Louisiana Coast, Photo by Journo Jen via Flickr

BROWN BAG SEMINAR: Predicting, Detecting, Understanding: Science to Inform Decision Making

The USGS and the Delta Stewardship Council are recruiting the next Delta Lead Scientist who is appointed by the Council based on a recommendation from the Delta Independent Science Board.  As part of the process, each candidate was invited to give a brown bag seminar presentation of their research and experience, as well as their vision for the Delta Science Program.  Ultimately another candidate was chosen, but here is Dr. Denise Reed’s brown bag seminar which discussed using models to make prediction and improve decision making.

Dr. Denise Reed is a coastal geomorphologist by training.  She has degrees in geography from the University of Cainbridge, and has lived and worked in the US for over 30 years.  Her work focuses on coastal marsh sustainability and the role of human activities in modifying coastal systems.  For much of her career, she has been directly involved in restoration planning in coastal Louisiana.  Her work in the Bay Delta has included field studies of tidal marshes.  She also served on numerous boards and advisory panels, and recently served on the development of a science plan for Delta smelt.  She is currently a member of the National Academy committee on Everglades Restoration and is a member of NOAA’s Science Advisory Board.

Dr. Reed began with giving her background.  She has worked on tidal marshes for most of her career, working in the marshes of southeast Essex near London, in southern Chile, and on the Louisiana coast, where she has been working since 1986.  At the beginning of her time in Louisiana, she spent a lot of time looking at sedimentation in coastal marshes and trying to determine if they would persist despite sea level rise and subsidence.  In the 1990s, she spent a lot of time on the Columbia River estuary as part of a land-marsh ecosystem research program funded by the National Science Foundation. 

In the late 1990s into the 2000s, Dr. Reed worked in the marshes in the Bay Delta as part of the BREACH program.  “Originally we looked at restored tidal marshes where tidal reintroduction had largely accidentally occurred on some flooded islands and thinking about what the state of these flooded islands was going to be,” she said.  “I was looking at the geomorphology aspects of that, and then in Breach 3, we got the opportunity to focus on Liberty Island up in the north Delta; it was more supervisory rather hands on, but it was a great project and a great opportunity to learn more about the system.”

In recent years, Dr. Reed’s work has focused more on the connection between science and decision making.  “Hurricane Katrina was a real wake up call,” she said.  “It was really a call to action that we really needed to be much more proactive and work out how that science and understanding that we had been developing throughout our careers could then be used for the benefit of thinking about the future of the system.”

The talk today is titled, ‘Predict, Detect, and Understand,’ and she is framing it as such as it’s a way to break down the science into pieces that can then be related to decision making in different ways.   

These are the kinds of things that we do,” she said.  “They are integral to this adaptive management process – I’m going to say they are necessary but not sufficient.  Often we think about these in isolation from one another.  ‘I’m a modeler’, ‘I’m a field scientist’, that kind of thing, and the idea that we really need all those together and we have a responsibility to make those joins, even if our own individual specialisms or interests to lead us to one or the other.”


She then presented a map of coastal Louisiana, noting the mouth of the Mississippi  River is on the right.  The coastline is about 500 kilometers (about 310 miles), and is composed of soft sediments, and in many areas, hundreds of feet of Holocene material. 

The system has been subject to massive coastal wetlands loss during the 20th century; the graph on the lower right of the slide is from a USGS study published in 2017 showing land loss from 1930 to 2010.  

In the 1930s, we had say 19.5 square kilometers and then by 2010, we were down to about 14.5 thousand, so massive loss of land,” said Dr. Reed.  “So what does that mean?  It means wetlands converting to open water.  This is not converting wetlands to golf courses and WalMarts.  It’s conversion of a system that’s dominated by emergent vegetation to one that’s dominated by open water.  In the top diagram, everything that’s a color on there, particularly red, would have been land back in the 1930s which is now open water.”

There was a very steep decline in the latter part of the 20th century, and when Dr. Reed came to Louisiana, people were beginning to recognize the problem and were thinking about what could be done about it.  So there were a number of coastal planning efforts undertaken by various state and federal agencies and they all had different perspectives.

During the 1990s and into the early 2000s, what we did was plan,” said Dr. Reed.  “We tried to think strategically.  We thought about big ideas, we never actually had any money to put them forward and to make them a reality.  We were stymied on the federal side by just the lack of recognition of the importance of the problem, or at least that’s what it felt like then in coastal Louisiana.”

Multiple state agencies became involved, with one agency working on restoration activities, and another agency or department worked on coastal flood risk and storm surges, and so on.  Then after Hurricane Katrina hit in 2005, a lot of things changed.  There was legislation that required agencies to work together and also mandated a comprehensive master plan for the coast to be developed.

The first plan was pulled together in 2007 in just eighteen months using existing information.  The plan is to be periodically updated, so it was updated in 2012 and 2017.  The next update is due in 2023.


They developed an analytical approach for the 2012 plan, refined it in 2017, and are continuing to work on it. 

The approach is centered around two decision drivers: risk reduction and restoration.  The metric for risk reduction is expected annual damages from hurricane storm surges and waves, and for restoration side, the major metric is wetland area.  They had about $25 billion for each of those, so the question was, how best to spend a certain amount of funding on risk reduction or restoration?  How should resources be allocated towards the specific goals?

There were a lot of ideas about what should be done about coastal restoration in Louisiana,” said Dr. Reed.  “There are 2 or 3 million people on the coast, they are all experts, and they’ve all got ideas, and many of them are very, very good ideas.  But those ideas have to be sifted through.”

So we have coastal projects and ideas; they are then analyzed with a set of predictive models and then the outputs of the predictive models go into this planning tool, a kind of optimization algorithm,” said Dr. Reed.  “We then go back and forth, say okay, this looks like a good set of projects for $25 billion, let’s look at them together, see how they work, and we go back and forth between the middle until CPRA, or the Coastal Protection and Restoration Authority, that is the agency in charge, determines that these projects will be a part of the plan.”

The modeling was a huge effort with very large numbers of people involved, from all kinds of different facets of the science enterprise, such as federal agencies, state agencies, NGOs, and consultants. 

This effort we put forth would not have been possible without substantial investment by the state in this science and the development of these tools,” she said.  “This was not a cheap operation.”

The model has tools to look at storm surges and risk assessment, but the hardest part was the integrated compartment model. 

Basically it takes this system with the Gulf of Mexico, with the barrier islands, with bays, and the wetlands of all different types, and thinks about the dynamics, the sediment movement, how salinity changes, how vegetation responds or how vegetated wetlands turn into the open water, and plays that out over time and space over a series of decades,” she said.

It also allows us to produce a number of other outputs which are used in the planning process, so we can know the decision driver for the restoration side is land, what’s the area that we have at the end of the day, we can actually talk about habitat for shrimp, we can actually talk about what happens to oil and gas related communities across the coast, and a number of other metrics that we calculate in order to say, if we do this set of projects, what kind of changes do we think are going to occur across the coast.”

The analysis is done across a number of environmental scenarios.  The graph is from the 2017 coastal master plan; the low scenario is shown in blue, medium in yellow, and the high scenario in red.  Dr. Reed said that they are largely driven by sea level rise rates; there are other aspects such as precipitation, evapotranspiration, and subsidence, but for the ease of discussion, the easiest way to think about it is in terms of sea level rise.

If we don’t do any projects, what happens 50 years into the future?,” she said.  “The bottom axis is time …  and when we add on sea level rise during the next 50 years or so, we see a continued massive decline if high rates of sea level rise come to pass.”

She presented a map (lower, left) of those results, which shows how much land would be lost to open water during the 50 years of simulation. “This is a massive potential loss of land,” said Dr. Reed.  “This is truly an existential threat to coastal Louisiana.”

She presented a graph (upper, right) that shows the model results for individual projects which allows them to stack up and rank each project’s effects.  The short dark colored bar is the benefit associated with the project after 20 years; the larger bar is the benefit of the project after 50 years, and the vertical bar is the mean of the two, she explained.  The people in coastal Louisiana recognize the problem and want to see something happen, but Dr. Reed said they also need to ensure the projects that are done will actually last and generate benefits in the long term. She pointed out that several marsh creation projects show a benefit after 20 years but no benefit at 50 years. 

At the end of the day, what we have is a selection of projects and this is the mix of different project types that is included in the 2017 coastal master plan,” she said.  “We also have as part of the 2017 Coastal Master Plan, a fairly structured process for updating the plan and tracking things in an adaptive management way.  This actually tells us what kinds of things we should be doing every year during this cycle of updates that we go through regularly.  We are currently in the process of developing or improving the systems models in order that we can use them and improve them for the next time around.”


Dr. Reed then turned to what was learned from the analysis.  She presented a map of the results of spending $25 billion on restoration, noting that all of the areas in red are still going to be lost to open water and the areas in green are the places that would have been lost if they didn’t implement the plan. 

So we’re like, can we do better than that?” she said.  “It really still doesn’t look very good and this is the medium scenario – you really don’t want to see the high.”

So in terms of what they learned, Dr. Reed explained that the integrated compartment model suggested wetlands could convert to open water under certain kinds of circumstances in the analysis; this happens when salt water comes in with a storm, so fresh forested wetlands and freshwater marshes are subject to salinity stress.  Other kinds of marshes, such as intermediate brackish and salt, are subject to different levels of inundation stress when they get flooded too frequently.  And marshes can erode at the edges from waves hitting the edge of the marsh over time.

She presented a map (lower, left), noting that the different types of marshes are distributed across the coast; green are the fresh marshes, salt marshes are red, and intermediate (just slightly salty with a different kind of vegetation) is shown in yellow.

There are the different mechanisms that cause land loss, and so by asking the model what is causing the land loss, they could glean some insights.  The numbers in the table on the slide (upper, right) are the land loss over time; this correlates to what was on the curve several slides ago.  The pie charts show the different causes of loss and how they contribute to land loss.  Yellow is marsh edge erosion, such as waves hitting the edge of the marsh day after day after day and during fronts and storms, causing expansion of open water bodies; the brown colors show the loss from flooding and inundation; and red is the salinity stress. 

Dr. Reed noted that the colors on the chart are the same as the previous.  “One thing you see is that land loss ramps up over time, but also there’s a difference in the contribution of the different factors over time,” she said.  “Marsh edge erosion is more important in the early years and it’s more important in the low scenarios.  When we get into the later years, it’s this inundation effect that really starts to dominate.”

She also noted that the red is loss of fresh wetlands due to salinity penetration, but it’s periodic and not consistent across scenarios.

This is all about diving in and trying to understand what the modeling is telling us,” Dr. Reed said.  “So we were looking at part of the southwestern part of the coast; the blue star is a particular basin called the Mermantou basin, and the blue arrows on the bar chart show these particular years show salinity loss in fresh wetlands.  This is occurring, not because of sea level rise or storms, but because of a set of interacting climatic conditions.  When we have conditions in the 50 year modeling sequence of events that gives us a really dry summer combined with low river inflows, then we get really highly salinities.  When those conditions converge, there can be really bad and really cause an increase in salinity.”

She presented a graph (upper, right) showing the effect over 50 years.  Summer precipitation is on the top chart, the middle is average flow anomaly, and the bottom chart is the maximum salinities. 

When we have negative surface water balance in the summer and low stream inflow, then we get high salinity,” she said.  “It makes sense, but this is not an effect that we thought was necessarily important.  I spent most of my career thinking, we need sediment, it’s one of the most limiting factors for the Louisiana coastal wetlands and it’s quite possible in the future that freshwater could be equally limiting.”

So the idea here is that doing this kind of diagnostic analysis really makes you think about what’s going on and it allows you to think forward,” she said.  “We’ve got to start solving the problems of tomorrow, not the problems of today.  The public wants things that address the problems they see out of the window, but what we see in this analysis is that the problems of today, we can solve those, we can put rocks all around the edges of the wetlands, but they are still going to go underwater in 30 or 40 years, so it’s not a long-term solution.”

So we use this to try to come up with some better projects, so for the 2023 plan so we can actually have a map that doesn’t look so red as a result of $25 billion of investment.


Dr. Reed recently helped put together a science plan for Delta smelt that targeted the effects of ambient environmental conditions and flow related management actions on Delta smelt.  She noted it was a fairly limited set of potential actions that could be done to benefit Delta smelt, and she was asked to put together some thoughts about how science could move forward that understanding.

I’m not going to talk much today about the science I recommended, but what I saw there in this circumstance was that I was brought in to try to help with was some kind of problems that are potentially more systemic than just this particular issue within the Delta,” she said.  “The idea was to try to be responsive to management needs, to be efficient and effective with limited resources, but also to generate timely usable information.  With Delta smelt, with the annual cycle for the species, timeliness is everything.  We can’t miss too many years.”

What was important was the idea of a structured process – that if we really want science to make a contribution to these management decisions, we have to be deliberate about that and we have to really think and plan that and not just be reacting all the time, she said.  So at the heart of the science plan was the idea to a three year science plan around flow-related management actions and changing environmental conditions in Delta smelt. 

Now, you never know how much water you’ve got for a flow-related action until the year that you’ve got the water, so planning three years in advance sounds a little bit futile, but my point in this plan was unless we really think about this in advance, we are always going to be reacting to these things and we’re never going to be ready,” said Dr. Reed.  “And we know what the array of potential management actions is, and we know what the array of potential scientific activities are, so let’s get our act together and think about what we might need to do, and then we can make adjustments as we go through.”

It’s not just monitoring conditions, but it’s also important to understand what kind of studies are needed and it’s also about estimating resource availability.  “We need to recognize that resources will be constrained,” she said.  “We know that there’s never enough money, but we have to do that and think about what science we can accomplish with the money that we have.  So the idea is having a process where you could go through that kind of filter of what’s a realistic level of resources over a three year cycle and then every year when you knew what management actions were going to be undertaken, you could come back and put the flesh on the bones for this particular year that is structured around predicting, detecting, and understanding and really thinking through the opportunities that you have with any year to advance the science to understand these management issues.”


Closing the Adaptive Management loop.  Dr. Reed noted this is something that has been talked about for 20 years, and it’s still difficult to do.  “I really think that we have to get serious about how we as scientists are going to contribute to that closure and how we’re going to work with decision makers in order to make that closure possible,” she said.  “Because unless we actually get to this piece where somebody’s using that information to make a decision, we haven’t actually made that work.”

So how do we make it work?  “I think it’s more than just communication, because communicate current understanding, adapt – like it just happens, but it doesn’t work like that,” said Dr. Reed.  “I think it really takes understanding the actual decisions.  Who is going to use this information and for what.  What’s the timeline for making that decision and what kind of information is going to be usable to them.  We’ve got all these other diagrams about how to do this, but we don’t really have the mechanics of this worked out. …  I really like the idea of interchange and dialog and back and forth.  It’s not just a one-way flow of information to the top.  We have to make that work.

Second is to encourage new technologies.  “We are already advancing and trying to use new things, and I want to encourage that,” she said.  “People are doing all kinds of things with drones these days, just amazing things.  There is the idea that you put an acoustic tag in a fish and when it gets eaten, the stomach acids converts the signal that the tag is giving off and I think that’s really cool!  I want to make sure that they’re not just research tools, but that they go into operations.  That it isn’t just something that we thought of working in a lab, but that we’re actually trying to use it to get some information for decisions.”

Increase integration in a lot of ways: Integration across disciplines, across issues, across the system – how these connections work, how the feedbacks occur, and what the spatial relationships are really important, she said.  There’s also integration across entities and institutions and multiple dimensions for this kind of integration, so it’s something we need to make sure we are being efficient about and leveraging all the resources we can, she said.

Part of the integration is the report from the Social Science Task Force which talks about the need to better integrate social science.  Dr. Reed said to her, it’s rather obvious, and she gave an example from Louisiana as it’s really a very central part of how they think in Louisana.

The Coastal Master Plan has five objectives, two of which are about land loss in these systems, one of which is about storm surge, another one is about working coasts, and another one is about culture and heritage.  When they are developing the plan, they are not just thinking about land and flood risks, but also how that interacts with all these activities on the coast. 

The energy industry is really important, and Louisiana is a great hub for trade,” Dr. Reed said.  “These are really important opportunities for our coastal population and also for how we relate to the rest of the country.  Culture is really important down there, too.  If you’ve been down to south Louisiana, my guess is that you’ve experienced that in some way.  Each one of those in and of itself is not de-faceted.  It’s not just people and demographics; it’s about how people are relating to their environment.”

Family connections are really important in Louisiana,” she continued.  “I’m sure the kind of social dynamics in the Delta are different … but I’m sure there’s a similar set of dynamics associated with how people associate with the landscape and the system here that similarly needs to be appreciated.  We could not do anything for coastal Louisiana without really thinking about how it affects the people down in the bayou.”

In Louisiana, they have come up with some ways to think about that.  She presented a slide from a monitoring plan, system-wide assessment of a monitoring plan, pointing out that it’s not just about the natural system, but it’s thinking about data sources and information about the economy, employment, housing, population, demographics and how that ecosystem dependence between what the people are doing and what the ecosystem is doing and how those link together, and what datasets can be brought together to inform that.  Dr. Reed acknowledged it was really tough to integrate that analysis into the framework for modeling; they are trying to make baby steps on that for the 2023 Master Plan. 

Obviously we’re in a big low lying coastal system with sea level rise coming,” she said.  “Migration and population movement is really important.  Where are people going to go and where are they going to leave from and how do we get ahead of that game.”

The other thing that’s important is to look at it through the lens of equity.  “The people who can afford to leave are going to leave.  I’m lucky; I live down the bayou, I can afford to go if I needed to, but there are lots of people who could not afford to go.  We have to make sure that the actions that we take and that we bring this into our science and that we help the agency understand the implications of that.  We’re still working on what that might be and how to do that, but this is very much in conscious in terms of the conversation we’re having down there in Louisiana.”


QUESTION: Do you have any ideas about how we can be more successful at turning what we’re discovering and knowing and what we think and talk about into actionable and useful recommendations, changes, and practices?

DR. REED:I’ll tell you how I would start. I would really be out there talking to people individually about what it is they need.  I feel like we need to make that connection directly.  I feel like there are a lot of issues or needs or concerns that don’t always get expressed in the kind of interactive forums that we have with people sitting around the table, so I think the idea of really trying to understand what it is they need and what kind of things that they need.  It might not always be deliverable.”

The other thing is the timescale and the format and the nature of the information that is then usable to them and so that then is a challenge because you’ve got lots of different audiences.  But somebody is going to make a decision and really you want everybody else be comfortable with that decision and that there’s some kind of transparency to the scientific information that went into informing the decision.  Not necessarily driving it but informing it, so I would start by really talking to the people around the DPIIC* table, but not necessarily at DPIIC.”

*Delta Plan Interagency Implementation Committee, which is a group of federal, state, and local agencies with responsibilities in the Delta that meet on a semi-annual basis to discuss implementation of the Delta Plan.

QUESTION: A lot of what you’re talking about relies on integrated models.  I’m a strong believer in integrated models but the issue that I see is that it’s difficult to decide what model to apply, particularly to biological systems because the data that’s available can often be confusing and resistant to conventional modeling approaches.  How do you think you should go about deciding which models to incorporate and validating those for use in the kind of prediction scenario and evaluation scenario that you’re suggesting?

DR. REED:Validation is a real problem when the conditions are changing so dramatically. And so that is just really difficult to do. …  I think there are issues with statistically generated models.  I’m a process geomorphologist, I’m not going to apologize for that.  I want to understand the processes and if there’s a relationship and I can’t understand why there’s a relationship, no matter what the statistics around it are then, I’m not going to feel comfortable trying to encourage decision makers to make a huge decision on a basis of that, albeit it might be the best way for us to capture the data that we have.  I’d like to focus on process-based interpretations, quantitative-based interpretations.”

Some people just don’t understand why we need more than one model.  We have to do a better job of explaining what each one does and what we can use them for.  I think the idea of deciding how to put these models together depends a lot on who is going to use it for what, and what their tolerance for validation would be.  We also have to do a better job explaining to people that, if we have some kind of integrated model that has ecological processes and hydrological processes, we’re probably going to get to the stage a whole lot better than we’re going to get productivity.  At least help people understand that that is just the way it is.  We might get this like this and that like that but actually that’s pretty good and we’re feeling good about it.”

So I think that a lot of the communication there.  Model equals model equals model is not something that we want to promulgate.  We have to do a better job of kind of fleshing that out.

QUESTION:  You talk about the need for infusing science into the decision making process and I think everyone would agree that’s a goal and there’s a lot of different ways to do that as you’ve talked about.  How do you measure success or how do you measure failure? 

DR. REED:That’s a really good question … One thing I have reconciled myself to over the years is, you’ve got to do the science, and they are going to use it how they area going to use it. It’s not my job to second guess their decision whether they use it or not … there are some issues in Louisiana which are contentious about the role of the oil and gas industry in 20th century land loss. It’s not helpful to blame the oil and gas industry.  It doesn’t move the ball forward.  I want to focus on what we can do about the situation that we have, not who caused the problem back then, and so the idea of getting people pointed towards thinking about the future.  For me that’s kind of success.

The integrated compartment model … People come and they want to use that for other things.  It’s been useful for that, so can we adapt it … so it’s basically a framework for thinking about things as much as a specific set of equations and algorithms that produce.  It’s a success in that people feel like having a model or having some simulation results is going to be useful to them.  That I find very gratifying.   There are other times when people take the results and misinterpret them.  That feels like failure in that we haven’t communicated about what this should be used for.”

It’s something worth tracking; we can track use, we can track application, but higher use and higher application doesn’t always equal success.  I’m not sufficiently naïve to think that the more people that use it, the better it is.  But if it helps them, it probably is.”

Print Friendly, PDF & Email