Changes between Version 10 and Version 11 of DevelopmentActivities/AccelerationSpinup

Timestamp:

2011-07-22T08:33:28+02:00 (13 years ago)

Author:

dsolyga

Comment:

--

DevelopmentActivities/AccelerationSpinup

@@ -131 +131 @@
 we are unable to represent the inter-annual variability of a carbon pool. This phenomenom is particularly
 striking for the carbon pools with a strong variability like the metabolic litter.
+
+
+=== Relative error ===
+
+The following figures show the evolution of the relative error for the algebraic. 
+We assume that the value found in the last restart file of the job represents the real "equilibrium".
+
+
+
+In order to reduce the oscillations due to the decimal part of the values, we "smooth" 
+the curves. Without this treatment, we obtain in the case of the metabolic litter (below) :
+
+[[Image(error_LITTER_MET_AB_inferior_to_0.5%.gif)]]
+
+So we use the following instruction in our ferret script :
+
+{{{
+let error_$1=100*abs($2 - $1_star[d=2,k=10,l=1:7000@sbx:40])/$2
+}}}
+
+This means we smooth our curves. Finally, we got :
+
+[[Image(error_LITTER_STR_AB_inferior_to_0.2%.gif)]] [[Image(error_LITTER_STR_BE_inferior_to_0.2%.gif)]]
+[[Image(error_LITTER_MET_AB_inferior_to_0.2%.gif)]]  [[Image(error_LITTER_MET_BE_inferior_to_0.5%.gif)]]
+[[Image(error_CARBON_ACTIVE_inferior_to_0.5%.gif)]]   [[Image(error_CARBON_SLOW_inferior_to_0.5%.gif)]]
+[[Image(error_CARBON_PASSIVE_inferior_to_2%.gif)]]  
+
 
 === Comparison with Forcesoil ===