Information Technology Reference
In-Depth Information
committee [3]. In the virtual landscape described above, each cell is either part of a
farm or in public domain. All farms can be fed with water pumped either in surface
water or in ground water through a pump entity (i.e. an instance of a class Pump)
located on a cell belonging to a farmer and in a specific compartment in that cell
(ground or surface). We assume that farmers have a full and perfect knowledge of
water needs and are able to pump whenever they need if water is available. Farmers
assess drought situation through water level at the location of their pump in the
relevant compartment (surface or groundwater) with comparison to their own
reference level.
Administration is represented by a single agent, an instance of PolicyMaker, with a
limited set of reference points and threshold associated to them and uses. We start
with only one reference point: the outlet of the largest basin in the area. When a
threshold is crossed, the rule forbids pumping to farmers in the associated area for the
stipulated water use. A major assumption for this agent is its incapacity of controlling
respect of restrictions. He has to rely on the willingness to comply of water users.
Other assumption regarding PolicyMaker is periodicity of its activity that is supposed
to be weekly: drought situation is assessed only every 7 days.
Water users are endowed with three possible attitudes regarding restriction rule
compliance: respect the rule (attitude := respect), respect the rule when agree on
drought situation (attitude := ownAssessment), don't respect the rule (attitude :=
noRespect). At each time step (the day), they decide to pump water according to their
needs, the rules, their attitude regarding the rule and their own assessment of drought
situation.
We assume at this level there are no direct interactions between users and
administration.
6
Model Implementation, Calibration and Verification
Model is implemented with Cormas platform (http://cormas.cirad.fr). It provides
several indicators to check its realism to empirical data: outflow for the various sub
basins, groundwater levels in each cell, but also water balance at cell and whole
county level. Outflows can be compared to observed discharge levels in the river
basin used to generate the maps for orders of magnitude as well as specific statistical
description of these flows, such as average value, variance, or the annual monthly
minimum flow with a return period of 5 years, QMNA5.
First level of model verification is checking that simulations comply with water
conservation at any scale. Figure 3 below shows the aggregated balance for the whole
basin along 3000 days. Absolute value of this balance is always below 0.2mm, which
is less than 1% of the average discharge at the main sub-basin outlet.
We calibrated the leakage parameter in order to be realistic in order of magnitude
of flow. Then we compared simulated mean flows and QMNA5, which is
representative of low water period, with observed ones over a 40 years period for the
Drome river valley and could calibrate the other hydrological parameters.
Search WWH ::




Custom Search