perspectives in handling uncertainty in european data
DESCRIPTION
Perspectives in handling uncertainty in European data. Philippe Crouzet European Environment Agency (EEA). Contents. Importance of uncertainty Uncertainty in the assessment process Uncertainty in the Policy questions proper Two open questions, among others Gross nutrient balances - PowerPoint PPT PresentationTRANSCRIPT
1 European Environment AgencyHarmoniRIB 21/09/2006
Perspectives in handling uncertainty in European data
Philippe CrouzetEuropean Environment Agency
(EEA)
2 European Environment AgencyHarmoniRIB 21/09/2006
Contents
• Importance of uncertainty• Uncertainty in the assessment process• Uncertainty in the Policy questions proper
• Two open questions, among others• Gross nutrient balances• Fragmentation of rivers
• Example Response in the EEA work• Catchment – water relationships
3 European Environment AgencyHarmoniRIB 21/09/2006
Uncertainty in the data process
data Monitoring
Processing Assessing
data
Other factors
Decision
Other factors
MeasuresImplementing
4 European Environment AgencyHarmoniRIB 21/09/2006
River fragmentation• Dams fragment rivers:
• Large volume and residence time settle sediments,• Large volume modifies water regime• Wall jeopardizes fauna migration
• 6700 large dams in Europe represent ~90% of stored volume (~1450 km3) and less than (??%) of walls
• Fragmentation vs. fish is more uncertain than sediment trapping and water regime changes, but:• Actual reservoir volume? Current discharge pattern?• True commissioning year• Has dam a lake?• Are all dams properly located
• Data base accuracy is key issue
5 European Environment AgencyHarmoniRIB 21/09/2006
6 European Environment AgencyHarmoniRIB 21/09/2006
Gross nutrient balances
• Conceptual• GNB = Input {N} –output {N}, can be >0 or
<0• Negative values do not compensate positive
values• Calculation scale and conceptual model
issues• Large calculation areas are smoothed:
problematic areas are faded, thus distorting results
7 European Environment AgencyHarmoniRIB 21/09/2006
GNB Sensitivity
Ref (NUTS4 + CLC)
NUTS3 + CLC
NUTS3 alone
Comparing catchments (298 units) surpluses by regression yields :S(dpt + CLC) = 0.94 S(ref) (R2 = 0.86) S(dpt) = 0.86 S(ref) (R2 = 0.82)Total surplus is comparable in the first case.
8 European Environment AgencyHarmoniRIB 21/09/2006
Implicit uncertainty in policies and assessment
• EEA provides support to policy effectiveness assessment:• Has a given policy delivered expected results on the
environment?• Were results observed at enforced time target?
• Difficult questions are raised:• Is result defined? How expressing “meeting of the target”
in accurate terms? • Can time the target is met be defined in variable conditions
within compliance constraints?• Any response being “bound uncertainty” “unbound
uncertainty” • Making reliable assessment difficult and improvable
9 European Environment AgencyHarmoniRIB 21/09/2006
Analysis of the WFD specifications• Demands programme of measures:
• to “prevent deterioration of the status” and achieve “good status... at the latest 15 years”, and “bring the bodies of water progressively to the required status”,
• Following “coherent and comprehensive overview of water status”, and survey it
• Through identification of “significant | main” pressures and
• Assessment of status at the “water body type” level, minimizing the impact of seasonal variability ..... Reflect “...changes in the WB as a result of changes...” {pressures}
• All terms pose practical and theoretical questions about their accuracy and uncertainty
10 European Environment AgencyHarmoniRIB 21/09/2006
Analysing trends• Stratified statistics were carried out
considering:• Catchments as statistical population, (comprehensive)• Pressures through aggregates of land cover, population
and livestock, defining strata in the catchments population, (significant pressures, WB types)
• Nitrate, phosphate and ammonium developments with time (progressively)
• On annual unbiased averages (...seasonal..)
• To compute time trends and year of target meeting (achieve... at the latest...)
11 European Environment AgencyHarmoniRIB 21/09/2006
Sample results•Intra strata variability,•Inter-years variability•Source data uncertainty•Stratum uncertainty•Orphan strata assessment•Increasing uncertainty with reduction in size•Stationarity, linearity?•Reporting choices
12 European Environment AgencyHarmoniRIB 21/09/2006
“Thickness of the line”
Data dispersion frame
Reporting uncertainties
Non compliant observations post
target meeting
13 European Environment AgencyHarmoniRIB 21/09/2006
Towards conclusion...
• Data related uncertainties now soundly addressed
• Assessment related uncertainties are necessary for accurate policy assessment
• They becomes considered at the assessment stage and addressed, but risk of shortage in data might make it inapplicable
• Lack of understanding the uncertainty attached to complex systems of data, key to spatial assessment .
Thanks for your attention