Problems in calculating tracers with source

[miaocb]

"vsource" and "msource" are used to calculate tr_el (schism_step.F90, line 7469-7487).
According to my understanding, "vsource" used here should be real sources associated with msource, not the sum of vsource, vsink, precipation and evaporation (schism_step.F90, 1602-1641).

[josephzhang8]

No, vsource in the code is the sum. If it's negative (net sink), tracer mass would leave the domain so msource should not be applied.

[miaocb]

Let's consider some scenarios:

(1) A grid element has one source ( 0.1m^3/s ) with tracer and one sink (-0.1m^3/s ), no precipation and evaporation. vsource=source+sink=0. "vsource" is zero, but the tracer can still be taken into the water, just like a bowl of water with one pipe putting ink into the bowl and one pipe taking water out of the bowl. The source and sink are in the same grid emelent, but are separated from each other.

(2) A grid element has one source (0.1m^3/s) with tracer, and evaporation is -0.1m^3/s，no sink, no precipation. vsource=source+evaporation=0. "vsource" is zero, but tracers can also be taken into water because evaporation does not take tracers out.

(3) A grid element has one source (0.1m^3/s) with tracer, and precipation is 0.1m^3/s with no tracer，no sink, no evaporation. vsource=source+precipation=0.2m^3/s. Precipation does not take tracers into water, so "vsource" can not be the sum of source and precipation in the calculation of tr_el.

[josephzhang8]

See below.

-Joseph Joseph Zhang Office: (804) 684 7466 Web: schism.wiki From: Paul ***@***.***> Sent: Wednesday, August 31, 2022 1:24 AM To: schism-dev/schism ***@***.***> Cc: Y. Joseph Zhang ***@***.***>; Comment ***@***.***> Subject: Re: [schism-dev/schism] Problems in calculating tracers with source (Issue #79) [EXTERNAL to VIMS received message] Let's consider some scenarios: 1. A grid element with one source ( 0.1m^3/s ) and one sink (-0.1m^3/s ), no precipation and evaporation. vsource=source+sink. "vsource" is zero, but the tracer can still be taken into the water, just like a bowl of water with one pipe putting ink into the bowl and one pipe taking water out of the bowl. The source and sink are in the same grid emelent, but are separated from each other.

> Since source and sink are in same elem, they cancel out and we assume the reaction time is short (<dt), so there is no net tracer change.

1. A grid element with one source (0.1m^3/s), with evaporation (-0.1m^3/s)，no sink, no precipation. vsource=source+evaporation. "vsource" is zero, but tracers can also be taken into water because evaporation does not take tracers out.

> Evap does take volume/mass/tracer out, and we assume this is to be equivalent to sink. Not perfect assumption, but workable.

1. A grid element with one source (0.1m^3/s), with precipation (0.1m^3/s)，no sink, no evaporation. vsource=source+precipation=0.2m^/s. Precipation does not take tracers into water, so "vsource" can not be the sum of source and precipation in the calculation of tr_el.

> Surely precip adds both volume and mass into the water. The tricky part is the tracer concentration of rain water. If nws=2, the rainwater is assume to have S=0 and ambient T..

In general, those assumptions we made are not perfect for sure, but serve as a starting point. Reply to this email directly, view it on GitHub<https://nam11.safelinks.protection.outlook.com/?url=https%3A%2F%2Fgithub.com%2Fschism-dev%2Fschism%2Fissues%2F79%23issuecomment-1232477493&data=05%7C01%7Cyjzhang%40vims.edu%7C5199ea72bca44704907f08da8b11086b%7C8cbcddd9588d4e3b9c1e2367dbdf1740%7C0%7C0%7C637975202533590973%7CUnknown%7CTWFpbGZsb3d8eyJWIjoiMC4wLjAwMDAiLCJQIjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVCI6Mn0%3D%7C3000%7C%7C%7C&sdata=Ot21PK0G7Z1i970opB2zt96ftj9x2%2BTf7szqSC3xOns%3D&reserved=0>, or unsubscribe<https://nam11.safelinks.protection.outlook.com/?url=https%3A%2F%2Fgithub.com%2Fnotifications%2Funsubscribe-auth%2FAFBKNZY7Y52B2UOQSOYCV33V33T7TANCNFSM6AAAAAAQA3UWYE&data=05%7C01%7Cyjzhang%40vims.edu%7C5199ea72bca44704907f08da8b11086b%7C8cbcddd9588d4e3b9c1e2367dbdf1740%7C0%7C0%7C637975202533590973%7CUnknown%7CTWFpbGZsb3d8eyJWIjoiMC4wLjAwMDAiLCJQIjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVCI6Mn0%3D%7C3000%7C%7C%7C&sdata=OWo6G8OM%2FkV88fpEOXp1yfGLeZBT0pBc1nhV85FApyk%3D&reserved=0>. You are receiving this because you commented.Message ID: ***@***.******@***.***>>

[miaocb]

A simple experiment is prepared for testing.
Main configuration:
1 element with size 10mx10m, no open boundary, depth is 5m
initial condition: temperature(20 degC), salinity(0)
1 source, no sink
vsource: 0 m^3/s
msource: temperature (20 degC), salinity (35)
impose_net_flux=1, precipation-evaporation: 0.001kg/m^2/s

The source is 0 m^3/s, and as we know precipation does not take salinity to the water, salinity should be 0, but salinity increases with time as below.

Files of the experiment is attached below.
coarse_prec.zip

[josephzhang8]

For T,S exchanges including air-sea and sources, the baroclinic configuration (ibc=0, ihconsv=isconsv=1) is the more accurate choice (by using bulk aerodynamic algorithm or boundary condition for S). I assume you were using 2D configuration (and the logic for impose_net_flux is complex so I've recently moved it to a CPP, i.e. not actively supported). As I said, the current method is not perfect. If you can demonstrate that with your choice (vsource<0 only for original sources) you can get consistently better results for _all_ cases, I'll be willing to change the code. I'm not convinced about that; e.g., with 1 source and 1 sink at same elem, your method would add tracer into the system. Thx.

…

-Joseph Joseph Zhang Office: (804) 684 7466 Web: schism.wiki From: Paul ***@***.***> Sent: Sunday, September 4, 2022 10:49 PM To: schism-dev/schism ***@***.***> Cc: Y. Joseph Zhang ***@***.***>; Comment ***@***.***> Subject: Re: [schism-dev/schism] Problems in calculating tracers with source (Issue #79) [EXTERNAL to VIMS received message] A simple experiment is prepared for testing. Main configuration: 1 element with size 10mx10m, no open boundary, depth is 5m initial condition: temperature(20 degC), salinity(0) 1 source, no sink vsource: 0 m^3/s msource: temperature (20 degC), salinity (35) impose_net_flux=1, precipation-evaporation: 0.001kg/m^2/s The source is 0 m^3/s, and as we know precipation does not take salinity to the water, salinity should be 0, but salinity increases with time as below. [??]<https://nam11.safelinks.protection.outlook.com/?url=https%3A%2F%2Fuser-images.githubusercontent.com%2F11263432%2F188350775-7ad4961d-b9a8-4663-b032-80431c25e636.png&data=05%7C01%7Cyjzhang%40vims.edu%7C22fb3b7ac57f4ccd9ce608da8ee92397%7C8cbcddd9588d4e3b9c1e2367dbdf1740%7C0%7C1%7C637979429235517150%7CUnknown%7CTWFpbGZsb3d8eyJWIjoiMC4wLjAwMDAiLCJQIjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVCI6Mn0%3D%7C3000%7C%7C%7C&sdata=ODGJCd3GRIG1HXaVxApYkNkZCPEO7Wdm38WynGU5caU%3D&reserved=0> Files of the experiment is attached below. coarse_prec.zip<https://nam11.safelinks.protection.outlook.com/?url=https%3A%2F%2Fgithub.com%2Fschism-dev%2Fschism%2Ffiles%2F9486066%2Fcoarse_prec.zip&data=05%7C01%7Cyjzhang%40vims.edu%7C22fb3b7ac57f4ccd9ce608da8ee92397%7C8cbcddd9588d4e3b9c1e2367dbdf1740%7C0%7C1%7C637979429235517150%7CUnknown%7CTWFpbGZsb3d8eyJWIjoiMC4wLjAwMDAiLCJQIjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVCI6Mn0%3D%7C3000%7C%7C%7C&sdata=5wxz6NQdqNNzUmy1ikt7MAIax2zJM2q9Ti3o6Lipflo%3D&reserved=0> - Reply to this email directly, view it on GitHub<https://nam11.safelinks.protection.outlook.com/?url=https%3A%2F%2Fgithub.com%2Fschism-dev%2Fschism%2Fissues%2F79%23issuecomment-1236493541&data=05%7C01%7Cyjzhang%40vims.edu%7C22fb3b7ac57f4ccd9ce608da8ee92397%7C8cbcddd9588d4e3b9c1e2367dbdf1740%7C0%7C1%7C637979429235517150%7CUnknown%7CTWFpbGZsb3d8eyJWIjoiMC4wLjAwMDAiLCJQIjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVCI6Mn0%3D%7C3000%7C%7C%7C&sdata=HjAQB3ohUQXzJe9jMeCLUYaqR6HN6mNqFMgUv%2BOQYjs%3D&reserved=0>, or unsubscribe<https://nam11.safelinks.protection.outlook.com/?url=https%3A%2F%2Fgithub.com%2Fnotifications%2Funsubscribe-auth%2FAFBKNZ4ZJC5KP6HHNJYAG2DV4VNQPANCNFSM6AAAAAAQA3UWYE&data=05%7C01%7Cyjzhang%40vims.edu%7C22fb3b7ac57f4ccd9ce608da8ee92397%7C8cbcddd9588d4e3b9c1e2367dbdf1740%7C0%7C1%7C637979429235517150%7CUnknown%7CTWFpbGZsb3d8eyJWIjoiMC4wLjAwMDAiLCJQIjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVCI6Mn0%3D%7C3000%7C%7C%7C&sdata=DKMQfjSKuZlijr4IsblIP2z9fx45CqMLCEayCsLY4%2BM%3D&reserved=0>. You are receiving this because you commented.Message ID: ***@***.******@***.***>>

[miaocb]

Hi Joseph,
I know that there may be no perfect method, but I think using vsoure <0 only for original sources is a slightly better choice.
(1) case1, one grid element with size 10mx10m, the source and sink are in the same element.
Results show salt are not taken into the water ( salinity is 0, top right panel in the figure ), since vsource and vsink cancel out according to current method.

(2) case2, refine the grid, with 4 grid elements of size 5mx5m. After refining the grid, the source and sink are in two separate elements. Results show that salt are taken into the water ( salinity reaches 35, top right panel in the figure ),. The results of case1 and case2 are not consistent.

(3) case3, same to case1 except using vsource <0 only for original sources.
The result is now consistent with case2 ( top right panel in the figure ).

I will check how other models (EFDC, Delft3d, Mike, etc.) deals with the case when sink and source are in the same element if needed.

[josephzhang8]

Hi Paul:

I’d argue that Cases 1 and 2 should NOT give same results; mesh resolution and placement of source/sinks do matter.

[josephzhang8]

Hi Paul: I'd argue that Cases 1 and 2 should NOT give same results; mesh resolution and placement of source/sinks do matter.

…

-Joseph Joseph Zhang Office: (804) 684 7466 Web: schism.wiki From: Paul ***@***.***> Sent: Saturday, September 10, 2022 4:34 AM To: schism-dev/schism ***@***.***> Cc: Y. Joseph Zhang ***@***.***>; Comment ***@***.***> Subject: Re: [schism-dev/schism] Problems in calculating tracers with source (Issue #79) [EXTERNAL to VIMS received message] Hi Joseph, I know that there may be no perfect method, but I think using vsoure <0 only for original sources is a slightly better choice. (1) case1, 1 grid element with size 10mx1m, the source and sink are in the same element. Salt are not taken into the water, since vsource and vsink cancel out according to current method. [??]<https://nam11.safelinks.protection.outlook.com/?url=https%3A%2F%2Fuser-images.githubusercontent.com%2F11263432%2F189475013-a409e3ef-437b-4d1b-900e-c0ee53557405.png&data=05%7C01%7Cyjzhang%40vims.edu%7C599483069b8a4b81af8e08da93073527%7C8cbcddd9588d4e3b9c1e2367dbdf1740%7C0%7C0%7C637983956425347242%7CUnknown%7CTWFpbGZsb3d8eyJWIjoiMC4wLjAwMDAiLCJQIjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVCI6Mn0%3D%7C3000%7C%7C%7C&sdata=0XUOhfPcIsX0iA%2FfJ6u29c8jfCCLTriRaSJZF7AFsWQ%3D&reserved=0> (2) case2, now refine the grid in case1, with 4 grid elements, now the source and sink are in two elements separately. After refining the grid, source and sink are in two elements, salt are taken into the water. The results of case1 and case2 are not consinstent. [??]<https://nam11.safelinks.protection.outlook.com/?url=https%3A%2F%2Fuser-images.githubusercontent.com%2F11263432%2F189475113-21fb824d-1f02-4004-a351-2b0d68962085.png&data=05%7C01%7Cyjzhang%40vims.edu%7C599483069b8a4b81af8e08da93073527%7C8cbcddd9588d4e3b9c1e2367dbdf1740%7C0%7C0%7C637983956425347242%7CUnknown%7CTWFpbGZsb3d8eyJWIjoiMC4wLjAwMDAiLCJQIjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVCI6Mn0%3D%7C3000%7C%7C%7C&sdata=cKpQon89FifQ0NkXpaePGojaRURpIBkwWiBMQZA4LVg%3D&reserved=0> (3) case3, same to case2, except using vsoure <0 only for original sources. The result is consistent with case2 with only 1 grid element. [??]<https://nam11.safelinks.protection.outlook.com/?url=https%3A%2F%2Fuser-images.githubusercontent.com%2F11263432%2F189475030-5fc77ccf-6d31-4a4b-8f2d-a43e7a0782a0.png&data=05%7C01%7Cyjzhang%40vims.edu%7C599483069b8a4b81af8e08da93073527%7C8cbcddd9588d4e3b9c1e2367dbdf1740%7C0%7C0%7C637983956425347242%7CUnknown%7CTWFpbGZsb3d8eyJWIjoiMC4wLjAwMDAiLCJQIjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVCI6Mn0%3D%7C3000%7C%7C%7C&sdata=csUwftrSTGvQ3SvnPGdjUmYz2G750Xud%2F6BqylhfEb0%3D&reserved=0> - Reply to this email directly, view it on GitHub<https://nam11.safelinks.protection.outlook.com/?url=https%3A%2F%2Fgithub.com%2Fschism-dev%2Fschism%2Fissues%2F79%23issuecomment-1242669365&data=05%7C01%7Cyjzhang%40vims.edu%7C599483069b8a4b81af8e08da93073527%7C8cbcddd9588d4e3b9c1e2367dbdf1740%7C0%7C0%7C637983956425347242%7CUnknown%7CTWFpbGZsb3d8eyJWIjoiMC4wLjAwMDAiLCJQIjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVCI6Mn0%3D%7C3000%7C%7C%7C&sdata=ZJgwP3I3L9nAtBa0pJi0aQpaDvvCy42c2zkob0UShh8%3D&reserved=0>, or unsubscribe<https://nam11.safelinks.protection.outlook.com/?url=https%3A%2F%2Fgithub.com%2Fnotifications%2Funsubscribe-auth%2FAFBKNZ5YYWOVMBQV35YR7XLV5RBXNANCNFSM6AAAAAAQA3UWYE&data=05%7C01%7Cyjzhang%40vims.edu%7C599483069b8a4b81af8e08da93073527%7C8cbcddd9588d4e3b9c1e2367dbdf1740%7C0%7C0%7C637983956425347242%7CUnknown%7CTWFpbGZsb3d8eyJWIjoiMC4wLjAwMDAiLCJQIjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVCI6Mn0%3D%7C3000%7C%7C%7C&sdata=HEXfkrudhJuVYg2x%2FxWCNA2q3ryb5Zm7d0h4U9r5Km0%3D&reserved=0>. You are receiving this because you commented.Message ID: ***@***.******@***.***>>