Linking typology clustering to river N and P loads to the NA Ocean LOICZVIEW clustering tool applied to the European North Atlantic coastal zone Linking typology clustering to river N and P loads to the NA Ocean The Hague, July 2001 Natacha Brion
Introduction European coast of the North Atlantic Ocean Outlet of large river systems (Rhine, Elbe, Seine, Loire, …) Dense population Industrialized Intense agriculture Systems extremely influenced by human activity --------> Impact on ecological functioning of the coastal system Objective: Find a classification of the different European Atlantic coastal environments based on typology data that can reflect the extend of the N and P load to the coastal zone.
Procedure abstract 1) Preparing the data to cluster: 2) Clustering: Choose most relevant typology data linked to N and P loads and extract them from the LOICZ-Typology data base. Import the data to an Excel sheet for appropriate filtration ---- .CSV file Extract from the Budget data base, corresponding sites with river N and P load data + add new data from literature. 4) Add budget data to typology data-------.CSV file 2) Clustering: In LOICZVIEW: upload the dataset and select ONLY the typology data. Cluster the data. 3) Evaluation: Visualize the clustered data and overlay the N-load data for evaluation.
Selecting Data to cluster: Typology Region: all Europe Data: Mean annual precipitation Temperature Basin population density Basin area Basin runoff % crop land coverage Target cells: Coastal cell -------------------------------------->create the data file and open in Excel Filtration: Remove all cells outside the geographic area of interest + remove cells with basin runoff = 0. ----------------------------------------->save as a .CSV file
Selecting Data to cluster: N and P loads Import the LOICZ budget data base in an excel sheet Remove all cells outside the considered area Result: Only 2 budgeted small estuaries in Ireland… Add new data from literature for other estuaries : Final 9 estuaries
Data of annual N and P loads are introduced in the previous typology data sheet at the corresponding CELL_ID ----------------> save as a>CSV file
Clustering Upload the data file and select all data except DIN and DIP load Make an MDL to estimate best number of clusters:
Cluster 0: high SUB BASIN AREA Cluster 1: low TEMP CRU ANNAVG, high PRECIP CRU TOTAL, low SUB BASIN AREA Cluster 2: high SUB BASIN POPULATION DENSITY, Cluster 3: high TEMP CRU ANNAVG, low SUB BASIN AREA, low SUB BASIN POPULATION DENSITY Cluster 4: high CELL PERCNT CROPLAND, Cluster 5: medium TEMP CRU ANN AVG, low PRECIP CRU TOTAL,
NITROGEN LOAD 1000 moles > year Class 0 > 5 000 000 4 points Cluster 0: high BASIN AREA Cluster 1: low TEMP, high PRECIP, low BASIN AREA Cluster 2: high POPULATION DENSITY, low BASIN AREA Cluster 3: high TEMP, low BASIN AREA, low POPULATION DENSITY Cluster 4: high PERCNT CROPLAND, low BASIN AREA, low BASIN POPULATION DENSITY Cluster 5: medium TEMP , low PRECIP , low BASIN AREA NITROGEN LOAD 1000 moles > year Class 0 > 5 000 000 4 points Class 1 1 000 000 - 5 000 000 7 points Class 2 100 000 - 1 000 000 1 Class 3 < 100 000 2 Class #0 Class #1 Class #2 Class #3 Cluster #0 100 14.29 0 0 Cluster #1 0 0 0 0 Cluster #2 0 28.57 100 0 Cluster #3 0 0 0 0 Cluster #4 0 14.29 0 0 Cluster #5 0 42.86 0 100
PHOSPHORUS LOAD 1000 moles / year Class 0 >100 000 8 points Cluster 0: high BASIN AREA Cluster 1: low TEMP, high PRECIP, low BASIN AREA Cluster 2: high POPULATION DENSITY, low BASIN AREA Cluster 3: high TEMP, low BASIN AREA, low POPULATION DENSITY Cluster 4: high PERCNT CROPLAND, low BASIN AREA, low BASIN POPULATION DENSITY Cluster 5: medium TEMP , low PRECIP , low BASIN AREA PHOSPHORUS LOAD 1000 moles / year Class 0 >100 000 8 points Class 1 10 000 - 100 000 4 Class 2 < 10 000 2 Class #0 Class #1 Class #2 Cluster #0 62.5 0 0 Cluster #1 0 0 0 Cluster #2 25 25 0 Cluster #3 0 0 0 Cluster #4 0 25 0 Cluster #5 12.5 50 100