October, 2019 Seasonal Forecasts. Part II, NOAA Comparison with JAMSTEC

Written by Sig Silber

We published the NOAA Long-Term Forecast Friday Night (here).  Now we compare the NOAA forecast for Alaska and CONUS with the JAMSTEC forecast. The differences in the weather forecast are less than we might have expected given that NOAA is forecasting ENSO Neutral with an El Nino bias and JAMSTEC is forecasting a weak El Nino with some Modoki characteristics. In addition, we also provide the JAMSTEC World Forecasts with special emphasis on Europe.

Please share this article – Go to the very top of the page, right-hand side for social media buttons. Please feel free to send this article to anyone who you think might find it interesting and or useful.

Now we compare the two NOAA forecasts for ENSO specifically the Nino 3.4 Index with the JAMSTEC forecast. We do it again later in the article in more detail but the comparison here may be easier to see. The first image is the Early-September IRI-CDC ENSO probabilities. The second image is the JAMSTEC proprietary forecast model for the Nino 3.4 Index. The third is the NOAA proprietary model CFSv2. We pay special attention to Winter because the forecasts differ the most in Winter. We have added a rectangular box in red in the second and third graphics to provide focus to the Nino 3.4 forecasts for Winter. Notice that one is above the +0.5C threshold for an El Nino value and the other is below it. We believe this explains the difference in the weather forecast for this Winter expecially temperature.

One of the reasons for doing this comparison is it provides the ability to create our own updated forecast as we see the assumptions made by each agency be confirmed or not. That is especially true when the actuals start coming within the range of the two sets of assumptions. Weather is not linear so it is more complex than scaling the solution to where the actuals are showing up relative to the two versions of the assumptions but with a little imagination one might be able to sort it out. You will be able to better understand what I have just said as we look at the two different forecasts.

C. Now to our full report. (Sections A and B were covered in Part I which can be accessed here)

This report is organized into a summary that has two tables of graphics that show the temperature and precipitation forecasts for the upcoming three seasons, a brief discussion of the ENSO assumptions by both NOAA and JAMSTEC and then the JAMSTEC Agency discussion and then a comparison of the two forecasts which is basically an expansion of the summary table.

Summary of the NOAA and JAMSTEC Forecasts

For those who want a quick synopsis of the two forecasts, below is a summary of the temperature and precipitation forecasts for three time-periods and from left to right the NOAA forecast for Alaska and CONUS (the contiguous mid-latitude U.S), then JAMSTEC for North America (which includes Canada and Mexico), and then JAMSTEC for Europe and surrounding areas. Larger graphics are provided later in the report. (the graphics in the Summary Table are not set up to be clicked on to enlarge). The Summary Table is kind of a tease to keep you reading but you can see the evolution of the weather pattern through Fall, Winter, and then Spring. For some readers, these two sets of summary graphics may be all the information they wish to look at and that is fine.

Temperature

	NOAA Alaska Plus CONUS	JAMSTEC North America	JAMSTEC Europe
Fall SON 2019 for JAMSTEC OND 2019 for NOAA
Winter DJF 2019-2020
Spring MAM 2020

Precipitation

	NOAA Alaska Plus CONUS	JAMSTEC North America	JAMSTEC Europe
Fall SON 2019 for JAMSTEC OND 2019 for NOAA
Winter DJF 2019-2020
Spring MAM 2020

We provided the full NOAA Discussion in Part I. The much shorter JAMSTEC Discussion was published on September 17, 2019 so we have that for our Report tonight.

Sep. 17, 2019 Prediction from 1st Sep., 2019

ENSO forecast:

As predicted earlier, the El Niño Modoki-like state is observed now. The SINTEX-F predicts the El Niño Modoki-like state will persist at least until autumn. We need to be careful of its impact, as it may be different from that of the canonical El Niño. Then, the model predicts that an El Niño-like pattern (looks a mixture of Modoki-type and canonical-type) will appear in the tropical Pacific from winter through the first half of year 2020.

Indian Ocean forecast:

As predicted earlier, a positive Indian Ocean Dipole quickly emerged in the May, and now reached a level similar to that of the strong events of 1994, 1997, and 2006. The model predicts that the positive Indian Ocean Dipole will persist in autumn, and then quickly decay in winter. We may observe co-occurrence of a positive Indian Ocean Dipole and an El Niño Modoki-like state in autumn; this is as we observed in 1994 and 2018.

Regional forecast:

On a seasonal scale, SINTEX-F predicts that most part of the globe will experience a warmer-than-normal condition in boreal autumn, except for southern part of the South American Continent, and some parts of West Africa. In boreal winter, most part of the globe will still experience a warmer-than-normal condition, except for some parts of U.S.A, Argentina, Scandinavia, and Greenland.

As regards to the seasonally averaged rainfall in boreal autumn, a wetter-than-normal condition is predicted for southern coastal area of Alaska, U.S.A., East and West Africa, Europe, most part of Southeast Asia. In contrast, some parts of Canada, Mexico, Brazil, southern Africa, western Russia, northern India, Indonesia, Philippines, Australia, and the Korean Peninsula will experience a drier-than-normal condition. Some of them will be partly due to the expected positive Indian Ocean Dipole (see schematic of a positive Indian Ocean Dipole event: http://www.jamstec.go.jp/aplinfo/sintexf/e/iod/about_iod.html). In particular, we notice that Indonesia and Australia may experience extremely drier than normal condition, owing to the expected co-occurrence of a positive Indian Ocean Dipole and an El Niño Modoki-like state. In boreal winter, a wetter-than-normal condition is predicted for Alaska, northwestern coastal area of Canada, northern part of the South American Continent, northern Argentina, eastern parts of southern Africa, Europe, western Russia, India and southern China. In contrast, southeastern Brazil, western part of southern Africa, Scandinavia, Australia, Indonesia, Philippines, and the Korean Peninsula will experience a drier-than-normal condition.

The model predicts most part of Japan will experience warmer- and wetter-than-normal in autumn as a seasonal average. In winter, most part of Japan will be still warmer, but slightly drier-than-normal.

It is useful to look at the forecasts for the Nino 3.4 Index used to determine if the Sea Surface Temperatures meet the criteria for an El Nino(+0.5 C or higher). Here is a brief summary of the Indices of the two Agencies and how they compare. Three of these graphics were presented earlier. We use them again here and add two more to provide a more detailed analysis.

Indices use a single number to convey information. Sometimes it is useful to look at the forecast Sea Surface Temperatures (SST) that correlate to the indices. I believe that the SST’s directly correlate with the JAMSTEC Nino 3.4 Index and their forecast but NOAA uses a variety of tools so it is not so easy to understand exactly how they do their forecast but there is a lot of information on that in their lengthy discussion which was presented in our Part I Report.

NOAA	JAMSTEC

Updates from JAMSTEC can be found here. Updates from NOAA can be found here. You have to look for the SST row and go to the right where it says “normalized with mask” and click on E3 which provides the latest model run.

All of these forecasts depend to a large extent on what is known about subsurface temperature anomalies. Here is the latest analysis. Updates can be obtained here.

Does the SOI confirm that ENSO Neutral Conditions apply? The SOI is one measure of the extent to what the atmosphere has been impacted by the Eastern Pacific surface pattern which is measured by the Nino 3.4 Index. After all, we are more interested in the impacts to weather in the atmosphere which impacts populated areas than we are about the temperature distribution of the ocean surface along the Equator in the Pacific Ocean.

I am showing the situation this month and last month.

SOI = 10 X [ Pdiff – Pdiffav ]/ SD(Pdiff) where Pdiff = (average Tahiti MSLP for the month) – (average Darwin MSLP for the month), Pdiffav = long term average of Pdiff for the month in question, and SD(Pdiff) = long term standard deviation of Pdiff for the month in question. So really it is comparing the extent to which Tahiti is more cloudy than Darwin, Australia. During El Nino we expect Darwin Australia to have lower air pressure and more convection than Tahiti (Negative SOI especially lower than -7 correlates with El Nino Conditions). During La Nina we expect the Warm Pool to be further east resulting in Positive SOI values greater than +7).

D. Now we begin our comparison of the NOAA and JAMSTEC Forecasts (Focus on next nine months i.e. three seasons)

In this Update, we compare the JAMSTEC temperature and precipitation forecast maps, which are for the World, with the NOAA temperature and precipitation forecast maps that cover only CONUS and Alaska. We do this primarily for educational purposes.

JAMSTEC works in three-month intervals which correspond to seasons and does not change the selection of months each time they update but does so every three months. At that time they drop one season and add another season further in the future. JAMSTEC provides maps for three seasons and the choice of seasons changes every three months. So for one out of three months, the first season lines up perfectly and this is not that month. So for JAMSTEC, we have Fall (SON), Winter(DJF) and Spring (MAM) and for NOAA we have OND for Fall even though OND overlaps with Winter, and we have Winter (DJF) and Spring (MAM) in the comparison.

It is a shame that we do not have the exact same months for the first season comparison but two of the three months are the same. This is a good time to comment on the timing associated with the preparation of the forecasts. For JAMSTEC, it is clear that that their forecasts are said to be based on an Nino 3.4 forecast dated September 1, 2019. And yet the discussion comes out much later this time on September 17, 2019, The forecast maps come out a few days earlier. So we do not know if everything is based on September 1 or if later information is utilized. We suspect that the JAMSTEC model takes a long time to run. For NOAA we have the opposite problem. The maps and discussion are always released on the Third Thursday of the month. But we do not know when they are prepared. There was one clue in the discussion this time. It is our belief that the forecasts and discussion are prepared a few days prior to the release on the Third Thursday. We observed that the forecasts for October seemed to be not consistent with the Week 3 – 4 forecast released the very next day. This is a common problem in meteorology and made more difficult the larger the agency. It should not normally impact the usefulness of the forecasts beyond the immediate next month. 

Presentation of the Forecast Maps.

I show the NOAA Maps first followed by the JAMSTEC maps. I extract North America from the Worldwide JAMSTEC map and use that to compare with the NOAA Maps. I also extract Europe including parts of North Africa and Western Asia from the Worldwide JAMSTEC maps and include those excerpted maps in the summary table at the beginning of this article.

Now we will compare the maps in that summary but with full-sized maps.

Fall

Temperature OND 2019 for NOAA

And here is the SON 2019 temperature forecast for North America that I extracted from the JAMSTEC World Forecast.

And now the JAMSTEC WORLD FORECAST (This month I am using the JAMSTEC comments for Fall and Spring and those comments include Alaska and CONUS. But my detailed comments on Alaska and CONUS appear with the larger graphics above)

Precipitation

NOAA (OND)

And here is the SON 2019 Precipitation Forecast for North America that I extracted from the JAMSTEC World Forecast.

And now the JAMSTEC WORLD FORECAST  (This month I am using the JAMSTEC comments for Fall and Spring and those comments include Alaska and CONUS. But my detailed comments on Alaska and CONUS appear with the larger graphics above)

Winter DJF 2019/2020

Temperature

NOAA

And here is the DJF 2019-2020 Temperature Forecast for North America that I extracted from the JAMSTEC World Forecast.

JAMSTEC WORLD FORECAST (This month I am using the JAMSTEC comments for Fall and Spring and those comments include Alaska and CONUS. But my detailed comments on Alaska and CONUS appear with the larger graphics above)

Precipitation

NOAA

And here is the DJF 2019/2020 Precipitation Forecast for North America that I extracted from the JAMSTEC World Forecast.

JAMSTEC WORLD FORECAST (This month I am using the JAMSTEC comments for Fall and Spring and those comments include Alaska and CONUS. But my detailed comments on Alaska and CONUS appear with the larger graphics above)

Spring (MAM – 2020)

Temperature

NOAA

And here is the MAM -2020 Temperature Forecast for North America that I extracted from the JAMSTEC World Forecast.

JAMSTEC WORLD FORECAST

Precipitation

NOAA

And here is the MAM 2020 NA precipitation I extracted from the JAMSTEC World Map

JAMSTEC WORLD FORECAST

D. Conclusion

Both NOAA and JAMSTEC have forecasts that in one way or another incorporate the value of Nino 3.4. Both forecast it to be positive and close to +0.5C.  NOAA sees it as under +0.5C and JAMSTEC sees it as slightly over +0.5C. JAMSTEC also sees Modoki characteristics in the pattern which probably only means the warmer water along the Equator in the Pacific is a bit farther west than usual for a typical El Nino. This shifts the Walker circulation west a bit. NOAA may take this into account but generally does not use Japanese terminology for the condition of the Equatorial Pacific Ocean which we share. I see that as a weakness in their forecasting approach. Also though it is possible to take Modoki into consideration using statistical methods, it is not easy to do so I suspect that NOAA does not do it. One the other hand, the dynamic methods of JAMSTEC build errors over time. So we have two flawed methodologies competing and thus comparing the results of the two flawed approaches is useful. 

Slight changes in the ENSO Phase are likely to change the actuals from what has been forecast. I do not have a lot of confidence in either of the forecasts but for different reasons. NOAA may not be properly taking into account the Modoki aspect of the situation and JAMSTEC may be overestimating the El Nino bias that is in their forecast which casts doubt on their DJF 2019-2020 forecast. It is the same problem as last month.