partial review comments [skip ci]

.buildkite/longruns_gpu/pipeline.yml

@@ -37,7 +37,7 @@ steps:
         artifact_paths: "land_longrun_gpu/*png"
         agents:
           slurm_gpus: 1
-          slurm_time: 08:00:00
+          slurm_time: 03:00:00
         env:
           CLIMACOMMS_DEVICE: "CUDA"
 

src/standalone/Vegetation/PlantHydraulics.jl

@@ -342,11 +342,15 @@ Computes the water flux given the absolute potential (pressure/(ρg))
 and the conductivity along
 the flow path between these two points.
 
-We currently assuming an geometric
-mean for mean K_sat between the two points. Previously,
+We currently assuming a harmonic
+mean for mean K_sat between the two points (see CLM Technical Documentation). Previously,
 we used the arithmetic mean (Bonan, 2019; Zhu, 2008),
 but then when the soil K was very low, root extraction would
-continue. This should be modified for compartments of differing sizes.
+continue. 
+
+Following CLM, this should be modified for compartments of 
+differing sizes because the water will travel different path lengths 
+in each compartment. Hence we should weight each K as K -> K/path length.
 """
 function flux(z1::FT, z2::FT, ψ1::FT, ψ2::FT, K1::FT, K2::FT) where {FT}
     K_eff = K1 * K2 / max(K1 + K2, eps(FT))