Using Predictions to Plan: Case Study – La Nina and the Missouri River (1)
Back in November I wrote an entry on whether or not we could use the prediction that we would have La Nina conditions in late 2011 and early 2012 to anticipate, for example, whether or not there would be a another historic flood in the Upper Missouri River. A little personal micro history: During August of 2011, I was at a meeting of a panel
which is writing a report on climate modeling. That meeting included climate-savvy water managers talking about the information from climate models they might find usable. During the meeting on the news
, there was the story that seasonal forecasts predicted there would La Nina conditions in late 2011 and early 2012 ( Climate Prediction Center Monthly Outlook
). I asked people at the meeting how they would use this information in their planning for 2012. To be fair, this question was out of the blue, but I had this idea that this seasonal prediction was definitive information when compared with the information that comes from century-long projections from climate models. The century long climate predictions might provide information that some characteristics of El Nino and La Nina will change. With adequate analysis of this information, interpretation of the information, and then guidance or translation of this information, then informed decisions about, for example, reservoir design might be made. But I was curious, given a forecast for a particular season, what would you do?
I have introduced a lot of terms in that paragraph. I will define some of them.
First for those who need information on El Nino and La Nina, these are names given to two parts of an oscillation observed in the tropical Pacific Ocean. In the El Nino phase, the eastern Pacific, off of Peru for instance, is warm. La Nina is the opposite, the eastern tropical Pacific is cold. This is our best known example of behavior where the atmosphere and ocean behave in concert together – and we have proven that we can predict it. (NOAA LaNina Page
, El Nino @ Wikipedia
) We have known for some time that these changes in the Pacific cause or influence preferential weather patterns in other parts of the world. This excites people about being able to do seasonal prediction. In this case there is some oceanic forcing of the weather – or perhaps, when the ocean is considered part of the weather prediction problem, there is information about what the weather might be like for a particular season in a particular place. Concretely, for example, when there is an El Nino, people who worry about floods in California go on high alert (for example
Translation and guidance - There is a lot of information that comes out of a weather and climate model. All practitioners of modeling know that you can’t simply read off the temperature in Des Moines 9 months from now, much less 90 years in advance. But there is the real possibility that there is usable information in the models if 1) we understand the mechanisms that are responsible for, say, stream flow in the Iowa River, and 2) we have an understanding of the ability or inability of the model to represent those mechanisms. That is, if we can find the right knowledge, often a matter of finding the right people, then we can put together this knowledge in a way that is usable. This is what I mean by translation. It is the translation of knowledge from one discipline expert to another in a way that makes that knowledge usable. That is, to provide guidance. (Lemos and Rood on Useful and Usable
OK – going down that path I introduced another term that I think demands more explanation. Mechanisms – when we look at a specific event like the 2011 Missouri River flood
, we look for what factors come together to cause the flood. In the article that was referenced in the November blog
, it was pointed out that there was an extraordinary snow cover on the Great Plains, and then a lot of rain on that snow, that caused melting, and collectively the accumulation of a lot of water that had to go downstream. So in this case, by mechanisms I mean what caused the event to happen. Perhaps the most important mechanisms that a climate model must represent to be usable for regional problems are those mechanisms that provide water to that region.
I am never quite sure if my style of writing is clarifying or just more confusing, but I get enough positive feedback that I think I clarify points for some – so I hope that the way I laid out this basic information makes sense. One more term - What I want to do is to translate information from observational studies and model predictions and make that information usable by someone. From my teaching the last 7 years, I have concluded that it is this translation of information that is the most essential missing ingredient in the usability of climate knowledge. There is a LOT of information and knowledge, but it is not easy to use.
So in this entry, I want to start the process of information translation. I warn in advance that this is a hazardous path. I am going to look at a few papers, in sub-disciplines of weather and climate, in which I am not expert. Hence, I am likely to make some mistakes, and I am hoping that doing this in public, motivates corrections of those mistakes. I take off down this path, because another thing I have discovered in the past seven years is that people who are not consummate experts in a subject are analyzing information and solving problems all over the world. And, I presume to imagine that I am more expert than most, and I presume to believe people when they tell me that I am reasonably good at translating information across discipline interfaces.
So I all start the analysis– and this is not irrelevant. I flew over a swath of the Great Plains last week, and I was struck by the lack of snow. I read Jeff Master’s blog on the extreme state of the Arctic Oscillation
. At the beginning of every problem I collect information. This information inventory process is essential. With a little luck, you will find information that when all brought together can be synthesized into a solution strategy or at least contribute to informed decision making. In fact, I have tried to structure a template to problem solving for a project I am involved in, and it is here at glisaclimate.org
. (What’s a GLISA
?) I collected together a bunch of references
that I thought might inform my translation. What, I am going to do now is extract the information from some of these references.
The first paper I am going to look at is by Bunkers et al. from the Journal of Climate in 1996
. I chose this paper for a couple of reasons. First, a lot has been written that 2011 Missouri River flood
had a La Nina influence. And, thinking about floods, one usually thinks about did it rain a lot? This paper is something of a sanity check, do we see changes in the rain in the Missouri River basin due to La Nina?Bunkers et al.
paper focuses on the “Northern Plains,” which is approximately North and South Dakota. The Missouri River and the Red River of the North are important drainages for these states, and they were in historic flood in 2011. The authors look at data as far back as the late 1800s. That is about as long as any record that we have in the United States. The short story of their findings is that they find that during El Nino, there is significantly enhanced precipitation in the months April through October that follow the onset of the El Nino. For the La Nina phase they find significantly less precipitation for the months May through August following the onset of La Nina. However, we cannot stop with the conclusion, El Nino = wet, La Nina = dry. El Nino and La Nina are often viewed as 2 year long events, and in the second year following the onset of El Nino it is usually a bit wetter than in years with neither an El Nino or a La Nina, but during April and May of that second year it is drier than average. The second year following the onset of the La Nina, it is in general dry. There is also temperature information in the paper, but I am going to keep my focus on precipitation for now.
Let’s recall the problem we are trying to address; namely, 2011 was a La Nina year with a huge flood on the Missouri River, and another La Nina is predicted for 2012, will we have a similar flood? One of the first things it makes sense to look at is the precipitation in the Missouri River basin. This paper looks at part of the Missouri River basin, and area where there were floods, and at least as far as La Nina is concerned we would expect less, not more, spring time precipitation. This seems contradictory to our 2011 experience.
Returning to the Bunker’s et al. paper, there are years when the relation described above did not hold. Bunker’s et al. extract seemingly robust signals, but there are exceptions to the rule. The exception to the rule requires us to consider the mechanisms that might be in play for a given year. We arrive therefore, at a problem of tailoring the information for a particular application. The relation that Bunkers et al. derived between El Nino / La Nina and precipitation in North and South Dakota is quite strong. So if you look at a climate model and it tells you that there will be more or less intense El Nino and La Nina cycles a century from now, the long-term water planner for Fargo might be able to anticipate the water system needed for her grand children. The statistical information might be enough – might, it requires more thought. For a particular season, however, we can’t use this information in isolation. It is just part of the portfolio.
So we have a sanity check that tells us that, indeed, there is documented variability of precipitation in the Missouri River basin, correlated with La Nina. But, at first blush, the La Nina variability in this region is towards drier conditions. We also, know, that what determines a flood is far more complex than “it rains a lot.” So while looking at the paper above gives us some good information, it motivates me to step back and think about all of the pieces – or mechanisms – that might work in concert to produce a flood. And it motivates me to seek whether or not such events are happenstance, or whether we can use our knowledge to anticipate, better, such extreme events. This series of blogs will go on for a while.
Figure 1. Characteristic position of wintertime jet streams during La Nina. From ClimateWatch Magazine
: “The jet streams are high-altitude, racing rivers of air that can influence the path of storms as they track over North America from the Pacific Ocean. The jet streams meander and shift from day to day, but during La Niña events, they tend to follow paths that bring cold air and storms into the Upper Missouri River Basin. Map based on original graphics from NOAA’s Climate Prediction Center
. Adapted by Richard Rivera & Hunter Allen.”Pilot Project on La Nina and the Missouri River Basin
Link to webinars