Longwave radiation budget problem in NoahMP LSM

[sudarsan3010]

I am using ERA-5 Land forcing with the following model configurations,

DYNAMIC_VEG_OPTION = 5
CANOPY_STOMATAL_RESISTANCE_OPTION = 1
BTR_OPTION = 1
SURFACE_RUNOFF_OPTION = 1
SUBSURFACE_RUNOFF_OPTION = 1
DVIC_INFILTRATION_OPTION = 1
SURFACE_DRAG_OPTION = 1
FROZEN_SOIL_OPTION = 1
SUPERCOOLED_WATER_OPTION = 1
RADIATIVE_TRANSFER_OPTION = 1
SNOW_ALBEDO_OPTION = 2
PCP_PARTITION_OPTION = 1
SNOW_THERMAL_CONDUCTIVITY = 1
TBOT_OPTION = 2
TEMP_TIME_SCHEME_OPTION = 1
GLACIER_OPTION = 1
SURFACE_RESISTANCE_OPTION = 1
SOIL_DATA_OPTION = 1
PEDOTRANSFER_OPTION = 1
CROP_OPTION = 0
IRRIGATION_OPTION = 0
IRRIGATION_METHOD = 0
TILE_DRAINAGE_OPTION = 0

FORCING_TIMESTEP = 3600
NOAH_TIMESTEP = 900
OUTPUT_TIMESTEP = 86400

SPLIT_OUTPUT_COUNT = 1
SKIP_FIRST_OUTPUT = .false.
RESTART_FREQUENCY_HOURS = -1

I have been trying to run the model for one year, but the model stops after a month (up to 20080123100) and gets stopped by giving the following error.

***DATE=2008-01-31_23:00:00 284.85693 0.95198 Timing: 0.07 Cumulative: 274.84 SFLX: 0.06
***DATE=2008-01-31_23:00:00 280.09055 2.24266 Timing: 0.07 Cumulative: 274.84 SFLX: 0.06
***DATE=2008-01-31_23:15:00 Timing: 0.08 Cumulative: 274.88 SFLX: 0.03
emitted longwave <0; skin T may be wrong due to inconsistent
input of VegFracGreen with LeafAreaIndex
VegFrac = 0.00000000 VegAreaIndEff = 0.00000000 TemperatureCanopy = 239.137390 TemperatureGrd = 111.317917
RadLwDownRefHeight = -176.621796 RadLwNetSfc = 176.037598 SnowDepth = 4.34467363
Note: The following floating-point exceptions are signalling: IEEE_UNDERFLOW_FLAG IEEE_DENORMAL
STOP Error: Longwave radiation budget problem in NoahMP LSM
***DATE=2008-01-31_23:15:00 285.39563 2.38023 Timing: 0.09 Cumulative: 274.90 SFLX: 0.04

[RAbolafiaRosenzweig]

Can you please check to make sure your input surface air temperature forcing is in units of Kelvin and varies within a physically realistic range? It looks like the canopy temperature is extremely cold (~ -34C) and the ground temperature is unrealistic (~ -162C), which makes me think that the surface air temp. forcing may not be in Kelvin. When you do this check, please also consider the other meteorological forcing variables and the corresponding units.

[sudarsan3010]

I have cross-checked the T2D and other meteorological forcings; every variable is in its range and units are appropriate. I have not changed anything from the Vtable. My study domain also covers the Himalayan region, with snow and glaciers.

From the last output obtained (at 2008013100), I am sharing below the screenshot of Ground Temperature (TG) and Canopy Air Temperature (TAH)

Ground Temperature (TG)

Canopy Air Temperature (TAH)

Is this error because I am using the hrldas-master from github, is there some changes in code in this version?

                                        --------------xxxx------------------xxxx-------------

2. Hrldas-v5.0.0, I am getting the following error. The model not even gives out the initial time step output, stops before that.

dst(1) = src(1)* 0.89286 + src(2)* 0.10714
layer_bottom(1:nsoil) = 0.0700 0.2800 1.0000 2.5500
layer_top(1:nsoil) = 0.0000 0.0700 0.2800 1.0000
Soil depth = 0.100000 0.300000 0.600000 1.000000
k, dst_centerpoint(k) = 1 5.00000007E-02
k, dst_centerpoint(k) = 2 0.250000000
k, dst_centerpoint(k) = 3 0.700000048
k, dst_centerpoint(k) = 4 1.50000000

k, src_centerpoint(k) = 1 3.50000001E-02
k, src_centerpoint(k) = 2 0.174999997
k, src_centerpoint(k) = 3 0.639999986
k, src_centerpoint(k) = 4 1.77499998
dst(1) = src(1)* 0.89286 + src(2)* 0.10714
At line 1725 of file module_hrldas_netcdf_io.f90
Fortran runtime error: Array bound mismatch for dimension 2 of array 'var' (4/32)

[tslin2]

maybe try using RADIATIVE_TRANSFER_OPTION = 2 or RADIATIVE_TRANSFER_OPTION = 3

[sudarsan3010]

I have tried using all the 3 options of Radiative transfer. It's still the same issue.

I have also changed the soil thickness to match with the Era-5 Land initialization, even that doesn't solve the issue.

[cenlinhe]

longwave error often occurs due to weird forcing values. My guess is that you have some unresolved coast points without any forcing values which caused the model crash.
If you used HRLDAS_forcing to generate your initial and forcing data, please try to increase the value of EXPAND_LOOP (e.g., to 10) in HRLDAS_forcing when you generate the initial and forcing data and re-run the model again.

[sudarsan3010]

I already reviewed the discussion threads and in the HRLDAS_forcing configuration, I used EXPAND_LOOP=10. I will check by increasing it further and update it here.

[ZhaoY-github]

It's likely that you are using a land use and land cover dataset that contains lake category, for example, MODIS lake=21, while your IST is still 1 (identified as soil) at certain grid cells, which makes the model confused.
Make sure XLAND in wrfinput and LANDMASK and LU_INDEX in geo_em.d01.nc are consistent. You could manual adjust the values in XLAND.
@cenlinhe any suggestion?

[sudarsan3010]

I realized my error. When using 'Prepare_ERA5-Land_Forcing.sh', it expects one file per year, but I processed it for each month, so the model terminated exactly at the end of the month.

Anyway, I have increased it to EXPAND_LOOPS=20 to overcome further errors.
Thank you all for your timely responses and suggestions. :)

[neha12-3]

i am also facing little bit similar issue.can anyone plz help

[sudarsan3010]

I think your image is not uploaded properly. It is not visible.