mosart1_0_28

Explicitly truncate the river length to a mininum value, rather than
just change the calculation of tlen for rlen_min.

Testing: Sean and Keith both ran long simulations with this change (in a branch off of 1_0_26)

Previous simulations showed that short rivers would accumulate water in them slowly over
time. This doesn't happen in longer rivers. But it affects the total water balance of the
system. Simulations by Sean Swenson and Keith Oleson show that this no longer happens
with this change.

Changes from Sean Swenson to add minimum value to rlen (length of main channel); rlen values
can be too small, leading to tlen values that are too large

Author: ekluzek

doc/ChangeLog

@@ -1,3 +1,24 @@
+===============================================================
+Tag name:  mosart1_0_28
+Originator(s): erik
+Date: Oct 05, 2017
+One-line Summary:  Explicitly truncate the river length to a mininum value, rather than
+                   just change the calculation of tlen for rlen_min.
+
+Testing: Sean and Keith both ran long simulations with this change (in a branch off of 1_0_26)
+
+Previous simulations showed that short rivers would accumulate water in them slowly over 
+time. This doesn't happen in longer rivers. But it affects the total water balance of the
+system. Simulations by Sean Swenson and Keith Oleson show that this no longer happens
+with this change.
+
+Changes from Sean Swenson to add minimum value to rlen (length of main channel); rlen values 
+can be too small, leading to tlen values that are too large
+
+M    src/riverroute/RtmIO.F90 - Explicitly set rlen to rlen_min early and remove code that was
+        calculating tlen based on rlen_min when rlen<rlen_min. Since rlen is changed this affects code
+        that is based on rlen, and can change the if statements that are triggered.
+
 ===============================================================
 Tag name:  mosart1_0_27
 Originator(s): erik

src/riverroute/RtmMod.F90

@@ -2551,6 +2551,16 @@ subroutine MOSART_init
    ! control parameters and some other derived parameters
    ! estimate derived input variables
 
+   ! add minimum value to rlen (length of main channel); rlen values can 
+   ! be too small, leading to tlen values that are too large
+
+     do iunit=rtmCTL%begr,rtmCTL%endr
+        rlen_min = sqrt(TUnit%area(iunit))                           
+        if(TUnit%rlen(iunit) < rlen_min) then
+           TUnit%rlen(iunit) = rlen_min
+        end if
+     end do     
+
      do iunit=rtmCTL%begr,rtmCTL%endr
         if(TUnit%Gxr(iunit) > 0._r8) then
            TUnit%rlenTotal(iunit) = TUnit%area(iunit)*TUnit%Gxr(iunit)
@@ -2574,18 +2584,7 @@ subroutine MOSART_init
               TUnit%hlen(iunit) = hlen_max   ! allievate the outlier in drainag\e density estimation. TO DO
            end if
 
-           ! rlen values can be too small, leading to tlen values that are 
-           ! too large; calculate tlen using rlen_min when necessary
-           ! an alternate fix (commented out above) would be to directly
-           ! adjust rlen, then all subsequent calculations based on rlen
-           ! would be consistent.  With this change, other calculations 
-           ! use the original rlen, e.g. twidth
-           rlen_min = sqrt(TUnit%area(iunit))                           
-           if(TUnit%rlen(iunit) < rlen_min) then                        
-              TUnit%tlen(iunit) = TUnit%area(iunit) / rlen_min / 2._r8 - TUnit%hlen(iunit)
-           else
-              TUnit%tlen(iunit) = TUnit%area(iunit) / TUnit%rlen(iunit) / 2._r8 - TUnit%hlen(iunit)
-           end if
+           TUnit%tlen(iunit) = TUnit%area(iunit) / TUnit%rlen(iunit) / 2._r8 - TUnit%hlen(iunit)
 
            if(TUnit%twidth(iunit) < 0._r8) then
               TUnit%twidth(iunit) = 0._r8