Northern Hemisphere
As noted in last year’s installment, the number of blocking events that occur annually has been higher since about 2000 than the previous 30-year period (1970- 1999), and a new publication (Lupo et al. 2019) will highlight these trends. During 2018, 50 blocking events occurred over the entire NH, which is higher than last year’s total (40) and quite a bit higher than the mean early 21st century occurrences (38). Since we typically expect +/- 8.5 events, 2018 was a “blocky” year. The persistence of 2018 blocking events was similar to their climatological mean for early 21st century blocks (about 9 days), and their intensity was close to the climatological mean strength as well.
Over the Atlantic Region (80 degrees W – 40 degrees E longitude) in 2018, there were 22 blocking events that occurred and this is almost 40% more than the regional mean. We have stated that the occurrence of blocking can be episodic, and during 2018, 13 of these Atlantic Region blocking events occurred over Eastern Europe and Western Russia. Three of these occurred during October and November in particular. The first one during mid-October caused a strong warm spell across much of Eastern Europe and Western Russia (Fig. 1a), and in some places the warmth was record setting. The latter two blocking events occurred during November and were about two weeks in duration each. The first of these was a moderately strong event, while the second was classified as strong. These events led to warmer than normal conditions over northeastern Europe and cooler than normal conditions from Ukraine to the Urals (Fig. 1b) during the month of November.
Within the Pacific Region (140 degrees E- 100 degrees W), the 2018 blocking occurrence (13) was close to the climatological normal (12) in number and duration (9-10 days). For the second straight year, most of these blocking events (11) occurred over the Northeast Pacific, but unlike last year, these were distributed throughout the year. One event occurred during mid-October (9-19 October), and combined with the Atlantic Region event described above, resulted in a very cold month for the western 2/3 of North America (Fig. 1a). Thus, North America was caught in the middle of a NH simultaneous blocking episode, which is not exactly rare. However, when the impact North America tend to anchor in persistent cool conditions. Also, Nunes et al. (2017) and references therein show extreme cold over North America is typically associated with blocking in the eastern Pacific Region. As we stated last year, the re-emergence of the Pacific Region “ridiculously resilient ridge” provided the impetus for more Pacific blocking. This also caused more blocking to occur throughout 2018 in the east Pacific. This prevalence for blocking over the eastern Pacific in 2018 led to Alaska experiencing a very warm year as seen in Fig. 1. The early winter saw very little snow over the interior of Alaska.
Figure 1. The Northern Hemisphere surface temperature anomaly (oC) for a) mid-October, 2018 (left), and b) November 2018 (right).
In 2018, it was the Continental Region (Weidenmann et al. 2002) that experienced more than double the number of events that during the previous year (2017). This was nearly 50% more than typical as well. These 15 blocking events were sprinkled over the Asian Continent throughout the year, and for the second year in a row, none occurred over North America in 2018. But, as shown in many studies, the occurrence of blocking over North America is comparatively rare.
a.The Southern Hemisphere
In the SH, there were 27 events during 2018 which is the most since record began to be kept in 1970. This breaks the record previously set in 2013 (24), and this is about a 65% greater frequency of occurrence over the annual climatological value (16.5). Weidenmann et al. (2002) demonstrated that most blocking events occur in the South Pacific and during the months of May and June. The record setting year was paced by the occurrence of 19 events over the South Pacific, and seven over the Indian Ocean sector. Normal for these two regions is 12 and three events, respectively. Like last year, the normal peak time only involved five SH block occurrences (late fall - May and June). Also, following 2017, the spring period from October to December saw six block occurrences. This time of the year is very quiet normally in the SH with respect to blocking activity. Most of the blocking events (16) occurred over the southwest Pacific from Australia to New Zealand and near the dateline throughout the year. This resulted in very warm temperatures over the western Pacific in 2018 (Fig. 2a), and Australian heat was often in the news in late 2018 into early 2019. Additionally Argentina and Brazil were cooler than normal.
The SH blocking of 2018 was a little less persistent than typical, the mean event lasting for seven days (compared to eight typically). During the year, only three events persisted for more than 10 days. These were a 17-day event near Australia in October, and two events (10 and 12 days) during the month of May. One of these May events occurred over the western Pacific and the other over the eastern Pacific. This double blocking event resulted in a temperature pattern for a 12 day period that mimicked the year overall in general (Fig. 2). Note than much of South America experienced cooler winter season temperatures at this time. In spite of the increased occurrence of SH blocking in 2018, the intensity of these events was very close to the climatological mean.
Figure 2: The Southern Hemisphere surface temperature anomaly (oC) for a) all of 2018 (left), and b) 24 May – 5 June 2018 (right).
In summary, for the third consecutive year, the number of blocking events globally was up (77 events). During 2018, there were 26% more blocking events globally than in 2017 (61) and this difference was accounted for by positive anomalies in both hemispheres. Only the NH Pacific and SH Atlantic showed blocking occurrences near the climatological norm, all other regions discussed were greater than normal. This year there were not any blocking events occurring in either hemisphere that made it onto the list of the top 20 strongest or persistent blocking events on record. Also, the duration and intensity of blocking in both hemispheres were very consistent with those which have occurred since 2000. Finally, blocking episodes were at least partly responsible for anomalous warm temperature conditions over Eastern Europe (especially the fall), the northeast Pacific and Alaska, and the entire western Pacific from Australia to New Zealand during 2018. Blocking also brought cooler conditions to the central USA and South America during their respective fall seasons, and over western Russia up to the Urals during the fall.
Bond, N.A., Cronin, M.F., Freeland H, and Mantua, N., 2015: Causes and impacts of the 2014 warm anomaly in the NE Pacific. Geophysical Research Letters, 42, 3414-3420. DOI: 10.1002/2105GL063306, 2015.
Lupo, A.R., A.D. Jensen, I.I. Mokhov, A.V. Timazhev, T. Eichler, and B. Efe, 2019: Changes in global blocking character during the most recent decades, Under Review, Atmosphere, January, 2019.
Pinheiro, M.C., Ullrich, P.A., and Grotjahn, R., 2018” Atmospheric blocking and intercomparison of objective detection methods: Flow field characteristics. Under Review, Climate Dynamics, January, 2019.
Wiedenmann, J.M., A.R. Lupo, I.I. Mokhov, and E. Tikhonova, 2002: The Climatology of Blocking Anticyclones for the Northern and Southern Hemisphere: Block Intensity as a Diagnostic. Journal of Climate, 15, 3459-3473.
Anthony R Lupo is a professor of Atmospheric Science specializing in the study of blocking anticyclone and jet stream dynamics at the University of Missouri and contributor to The Global Climate and Weather Center.
To learn more about other global weather research topics, click here!
© 2019 Meteorologist Anthony Lupo
DISCUSSION: In order for cloud droplets to form, they must have something to condense onto in our atmosphere (e.g., grain of dust, sea salt, etc.) Thus, it is thought that aircraft, for example, can aid cloud formation by emitting particles in their exhaust in addition to water vapor. However, scientists in Finland conducted a study where they found a different way for aircraft to potentially enhance precipitation processes. The picture above (image credit: Michael Bryant-Mode) is a dramatic illustration of the potential interaction between aircraft and clouds.
We have to first understand some basic ideas about precipitation formation before understanding the results of the study. When cloud droplets or cloud ice crystals first form, they are too small to fall as precipitation. In a pure liquid or pure frozen cloud, bigger droplets/crystals fall faster than smaller ones, collide with and stick to the smaller droplets/crystals, and eventually become large enough to fall as precipitation. In our atmosphere, water often doesn't freeze at 32 degrees Fahrenheit, but can exist in liquid form down to -40 degrees Fahrenheit (i.e., supercooled water). Thus, clouds can and often do contain a mixture of ice and liquid water. In this situation, the ice grows at the expense of the liquid, and this growth process is often much quicker than if the cloud was pure ice or pure liquid.
Imagine there is a cloud of supercooled liquid water (no ice) through which an aircraft flies. As the plane's wings and/or propeller moves through the air, the pressure and density of the air change such that temperature rapidly drops in a small area. This temperature decrease can result in the freezing of some of the supercooled water which can then trigger the accelerated ice growth process described above. These large ice crystals can then fall faster, collide with other smaller ice crystals, and grow even faster. Thus, even if aircraft produced no exhaust, the study from the Finish scientists indicated that 6-14 times more precipitation could be produced over a small area than if no aircraft flew through the cloud.
This is an example of another way that human activity can potentially influence the weather, perhaps in a way that we haven't thought of before.
To learn more about other topics in weather research, be sure to click here!
© 2019 Meteorologist Dr. Ken Leppert II