diff --git a/README.md b/README.md
new file mode 100644
index 0000000000000000000000000000000000000000..bebb7a72d683e1cd1d47517686003cfc70880337
--- /dev/null
+++ b/README.md
@@ -0,0 +1,20 @@
+In ORCHIDEE the graph of the water routing through rivers was generated on the fly in the main code.
+
+This solution is not ideal for various reasons :
+
+    It is difficult to parallelize as in ORCHIDEE the domain decomposition does not foresee exchange of halos.
+    The routing graph produced in ORCHIDEE could not easily be verified before a simulation is launched.
+    We plan evolution of the routing scheme which will require a large number of extra information (floodplains, dams, adduction networks, ...) and it will be complicated to add this within the full model.
+
+This pre-processor should be able to generate the full information needed by ORCHIDEE to route the water 
+to the ocean or return it to evaporative processes. It required the grid of the model and the land/sea mask. All the rest should be ancillary information.
+
+Main code : RoutingPreProc.py
+
+F90subroutine : Subroutines from ORCHIDEE which build the routing graph.
+
+DomainDecompTests : Code to test the domain decomposition and paralelization tools.
+
+SampleScripts : Jobs which are used for some standard tests.
+
+TestConfigs : Examples of run.def files used to test the pre-processor. 
\ No newline at end of file