ET is not fully sensitive to Rainfall -> total outflow always >> inflow

[MostafaGomaa93]

Hi all, @bobzsu @Crystal-szj @yijianzeng @Yunfei-Wang1993 @EntingTang @prajzwal08
I did some trials for 3 different sites: Loobos (Netherlands), one site in Spain and one in Italy. The simulations were run without groundwater and using the default parameters generated by PyStemmusScope. For each site, I conducted two experiments: one with full precipitation and another with 50% precipitation, so in total, 6 models, each model ran over 2 years.

Results show that output (ET) is always >> input (effective precipitation) in all models, except for the first one (Loobos, full precipitation), where rainfall was too high. But with 50% rainfall, we see the problem more obvious. The imbalance will become even higher if we reduce the hydraulic conductivity, so runoff will increase. Also, if groundwater is activated, groundwater recharge will be added to the output and more imbalance.

My feeling is that the ET calculation is somehow not very sensitive to rainfall and soil moisture, or soil moisture is not reduced appropriately after ET is removed from the soil

See please below the table of the flux values and the ET figures

Update

• Trials with no rainfall for the 3 sites are added. Table and Figures are updated
• Change in storage and flux at bottom boundary is added to the Table
• Water stress factor and soil moisture figures are added
• Simulation named (NL-Loo full rain) looks not realistic at all and has lots of issues (so we may disregard it from the discussion)

[MostafaGomaa93]

Units are mm/yr

Model	Year	Precipitation	Runoff	Effective precip	ET	Delta Storage	Flux at Bottom Boundary
NL-Loo full rain	2016	623	79	543	479	-899
	2017	754	221	533	400	-354
	total	1377	301	1076	879	-1257
NL-Loo half rain	2016	311	41	270	438	-988	-2265
	2017	377	47	330	395	66	-2632
	total	689	88	600	834	-922	-4897
NL-Loo no rain	2016	0	0	0	317	-975	-1465
	2017	0	0	0	214	188	-734
	total	0	0	0	532	-787	-2199
ES-Abr full rain	2016	418	140	277	519	-95	-3
	2017	291	121	170	402	-16	-4
	total	709	261	448	921	-111	-7
ES-Abr half rain	2016	209	41	167	465	-95	-4
	2017	146	40	105	358	-36	-3
	total	355	81	272	823	-131	-7
ES-Abr no rain	2016	0	0	0	289	-156	-4
	2017	0	0	0	204	-13	-4
	total	0	0	0	493	-169	-8
IT-BCi full rain	2016	842	321	521	705	-479	-350
	2017	665	248	416	528	16	6
	total	1507	569	937	1233	-463	-344
IT-BCi half rain	2016	421	102	319	571	-511	-317
	2017	332	78	254	384	-10	22
	total	753	180	573	955	-521	-295
IT-BCi no rain	2016	0	0	0	333	-590	-285
	2017	0	0	0	167	-33	47
	total	0	0	0	500	-624	-238

[MostafaGomaa93]

Loobos site

Spain site (ES-Abr)

Italy site (IT-BCi)

[Yunfei-Wang1993]

Hi Mostafa, This issue is likely related to the bottom boundary condition of STEMMUS. If you want to check the water balance, you should take into account both the change in soil water storage and the water flux at the bottom boundary. I hope this serves as a useful reference. Best regards, Yunfei

…

________________________________ 发件人: Mostafa Gomaa Daoud ***@***.***> 发送时间: 2025年9月11日 17:58 收件人: EcoExtreML/STEMMUS_SCOPE ***@***.***> 抄送: Wang, Yunfei (UT-ITC) ***@***.***>; Mention ***@***.***> 主题: [EcoExtreML/STEMMUS_SCOPE] ET is not fully sensitive to Rainfall -> total outflow always >> inflow (Issue #285) [https://avatars.githubusercontent.com/u/54531356?s=20&v=4]MostafaGomaa93 created an issue (EcoExtreML/STEMMUS_SCOPE#285)<#285> Hi all, @bobzsu<https://github.com/bobzsu> @Crystal-szj<https://github.com/Crystal-szj> @yijianzeng<https://github.com/yijianzeng> @Yunfei-Wang1993<https://github.com/Yunfei-Wang1993> @EntingTang<https://github.com/EntingTang> I did some trials for 3 different sites: Loobos (Netherlands), one site in Spain and one in Italy. The simulations were run without groundwater and using the default parameters generated by PyStemmusScope. For each site, I conducted two experiments: one with full precipitation and another with 50% precipitation, so in total, 6 models, each model ran over 2 years. Results show that output (ET) is always >> input (effective precipitation) in all models, except for the first one (Loobos, full precipitation), where rainfall was too high. But with 50% rainfall, we see the problem more obvious. The imbalance will become even higher if we reduce the hydraulic conductivity, so runoff will increase. Also, if groundwater is activated, groundwater recharge will be added to the output and more imbalance. My feeling is that the ET calculation is somehow not very sensitive to rainfall and soil moisture. See please below the table of the flux values and the ET figures ― Reply to this email directly, view it on GitHub<#285>, or unsubscribe<https://github.com/notifications/unsubscribe-auth/ALXM2AZGUE6GKZB5TFISBWL3SFBVHAVCNFSM6AAAAACGHC5ZX6VHI2DSMVQWIX3LMV43ASLTON2WKOZTGQYDKNZRGE4DIOA>. You are receiving this because you were mentioned.Message ID: ***@***.***>

[MostafaGomaa93]

This issue is likely related to the bottom boundary condition of STEMMUS. If you want to check the water balance, you should take into account both the change in soil water storage and the water flux at the bottom boundary. I hope this serves as a useful reference.

Hi Yunfei,
Total inflow = Total outflow + Change in Storage
Precipitation = Runoff + ET + Flux at bottom boundary + Change in Storage

After removal of runoff (Runoff = Precipitation - Effective precipitation), the equation will be like the following:
Effective precipitation = ET + Flux at bottom boundary + Change in Storage

If no groundwater -> Flux at the bottom boundary is negligible. Then the equation will be
Effective precipitation = ET + Change in Storage

With the numbers I showed above in the table, the Change in Storage will always be a high negative value, meaning that the soil is always losing water even if there is rainfall. And this is not physically correct. Change in Storage should be close to zero per each year, and Effective precipitation should be close to ET per each year

[Yunfei-Wang1993]

This issue is likely related to the bottom boundary condition of STEMMUS. If you want to check the water balance, you should take into account both the change in soil water storage and the water flux at the bottom boundary. I hope this serves as a useful reference.

Hi Yunfei, Total inflow = Total outflow + Change in Storage Precipitation = Runoff + ET + Flux at bottom boundary + Change in Storage

After removal of runoff (Runoff = Precipitation - Effective precipitation), the equation will be like the following: Effective precipitation = ET + Flux at bottom boundary + Change in Storage

If no groundwater -> Flux at the bottom boundary is negligible. Then the equation will be Effective precipitation = ET + Change in Storage

With the numbers I showed above in the table, the Change in Storage will always be a high negative value, meaning that the soil is always losing water even if there is rainfall. And this is not physically correct. Change in Storage should be close to zero per each year, and Effective precipitation should be close to ET per each year

Hi Mostafa,

The flux at the bottom boundary should not be neglected, especially if you set the bottom boundary condition as a specific water potential.

Another point is that when I ran STEMMUS-SCOPE at permanent wetland sites in Canada (with the bottom boundary condition set to gravity), the simulated annual ET was around 400 mm, which corresponds well to the annual precipitation of 400–500 mm. However, the actual annual ET at these sites is closer to 800 mm. This suggests that STEMMUS-SCOPE responds well to precipitation, but may underestimate ET under such conditions.

[Yunfei-Wang1993]

This issue is likely related to the bottom boundary condition of STEMMUS. If you want to check the water balance, you should take into account both the change in soil water storage and the water flux at the bottom boundary. I hope this serves as a useful reference.

Hi Yunfei, Total inflow = Total outflow + Change in Storage Precipitation = Runoff + ET + Flux at bottom boundary + Change in Storage

After removal of runoff (Runoff = Precipitation - Effective precipitation), the equation will be like the following: Effective precipitation = ET + Flux at bottom boundary + Change in Storage

If no groundwater -> Flux at the bottom boundary is negligible. Then the equation will be Effective precipitation = ET + Change in Storage

With the numbers I showed above in the table, the Change in Storage will always be a high negative value, meaning that the soil is always losing water even if there is rainfall. And this is not physically correct. Change in Storage should be close to zero per each year, and Effective precipitation should be close to ET per each year

Another test you can try is setting rainfall to zero and bottom water potential is set to zero; you will still obtain a high ET because the bottom water flux becomes very large.

[MostafaGomaa93]

Another test you can try is setting rainfall to zero, and bottom water potential is set to zero; you will still obtain a high ET because the bottom water flux becomes very large.

Yes, this is correct because when the bottom water potential = zero, it refers to the position of the water table (saturated zone), so even with no rainfall, there could be high ET indeed. But this is not the default case of STEMMUS-SCOPE.

The default setting of STEMMUS-SCOPE, which I also used in these trials, is that the bottom boundary is free drainage. I have added now to the table and figures above, additional model trials that run with no rainfall at all. With no rainfall, soil mositure is very low and flux at the bottom boundary is neglected.

However, I am still surprised how the model is producing ET without rainfall and with a free drainage boundary condition

[Yunfei-Wang1993]

Another test you can try is setting rainfall to zero, and bottom water potential is set to zero; you will still obtain a high ET because the bottom water flux becomes very large.

Yes, this is correct because when the bottom water potential = zero, it refers to the position of the water table (saturated zone), so even with no rainfall, there could be high ET indeed. But this is not the default case of STEMMUS-SCOPE.

The default setting of STEMMUS-SCOPE, which I also used in these trials, is that the bottom boundary is free drainage. I have added now to the table and figures above, additional model trials that run with no rainfall at all. With no rainfall, soil mositure is very low and flux at the bottom boundary is neglected.

However, I am still surprised how the model is producing ET without rainfall and with a free drainage boundary condition

What is the total soil depth in your simulation? Could you calculate the change in water storage and plot the soil moisture together with the water stress factor for IT-BCi? As shown in the table, when rainfall is set to 0, the simulated ET decreases from 333.3 mm to 167.2 mm. I think the soil moisture should be the source of the ET when there is no rainfall.

[MostafaGomaa93]

What is the total soil depth in your simulation? Could you calculate the change in water storage and plot the soil moisture together with the water stress factor for IT-BCi? As shown in the table, when rainfall is set to 0, the simulated ET decreases from 333.3 mm to 167.2 mm. I think the soil moisture should be the source of the ET when there is no rainfall.

• Total soil depth is 5 m.
• The change in water storage is added to the table.
• In Loobos, change in storage seems not realistic (too high).
• For the other two sites (Spain and Italy), change in storage in the first year is still too high and unrealistic, while in the 2nd year, change in storage value looks realistic (low value or close to zero)

[MostafaGomaa93]

Water stress factor for Spain site (ES-Abr)

Water stress factor for Italy site (IT-BCi)

[MostafaGomaa93]

Soil moisture for Spain site (ES-Abr)

Soil moisture for Italy site (IT-BCi)

[MostafaGomaa93]

@Yunfei-Wang1993
Flux at the bottom boundary is added to the table above

[Yunfei-Wang1993]

Soil moisture for Spain site (ES-Abr)

### Soil moisture for Italy site (IT-BCi)

Hi Mostafa, could you check the bottom boundary condition for water potential? Also, what is the PFT of ES-Abr? It seems that h = 0. I know that IT-BCi is cropland, so the boundary condition is set to 0 in the model.

[MostafaGomaa93]

Hi Mostafa, could you check the bottom boundary condition for water potential? Also, what is the PFT of ES-Abr? It seems that h = 0. I know that IT-BCi is cropland, so the boundary condition is set to 0 in the model.

For IT-BCi, bottom boundary is a specified head with a very low negative value = -9e-10 ~ zero (as stated here in the code)
For NL-Loo and ES-Abr, bottom boundary is free drainage

[yijianzeng]

@MostafaGomaa93 Dear Mostafa, we need to look into all the steps in your workflow in order to understand how you do the water balance calculation. Please help create a repository, particularly for this purpose, with a notebook (explained with markdowns) and all the figures/data you used in this issue.