It’s a national embarrassment. It has resulted in large unnecessary costs for the U.S. economy and needless endangerment of our citizens. And it shouldn’t be occurring.
What am I talking about? The third rate status of numerical weather prediction in the U.S. It is a huge story, an important story, but one the media has not touched, probably from lack of familiarity with a highly technical subject. And the truth has been buried or unavailable to those not intimately involved in the U.S. weather prediction enterprise. This is an issue I have mentioned briefly in previous blogs, and one many of you have asked to learn more about. It’s time to discuss it.
Weather forecasting today is dependent on numerical weather prediction, the numerical solution of the equations that describe the atmosphere. The technology of weather prediction has improved dramatically during the past decades as faster computers, better models, and much more data (mainly satellites) have become available.
- Supercomputers are used for numerical weather prediciton.
U.S. numerical weather prediction has fallen to third or fourth place worldwide, with the clear leader in global numerical weather prediction (NWP) being the European Center for Medium Range Weather Forecasting (ECMWF). And we have also fallen behind in ensembles (using many models to give probabilistic prediction) and high-resolution operational forecasting. We used to be the world leader decades ago in numerical weather prediction: NWP began and was perfected here in the U.S. Ironically, we have the largest weather research community in the world and the largest collection of universities doing cutting-edge NWP research (like the University of Washington!). Something is very, very wrong and I will talk about some of the issues here. And our nation needs to fix it.
But to understand the problem, you have to understand the competition and the players. And let me apologize upfront for the acronyms.
In the U.S., numerical weather prediction mainly takes place at the National Weather Service’s Environmental Modeling Center (EMC), a part of NCEP (National Centers for Environmental Prediction). They run a global model (GFS) and regional models (e.g., NAM).
The Europeans banded together decades ago to form the European Center for Medium-Range Forecasting (ECMWF), which runs a very good global model. Several European countries run regional models as well.
The United Kingdom Met Office (UKMET) runs an excellent global model and regional models. So does the Canadian Meteorological Center (CMC).
There are other major global NWP centers such as the Japanese Meteorological Agency (JMA), the U.S. Navy (FNMOC), the Australian center, one in Beijing, among others. All of these centers collect worldwide data and do global NWP.
The problem is that both objective and subjective comparisons indicate that the U.S. global model is number 3 or number 4 in quality, resulting in our forecasts being noticeably inferior to the competition. Let me show you a rather technical graph (produced by the NWS) that illustrates this. This figure shows the quality of the 500hPa forecast (about halfway up in the troposphere–approximately 18,000 ft) for the day 5 forecast. The top graph is a measure of forecast skill (closer to 1 is better) from 1996 to 2012 for several models (U.S.–black, GFS; ECMWF-red, Canadian: CMC-blue, UKMET: green, Navy: FNG, orange). The bottom graph shows the difference between the U.S. and other nation’s model skill.
You first notice that forecasts are all getting better. That’s good. But you will notice that the most skillful forecast (closest to one) is clearly the red one…the European Center. The second best is the UKMET office. The U.S. (GFS model) is third…roughly tied with the Canadians.
Here is a global model comparison done by the Canadian Meteorological Center, for various global models from 2009-2012 for the 120 h forecast. This is a plot of error (RMSE, root mean square error) again for 500 hPa, and only for North America. Guess who is best again (lowest error)?–the European Center (green circle). UKMET is next best, and the U.S. (NCEP, blue triangle) is back in the pack.
Lets looks at short-term errors. Here is a plot from a paper by Garrett Wedam, Lynn McMurdie and myself comparing various models at 24, 48, and 72 hr for sea level pressure along the West Coast. Bigger bar means more error. Guess who has the lowest errors by far? You guessed it, ECMWF.
I could show you a hundred of these plots, but the answers are very consistent. ECMWF is the worldwide gold standard in global prediction, with the British (UKMET) second. We are third or fourth (with the Canadians). One way to describe this, is that the ECWMF model is not only better at the short range, but has about one day of additional predictability: their 8 day forecast is about as skillful as our 7 day forecast. Another way to look at it is that with the current upward trend in skill they are 5-7 years ahead of the U.S.
Most forecasters understand the frequent superiority of the ECMWF model. If you read the NWS forecast discussion, which is available online, you will frequently read how they often depend not on the U.S. model, but the ECMWF. And during the January western WA snowstorm, it was the ECMWF model that first indicated the correct solution. Recently, I talked to the CEO of a weather/climate related firm that was moving up to Seattle. I asked them what model they were using: the U.S. GFS? He laughed, of course not…they were using the ECMWF.
A lot of U.S. firms are using the ECMWF and this is very costly, because the Europeans charge a lot to gain access to their gridded forecasts (hundreds of thousands of dollars per year). Can you imagine how many millions of dollars are being spent by U.S. companies to secure ECMWF predictions? But the cost of the inferior NWS forecasts are far greater than that, because many users cannot afford the ECMWF grids and the NWS uses their global predictions to drive the higher-resolution regional models–which are NOT duplicated by the Europeans. All of U.S. NWP is dragged down by these second-rate forecasts and the costs for the nation has to be huge, since so much of our economy is weather sensitive. Inferior NWP must be costing billions of dollars, perhaps many billions.
The question all of you must be wondering is why this bad situation exists. How did the most technologically advanced country in the world, with the largest atmospheric sciences community, end up with third-rate global weather forecasts? I believe I can tell you…in fact, I have been working on this issue for several decades (with little to show for it). Some reasons:
1. The U.S. has inadequate computer power available for numerical weather prediction. The ECMWF is running models with substantially higher resolution than ours because they have more resources available for NWP. This is simply ridiculous–the U.S. can afford the processors and disk space it would take. We are talking about millions or tens of millions of dollars at most to have the hardware we need. A part of the problem has been NWS procurement, that is not forward-leaning, using heavy metal IBM machines at very high costs.
2. The U.S. has used inferior data assimilation. A key aspect of NWP is to assimilate the observations to create a good description of the atmosphere. The European Center, the UKMET Office, and the Canadians using 4DVAR, an advanced approach that requires lots of computer power. We used an older, inferior approach (3DVAR). The Europeans have been using 4DVAR for 20 years! Right now, the U.S. is working on another advanced approach (ensemble-based data assimilation), but it is not operational yet.
3. The NWS numerical weather prediction effort has been isolated and has not taken advantage of the research community. NCEP’s Environmental Modeling Center (EMC) is well known for its isolation and “not invented here” attitude. While the European Center has lots of visitors and workshops, such things are a rarity at EMC. Interactions with the university community have been limited and EMC has been reluctant to use the models and approaches developed by the U.S. research community. (True story: some of the advances in probabilistic weather prediction at the UW has been adopted by the Canadians, while the NWS had little interest). The National Weather Service has invested very little in extramural research and when their budget is under pressure, university research is the first thing they reduce. And the U.S. NWP center has been housed in a decaying building outside of D.C.,one too small for their needs as well. (Good news… a new building should be available soon).
4. The NWS approach to weather related research has been ineffective and divided. The governmnent weather research is NOT in the NWS, but rather in NOAA. Thus, the head of the NWS and his leadership team do not have authority over folks doing research in support of his mission. This has been an extraordinarily ineffective and wasteful system, with the NOAA research teams doing work that often has a marginal benefit for the NWS.
5. Lack of leadership. This is the key issue. The folks in NCEP, NWS, and NOAA leadership have been willing to accept third-class status, providing lots of excuses, but not making the fundamental changes in organization and priority that could deal with the problem. Lack of resources for NWP is another issue…but that is a decision made by NOAA/NWS/Dept of Commerce leadership.
This note is getting long, so I will wait to talk about the other problems in the NWS weather modeling efforts, such as our very poor ensemble (probabilistic) prediction systems. One could write a paper on this…and I may.
I should stress that I am not alone in saying these things. A blue-ribbon panel did a review of NCEP in 2009 and came to similar conclusions (found here). And these issues are frequently noted at conferences, workshops, and meetings.
Let me note that the above is about the modeling aspects of the NWS, NOT the many people in the local forecast offices. This part of the NWS is first-rate. They suffer from inferior U.S. guidance and fortunately have access to the ECMWF global forecasts. And there are some very good people at NCEP that have lacked the resources required and suitable organization necessary to push forward effectively.
This problem at the National Weather Service is not a weather prediction problem alone, but an example of a deeper national malaise. It is related to other U.S. issues, like our inferior K-12 education system. Our nation, gaining world leadership in almost all areas, became smug, self-satisfied, and a bit lazy. We lost the impetus to be the best. We were satisfied to coast. And this attitude must end…in weather prediction, education, and everything else… or we will see our nation sink into mediocrity.
The U.S. can reclaim leadership in weather prediction, but I am not hopeful that things will change quickly without pressure from outside of the NWS. The various weather user communities and our congressional representatives must deliver a strong message to the NWS that enough is enough, that the time for accepting mediocrity is over. And the Weather Service requires the resources to be first rate, something it does not have at this point.
* * *
Saturday, April 7, 2012
Lack of Computer Power Undermines U.S. Numerical Weather Prediction (Revised)
In this blog, I will describe in some detail one major roadblock in giving the U.S. state-of-the-art weather prediction: inadequate computer resources. This situation should clearly have been addressed years ago by leadership in the National Weather Service, NOAA, and the Dept of Commerce, but has not, and I am convinced will not without outside pressure. It is time for the user community and our congressional representatives to intervene. To quote Samuel L. Jackson, enough is enough. (…)
|Courtesy of Bill Lapenta, EMC.|
If you would expect the U.S. has a lot more computer power to balance all these responsibilities and tasks, you would be very wrong. Right now the U.S. NWS has two IBM supercomputers, each with 4992 processors (IBM Power6 processors). One computer does the operational work, the other is for back up (research and testing runs are done on the back-up). About 70 teraflops (trillion floating points operations per second) for each machine.
|NCEP (U.S.) Computer|
The UKMET office, serving a far, far smaller country, has two newer IBM machines, each with 7680 processors for 175 teraflops per machine.
Here is a figure, produced at NCEP that compares the relative computer power of NCEP’s machine with the European Centre’s. The shading indicates computational activity and the x-axis for each represents a 24-h period. The relative heights allows you to compare computer resources. Not only does the ECMWF have much more computer power, but they are more efficient in using it…packing useful computations into every available minute.
|Courtesy of Bill Lapenta, EMC|
The Canadians? They have TWO machines like the European Centre’s!
So what kind of system does NCEP require to serve the nation in a reasonable way?
To start, we need to double the resolution of our global model to bring it into line with ECMWF (they are now 15 km global). Such resolution allows the global model to model regional features (such as our mountains). Doubling horizontal resolution requires 8 times more computer power. We need to use better physics (description of things like cloud processes and radiation). Double again. And we need better data assimilation (better use of observations to provide an improved starting point for the model). Double once more. So we need 32 times more computer power for the high-resolution global runs to allow us to catch up with ECMWF. Furthermore, we must do the same thing for the ensembles (running many lower resolution global simulations to get probabilistic information). 32 times more computer resources for that (we can use some of the gaps in the schedule of the high resolution runs to fit some of this in…that is what ECMWF does). There are some potential ways NCEP can work more efficiently as well. Right now NCEP runs our global model out to 384 hours four times a day (every six hours). To many of us this seems excessive, perhaps the longest periods (180hr plus) could be done twice a day. So lets begin with a computer 32 times faster that the current one.
Many workshops and meteorological meetings (such as one on improvements in model physics that was held at NCEP last summer—I was the chair) have made a very strong case that the U.S. requires an ensemble prediction system that runs at 4-km horizontal resolution. The current national ensemble system has a horizontal resolution about 32 km…and NWS plans to get to about 20 km in a few years…both are inadequate. Here is an example of the ensemble output (mean of the ensemble members) for the NWS and UW (4km) ensemble systems: the difference is huge–the NWS system does not even get close to modeling the impacts of the mountains. It is similarly unable to simulate large convective systems.
|Current NWS( NCEP) “high resolution” ensembles (32 km)|
|4 km ensemble mean from UW system|
A real champion within NOAA of the need for more computer power is Tom Hamill, an expert on data assimilation and model post-processing. He and colleagues have put together a compelling case for more NWS computer resources for NWP. Read it here.
Back-of-the-envelope calculations indicates that a good first step– 4km national ensembles–would require about 20,000 processors to do so in a timely manner–but it would revolutionize weather prediction in the U.S., including forecasting convection and in mountainous areas. This high-resolution ensemble effort would meld with data assimilation over the long-term.
And then there is running super-high resolution numerical weather prediction to get fine-scale details right. Here in the NW my group runs a 1.3 km horizontal resolution forecast out twice a day for 48h. Such capability is needed for the entire country. It does not exist now due to inadequate computer resources.
The bottom line is that the NWS numerical modeling effort needs a huge increase of computer power to serve the needs of the country–and the potential impacts would be transformative. We could go from having a third-place effort, which is slipping back into the pack, to a world leader. Furthermore, the added computer power will finally allow NOAA to complete Observing System Simulation Experiments (OSSEs) and Observing System Experiments (OSEs) to make rational decisions about acquisitions of very expensive satellite systems. The fact that this is barely done today is really amazing and a potential waste of hundreds of millions of dollars on unnecessary satellite systems.
But do to so will require a major jump in computational power, a jump our nation can easily afford. I would suggest that NWS’s EMC should begin by securing at least a 100,000 processor machine, and down the road something considerably larger. Keep in mind my department has about 1000 processors in our computational clusters, so this is not as large as you think.
|For a country with several billion-dollar weather disasters a year, investment in reasonable computer resrouces for NWP is obvious.|
Yes, a lot of money, but I suspect the cost of the machine would be paid back in a few months from improved forecasts. Last year we had quite a few (over ten) billion-dollar storms….imagine the benefits of forecasting even a few of them better. Or the benefits to the wind energy and utility industries, or U.S. aviation, of even modestly improved forecasts. And there is no doubt such computer resources would improve weather prediction. The list of benefits is nearly endless. Recent estimates suggest that normal weather events cost the U.S. economy nearly 1/2 trillion dollars a year. Add to that hurricanes, tornadoes, floods, and other extreme weather. The business case is there.
As someone with an insider’s view of the process, it is clear to me that the current players are not going to move effectively without some external pressure. In fact, the budgetary pressure on the NWS is very intense right now and they are cutting away muscle and bone at this point (like reducing IT staff in the forecast offices by over 120 people and cutting back on extramural research). I believe it is time for weather sensitive industries and local government, together with t he general public, to let NOAA management and our congressional representatives know that this acute problem needs to be addressed and addressed soon. We are acquiring huge computer resources for climate simulations, but only a small fraction of that for weather prediction…which can clearly save lives and help the economy. Enough is enough.