merged cross section dependence of ktdurham (with CKKW-L matching)

Dear Sir,

When I run a simple matching process :
p p > l+ l-
+ p p > l+ l- j
+ p p > l+ l- j j

with CKKW-L and different ktdurham scale :
 5.0 (or 22.5) = ktdurham
 0.4 = dparameter
mll>50
nEvents 25000
cm energy 4000*4000

Merging:TMS = 5.0 (or 22.5)
Merging:Process = pp>l+l-
Merging:nJetMax = 2
SysCalc:fullCutVariation = off
Merging:applyVeto=on

In this case, the 0 jet cross section is ~860 pb.
The merged cross section for ktdurham = 5.0 give ~350 pb,
and for ktdurham = 22.5 give ~700 pb.

In my understanding through MLM matching, the merged cross section will larger than/close to 0j cross section.
And usually the larger the xqcut, the smaller the cross section.

May I ask what is the definition of ktdurham? And how to understand this merged cross section given by CKKW-L matching?
Or is I do something wrong?

Thanks!

Best,
Jung