Shuttleworth-Wallace: Inconsistent behavior in wet to dry leaves transition

[philippkraft]

The potential transpiration rate PTR is reduced by the actual interception rate AIR- this is copied from BROOK90. However, AIR is derived from PIR with an own mechanism. This leads to a raise of ground evaporation GER if PIR>AIR>PTR. GER has the roughly same value for AIR=PIR>PTR and PIR>PTR>=AIR.

[philippkraft]

Effect can be seen with demo/surface-sw-et.py and was revealed by #35.

Problematic code:

cmf/cmf/cmf_core_src/upslope/vegetation/ShuttleworthWallace.cpp

Lines 865 to 878 in 51d20e4

	AIR = piecewise_linear(VOL_IN_CANOPY,0,MAX_VOL_IN_CANOPY,0,PIR);
	PTR -= AIR;
	if (PTR>0.001) {
	TBYLAYER(1,PTR,DISPC,ALPHA,KK,RROOTI,RXYLEM,PSITI,lc, RHOWG * pF_to_waterhead(4.2),1,ATR_sum,ATR);
	if (ATR_sum < PTR)
	SWGE(RN,RNground,w.e_s-w.e_a,RAA,RAS,RSS,DELTA,ATR_sum + AIR,GER);
	} else {
	PTR = 0.0;
	ATR = 0.0;
	ATR_sum=0.0;
	ATR = 0.0;
	SWGE(RN,RNground,w.e_s-w.e_a,RAA,RAS,RSS,DELTA,ATR_sum + AIR,GER);
	}

[philippkraft]

In BROOK90 "the canopy is considered to be either completely wetted or completely dry" (SUBROUTINE INTER), which is not the case for cmf / ShuttleworthWallace, where the canopy is only completely wet, when the canopy storage is completely filled.