Scenario+Ideas

George Hess //2010 Nov 5//
 * Triangle Ecosystem Services – Project Questions and Products Outline**


 * Study area:** Upper Neuse River Basin, approx 2,000 sq-km

(1) Biodiversity vs ES questions. [ A set of questions about the concordance of high levels of ecosystem services and areas of conservation concern. ]

A. How well do the areas that provide high levels of ecosystem services overlap areas of known biodiversity and ecological value? [ How well does conserving areas of high ES also conserve biodiversity? ]
 * Use 2001 data only.
 * Stack the normalized, biophysical data for carbon, nutrient retention, pollination
 * Use same type of diversity data as Hess et al 2006 – significant natural heritage areas and element occurrences. This allows comparison with other approaches in that article as well. [ Significant Natural Heritage Areas are defined by the Natural Heritage Program to be areas that contain species and ecosystems of conservation concern. Element Occurrences show the location of individual species and ecosystems of conservation concern. ]
 * Calculate representation (% element occurrences covered), completeness (% element occurrences covered), and overlap (% significant natural heritage areas) covered by areas of high (top quartile), top half (top two quartiles), and top 3/4 (top three quartiles) of ecosystem services areas.
 * Output: table showing results, like T3 in Hess etal 2001.


 * We are also trying to use NC GAP data on individual species to get species richness for areas of high ES - the procedure to do this is complex and still being developed.


 * Note: I think to answer this question effectively we need to first make a map of the areas we would suggest saving because they are of high ecosystem service value and then see if there is overlap with the biodiversity sites. So if we say that the areas of highest concern are the groups of pixels that have the highest quartile of stacked services - then look at completeness and overlap, etc. I think this is what George is saying, but we have to find a way to represent the areas we would suggest provide the most value first. Also, we need to consider if the areas that are of high ecosystem service value will still be if the area around them is developed. Perhaps we suggest protecting a buffer of area around the groups of pixels that have the highest ecosystem service values. <>

B. What is the level of ecosystem services in areas of known biodiversity and ecological value compared to other areas? [ If you conserve biodiversity, what's the level of ES in those areas? ]
 * Same base data as A
 * Clip stacked ES data with SNHA boundaries
 * This time, create a histogram of the quartiles of ES found in the significant natural heritage areas. X-axis is quartile (0, 1-25, 26-50, 51-75, 76-100) and y-axis is area (sq-km) - or do a line graph as suggested in the sketches below.

C. What proportion of the total area of high levels of ecosystem service (top quartile) in the Upper Neuse Watershed are overlapped by areas of known biodiversity? Same for top half, top 3/4. [ If you conserve diversity, are you also conserving areas of high ES? ]
 * Same base data as B.
 * Used clipped and total values to develop proportions.

(2) Payment for Ecosystem Services question

How does the dollar value of the stacked ecosystem services – carbon, nutrient retention – compare to property values in the study region? The notion is that if property values are much higher than ES values, it’s hard for ES to compete.
 * Use 2001 data only.
 * Create GIS layer of stacked value of ecosystem services in dollars.
 * Tax parcel data for portions of 6 counties – Durham, Franklin, Granville, Orange, Person, Wake – all available from Triangle Land Conservancy for the study area!! It’s the most current reevaluation: Durham (2008), Franklin (2004), Granville (2010 or 2004), Orange (2009), Person (2005), Wake (2008).
 * Create GIS layer of tax value. We’ll need to rasterize the tax parcel data somehow so that we can compare to ecosystem services – need to convert tax data to $/pixel. Value and area are in data, so this can be done – do calculation to get $/900sq-m.
 * We need to mask out the developed areas somehow and focus on the areas on their fringe (highest chance of conversion) and radiating our from there. Details unclear.
 * Divide: tax/ES so that values >1 are where property >ES and by what factor; values <1 ES > tax. This avoids us needing to decide how many years of ES payments to compare to property values – we just show how much bigger / smaller they are.
 * Some kind of range analysis - high, med, low $$ values - does it matter?

Here are some graph sketches (added 28 Oct): [|GraphSketches.pdf] (Not updated to reflect changes in questions) Rough tax parcel map - has data issues that need to be worked through - [|TaxParcelMapRough.png]


 * OLDER DISCUSSION FROM 27 Oct DRAFT.**

(1) How has / will the level of ecosystem services in the study area changed … 1992, 2001, 2010 (projection), 2030 (projection)? We will evaluate biophysical values of carbon, nutrient retention, and pollination. (I think we could completely remove this question or if we keep it, we should restructure the subsequent questions. In my mind, proceeding with temporal changes in ES is not interesting in itself. Instead it would need to be casted in a broader framework of conservation planning through future land acquisition, land configuration, etc. I think there is a cohesive and exciting story if we proceed with questions 2-4, see below. - Nyeema) <> <> <> <> [I think this option is the most viable and most readily achieved. It has the potential to show the negative impact of human populations over time and the need for a different approach. Implications are what is most important to me personally, but I am more than willing to go along with whatever.-Shawn] <> [Nelson et al. were able to show that conservation led to gains across the board. If we examined what types of relationships are expected in the Triangle, compared it to high/low scenarios (i.e. extreme conservation vs. extreme development), we would be able to show if services in NC differ from Oregon. Additonally, we would be assessing if there is a clear link between the biophysical provision of ES and their ultimate use by people. This is an issue that Nelson et al. suggests for further examination. Thus, we would already have probable cause. The latter (4th question) only looks at the final output and is less likely to contribute to the documented literature. -Shawn] < Recent studies that incorporate the spatial distributions of biological benefits and economic costs in conservation planning have shown that limited budgets can achieve substantially larger biological gains than when planning ignores costs. -Neal>> <>
 * Four Questions:**


 * Output: graphs of total carbon stored through time. <> <>, total nutrients retained, pollination ( kevin/alicia- agree with meg: we can't graph each pixel, we have to look at the map.) <> <<would it be possible to discuss this in class because I don't think I understand all of what GIS can do. where would these totals come from. does GIS sum up everything giving us totals to compare through the years?-alicia>>
 * Output: maps of each through time - a panel with the services going across the top and the years down the side – 4 maps for each service, simplified to some quartile or quintile representation of the service levels. Maps will need to be small and simple <<so all the maps fit into one diagram? meg>> <<Yup - george>>.
 * Get models running for 2001 (all) – reclassify (using GIS reclass tool) the biophysical output maps into quartiles (0, 1-25, 26-50, 51-75, 75-100) – KEVIN, can you make a grey-scale color scheme that will print well for that? Darker is higher <<why do we need to reclass? or is this for Kevin's ears only?>>. <<To create the final graph (area with different levels of service) we need to create breakpoints somewhere - hence the reclass. Also, to do the comparison against biodiversity stuff I think we'll want to pick the top service areas. Perhaps I'm not thinking of this correctly. Maybe if I sketch the graphs out on paper things will clarify. george>>
 * Data for 1992 already available – once 2001 is done, do 1992 same way. This depends on getting the best parameters we can for the models.
 * HESS is getting projected data for 2010, 2020, and 2030 from TLC – will make available this week or early next. KEVIN will likely need to help combine with 2001 data.
 * For each service, collect the number of pixels in the 75-100 quartile (just open the attribute table) for each year and convert to area (each pixel is 900sq-m – convert total to sq-km) and put in a spreadsheet for plotting. Need someone to design a standard graph. X-axis is year (scaled appropriately) and Y-axis is area (sq-km).

We could frame the paper to explore how stacked ecosystem services compare to biodiversity and land value. In doing so, we would determine whether conserving areas of high ecological value (based on rare habitats, community composition, and species diversity) inherently protects ecosystem services in a developing area. Additionally, we would explore whether the economic return/value of ES can outcompete land values, providing just cause to deter future development scenarios. I think this is an attainable project that makes a novel contribution. - Nyeema ﻿ ﻿ <<This is an interesting point. I think this is what I was expecting when we left class on Monday - Chris >> ﻿ ﻿<<So, we would need baseline data for conservation. That is numbers, in terms of dollars, it will take to conserve these priority areas through, I assume, outright purchase. This is OK, but what about working lands? I've been looking into these conservation 'banking' organizations like Chesapeake EcoFinance Co. and Conservation Marketplace of Minnesota that organize payments for ecosystem services. Seems like the promise of ES is to make payments for conservation. Payments to farmers and other landowners to provide conservation might be cheaper than outright land purchase - as it would be a payment for income lost, not necessarily for purchased land. Perhaps we could compare baseline tax dollars (assuming purchase land) and conservation dollars (i.e. NRCS, Soil and Water, other Farm Bill dollars) and ask whether either of these approaches can keep up with the ES values we identify. This seems really similar to question 4, just without the time change. I can't see how not including time change would be useful to us. The most important thing is not the final number, in terms of value$, but change in value, because we know our final values will be wrong and full with assumptions. What's important is change. And, if we are thinking sensitivity analysis, then let's do it to land values and show how high land values discourage conservation (and at what point), or perhaps they don't. This question seems to ask a lot, which is part of the problem, and opportunity. -NEal>> <<IF we could get some figures for annual income per unit land area for typical crops here (corn, soy, tobacco, etc) we could take a shot at this for ag land, but it does require more assumptions about what's planted where, etc - we'd have to frame very carefully - perhaps comparing potential ES payments against the highest possible income for crop here - george>>

(2) How well do the areas that provide high levels of ecosystem services match areas of known biodiversity and ecological value? <span style="color: #800000; font-family: 'Lucida Sans Unicode','Lucida Grande',sans-serif;">[I am neutral on this topic. -Shawn] <span style="color: #ff7000; font-family: 'Lucida Sans Unicode','Lucida Grande',sans-serif;"><<I'm beginning to really like this question but I'm not sure how accurate our analysis will be (it will ultimately depend upon the data you guys find) - Chris>>
 * Use 2001 data only.
 * Use same type of data from Hess et al 2006 – significant natural heritage areas and element occurrences. This allows comparison with other approaches in that article as well. (HESS has these data from TLC for study area. NYEEMA is also looking at other data from Wildlife Resources Green Growth Toolbox.)
 * We can also make comparisons of stacked ES to a metric of species richness (to actually have a biodiversity component). I will try to obtain this data from NC GAP that includes 414 vertebrate species. -Nyeema <<That would be an interesting addition - george>> ﻿ <<I like this point as well - Chris>>
 * Calculate representation (% species and ecosystems covered), completeness (% element occurrences covered), and overlap (% significant natural heritage areas) covered by areas of high (top quartile) and top half (top two quartiles) of ecosystem services areas
 * Output: table showing results, like T3 in Hess etal 2001.

(3) What is the level of ecosystem services in areas of known biodiversity and ecological value compared to other areas? <span style="color: #800000; font-family: 'Lucida Sans Unicode','Lucida Grande',sans-serif;">[I am neutral on this topic. -Shawn]
 * Same data as #2.
 * This time, create a histogram of the quartiles of ES found in the significant natural heritage areas. X-axis is quartile (0, 1-25, 26-50, 51-75, 76-100) and y-axis is area (sq-km) <<Actually, I can create this on a line - see attached sketches - george>>

(4) How does the dollar value of the stacked ecosystem services – carbon, nutrient retention, pollination (kevin/alicia-- we only have a relative value from 0-1; perhaps gather replacement costs of managed pollinators??) – compare to property values in the study region? The notion is that if property values are much higher than ES values, it’s hard for ES to compete <<or, we know how much additional money is needed, i.e. grant, etc. -Neal>>. <span style="color: #800000; font-family: 'Lucida Sans Unicode','Lucida Grande',sans-serif;">[I do not think the outputs from the models should be manipulated to show this information. We know the models are dependent on hypothesized inputs and I do not think the data is defensible -Shawn] <span style="color: #0000ff; font-family: 'Lucida Sans Unicode','Lucida Grande',sans-serif;"><<What makes this more hypothetical and less defensible than the first question? It's all approximation based on best-available information, no? How defensible are growth projections needed for the first question? Also, this is why you might do a sensitivity analysis (as Nyeema noted) - if the conclusion doesn't change much with large changes in the parameters, then it doesn't matter if the parameters are a little off. george>> <span style="color: #800000; font-family: 'Lucida Sans Unicode','Lucida Grande',sans-serif;">.[Again, I will do whatever people want to do. Both questions have a large number of assumptions. When I read the first question, I see that we are already examining the function of each ecosystem to provide a certain service. I can understand this type of relationship more readily than comparing prices of ecosystem service value to land value. This question matters more to an economist and I am definitely not that. I am more concerned with the nuts and bolts of the process. The amount of work I have to do on either question is about the same so I do not care which question we answer. Although, the latter question has two sections. Can we finish that in time for a presentation? I am all about evaluating how much time we have and what can be realistically achieved.] <span style="color: #ff7000; font-family: 'Lucida Sans Unicode','Lucida Grande',sans-serif;"><<I think this question may be our most important contribution for Taylor and WWF (based upon what results we get). Does this make the ES concept unimportant in urban areas of developed countries? How high would carbon prices have to be to allow ES to compete? I disagree with Shawn's take on this question. Some of the inputs are hypothesized, yes, but we're making approximations on a large scale. As long as we stay on that spatial scale, I think we'll continue to be fine. Also, I think this is my favorite question, so I'm a bit impartial to it - Chris>>


 * Use 2001 data only.
 * Create GIS layer of stacked value of ecosystem services in dollars.
 * Tax parcel data for portions of 6 counties – Durham, Franklin, Granville, Orange, Person, Wake – all available from Triangle Land Conservancy for the study area!! It’s the most current reevaluation: Durham (2008), Franklin (2004), Granville (2010 or 2004), Orange (2009), Person (2005), Wake (2008).
 * Create GIS layer of tax value. We’ll need to rasterize the tax parcel data somehow so that we can compare to ecosystem services – need to convert tax data to $/pixel. Value and area are in data, so this can be done – do calculation to get $/900sq-m.
 * Divide: tax/ES so that values >1 are where property >ES and by what factor; values <1 ES > tax. This avoids us needing to decide how many years of ES payments to compare to property values – we just show how much bigger / smaller they are.
 * Is there a sensitivity analysis in here for the monetary values selected for each ecosystem service? -Nyeema ﻿<<seems like for this one we would need to do some kind of high/ low because (at least for pollination) we are just calculating replacements costs and farmers vary in approaches to managed bee species- alicia>>

Here are some graph sketches (added 28 Oct): [|GraphSketches.pdf] Rough tax parcel map - has data issues that need to be worked through - [|TaxParcelMapRough.png]

Hess, G.R., F.H. Koch, M.J. Rubino, K.A. Eschelbach, C.A. Drew, and J.M. Favreau. 2006. Compraing potential effectiveness of conservation planning approaches in central North Carolina, USA. //Biological Conservation// 128(3): 358-368.

OLDER STUFF BELOW THIS LINE

10/18/2010

Revisited

IF you can conserve different areas, where should you conserve? -->Where would you get the highest return?

Land use/Land cover changes over time -->Would different conservation scenarios conserve different levels of carbon? -->How simplified is the model spatially?-->Would require compact growth and sprawled growth scenarios.

Determine how the model is running? -->Are the numbers we creating real? Do we know what affects these numbers? ---> User error? Complexity of model?

Water Model -->Identfiy where the data layers are "shaky"

-->Analyze data model and see how it works: 1) Work on the sample data and see if any new values are produced in the water yield model.(Change parameters and see what outputs changes), -->George, Neal, Meg -->Melissa: Review literature and think about realities that we want to ask for this model

2) Work on upper neuse data and see if there are errors Shawn, Kevin

10/11/2010 Research Scenarios 1) How much LULC change makes difference in model? Will this demonstrate use of model scale?

2) How much $ TLC put into conservation? What is the return on their investment? Additional factors/justifications?

3) Water sensitivity analysis

4) Carbon sensitivity analysis

5) Bigger area? over time?

6) How does urban sprawl change it?

7) How much conservation is needed to make change?

8) Test model with urbanization best case/ worst case scenarios (bracketing) -->Would account for drought and hurricane in water -->Would account for LULC change for carbon, biodiversity, and pollination -->Would test the model in relation to LULC

10) Range analysis or sensitivity analysis for biodiversity. WHat is the value of each habitat or species? --> What is the level of threat to that species or habitat? --> How far the threat penetrates the habitat? -->George can see a graph where he sees the highest quality numbers on the y-axis and the habitat values on the x-axis. Is there somewhere in the curve with a dramatic change?