I'm conducting a research on risk assessment of river basin pollution using GIS. I'm trying to find ways to incorporate socioeconomic variables in the risk assessment. Are there any socioeconomic index or calculation that I can use to predict the risk of pollution in a river basin? Lastly, what kind of data do I need to assess the impact of socioeconomic on river basin pollution?
The US Forest Service developed a Framework to assess watershed condition about 15 years ago. All the watersheds on the National Forests were assessed using this framework. It used parameters such as population, land use, etc. The various factors were assessed in GIS and evaluated by in interdisciplinary team on each forest. Land uses with high potential or risk were urban areas, numbers of point source discharges, agriculture especially annual cropping, barren areas, road density and similar land uses or activities. Forests and grasslands would be examples of land uses with low risks. In general, much of the analysis was developed by hydrologists and soil and water experts, and for most National Forests, Th assessment teams included this level of expertise to help evaluate, rate and compare watershed conditions. Each forest then identified one or more watersheds for more intensive planning and treatment directed to improve conditions. Much of the rating required some professional guidance in the ratings, such as for example, a watershed with less then 1 mile of road per square mile might have low risk for that factor. Numerous factors were considered and rated. This level of detail might be more than you were searching for.
Very good question!
I concur with Hansen's comments.
So, let me take a slightly more analytical approach. First, I assume that you are not limited to establishing the set of variables (e.g,, as you would using GIS) but to describe and predict outcomes -- hence the term risk. Ask implies a measure of uncertainty such as probabilities. The SES variables will include a subset of risk factors. In some situation you might reduce the number of subsets of risk factors using methods such PCA or factor analyses. In general, you need some function R=f(X1 ... Y1 ....) and assure that it is consistent with what you are representing. Theoretical considerations will determine the model form (how R relates to the inputs variables you use). Indeces can seldom be correct: most indices I know summarize some -- but not all - aspects of the combinations of variables used to construct the index. Your models should be statistical and thus include probabilistic uncertainty. See linear statistical models ( they include non-linear functions) on the net.
Regarding your last point, a reasonable solution would be to develop a system of equations (linear, to begin with) that relates input variables to out put variables. For instance, oxygen uptake is a function of certain river specific physical parameters, including pollution). At the same time, land values decline as river pollution increases. And, land values are a function of distance to the river, industrial development nearby, air pollution (it deposits particles), water pollution from line and point sources into the river. So, a plausible risk analysis is relatively demanding. The question is: What is that you really need and for what purpose?.
I hope this helps a little.
best,
paolo
Paolo F Ricci
The risk map that I'm going to make is consisting of three main components: Vulnerability, Hazard & Socioeconomic factor. The purpose of my study is to assess the susceptibility/risk of river to pollution(in watershed scale). So the source of pollution can be point source and non-point source. The reason I'm adding a socioeconomic factor is to see the impact of socioeconomic on river, especially how it contributes to water pollution. Or I use the socioeconomic factor to assess how vulnerable the population in the watershed when water pollution occurs. Recently I found Socioeconomic Vulnerability Index (SoVi) but I have no idea on how to incorporate it in GIS and my risk mapping process.
So far the socioeconomic variables that I'm planning to use is:
1. Population
2. Land use change
3. Animal husbandry output
4. Industry output
5. Gross domestic product
I would tend to take a broad economic view, seeking to monetize the effects, which would admittedly require care (possibly involving polling) in treating the more subjective values.
Article Influences of topographic and socio-economic attributes on t...
This article may help you.
Hi All,
These conversations/issues/points are quite valuable. I would think that rent theory from economics, land value theory and evaluation of intangibles are critical to any form analysis. A first issue is that methods to monetize intagibles are approximate and thus go well beyond probabilistic uncertainty. Yet, they seem to be the best under the circumastances you might have. Take a look at the work done by U.S agencies iDept. of Agriculture, US Army Corps of Engineers and the relevant literature in land economics). A number of methods require surveys ... again a well-treaded road but expensive and time consuming. Moreover, it requires expert help to develop questionnaire, their statistical analyses, interpretations and so on.
Regarding risk maps: you are the one who should set its granularity and purpose. Vulnerability needs a formal definition (e.g., probability of occurring in 100 years) unless you can develop an appropriate index relating probability of occurrence to economic loos or something like that. If you multiply (pr * $) you get an expected value (e.g., the arithmetic mean the pr = 1/n, over n objects that are homogeneous). Not really meaningful in most cases you may face.
Any index can mask effects The mathematical form of the index is critical in this context.
An aggregation through an index has to be take with several grains of salt. Clearly, it should not be simple averaging because some averages are compensatory. Any measure of central tendency (mean, mode etc.) should include the variability about each. Actually, you might prefer to call averaging the expected value of the random variable(s) you deal with).
All of this points to a number of technical issues that, f not considered, make results questionable.
All the best,
Paolo
- Badges
- Science topic
- Similar topics
- Engineering
- Petroleum Engineering
- Gas