Assumptions and idea for metrics

Food supply, food security and the independence of an economy in obtaining food are undoubtedly one of the primary indicators for quality of life. It is therefore interesting how difficult it is to find information on the supply, production and consumption of food. This information certainly exists in society, where shops, distributors, importers and manufacturers are well aware of the product that goes through the business, but it also involves numerous entities with different operations. Direct data collection from these sources would therefore be costly and would impose a considerable burden of response on parties who might be getting little back.

There would be the possibility of getting this data through direct access to cashier systems, stock control systems and delivery notes, but such collection is technically complex, the amount of data is enormous and, most importantly, not popular with data owners.

It is therefore worth trying your best to prepare a mathematical model that takes the data that Statistics Iceland already has in its possession for other statistical purposes. Such mathematical model can never get a complete picture of food supply, or food flow through the economy, but the model can get closer to reality than we stand today. The purpose of this project is therefore to prepare a measurement of how much food is available Icelandic economy after exporting the same product. In addition, it is hoped to assess the dependence on this food supply on imports.

Here supply was defined according to the following equation:

\[ FA = m(\textrm{IMP})+m(\textrm{P})-m(\textrm{EXP}) \] where m(IMP) is the quantity (in kg) of the imported product, m(EXP) is the quantity exported and m(P) the quantity derived from the production of food.

Foodstuffs are defined here as the main source of food on the table in the northern hemisphere, such as meat products, dairy products, eggs, vegetables and the like. By defining food according to European tradition, the food traditions in Asia, Africa, the South and Oceania are ignored. This causes some inaccuracy, as a large part of the food traditions as well as a fairly large proportion of slaughtered animals, such as blood, fat, bones and skins, are overlooked. Here it is not intended to look down on these traditions or write off valuable exports, such dried fish, roe and flour products, but this product cannot usually be bought in stores in Iceland.

Here, the term “food independence” is also defined as the ratio

\[ D = \frac{m(\textrm{DP})-m( \textrm{EXP})}{m(\textrm{DP})+m(\textrm{IMP})+m(\textrm{IP})} \] Here it is assumed that m(DP) is the amount of food produced from domestic raw materials, while m(IP) is the amount of food produced from imported raw materials. Here, the term “domestic raw material” is the raw material that companies within the Icelandic of the economy obtain from nature and deliver into the economy. Domestic fishing in other nation’s waters is not included if the goods are not delivered in Iceland. Domestic fishing in Icelandic waters and landings abroad is included, but this latter quantity reappears in export figures. The term “imported raw materials” can be the landing by foreign operators of catches that come from Icelandic waters.

In reality, producers almost always use some imported product to produce food from domestic raw materials, but here this separation is made on the numerical frame even though it is not present on the table.

This D measure can be from a negative number, for example when m(DP) = 0 and m(EXP) > 0, but this can easily occur when imported raw materials are used. Alternatively, the actual exported product may be heavier than the calculated amount in m(DP), since the production consists of adding salts, sugar, oil or water to the product. The model based does not attempt to assess all the diversity on which food processing is based.

If the calculated value for D is zero or close to zero, it can be assumed that the product is not independent of the output of the economy and that the output of the product is dependent on import, or that the product is produced primarily for export. The calculated value of D close to one indicates that the product is largely autonomous within the economy.

Here it is effervescent to subtract the exports in the denominator of D, however this makes the scale unnecessarily susceptible to exports, especially when exports and production are of the same magnitude.

Classification of food and definition of food flow paths throughout the economy

This design uses the classification system of materials from the Economy-Wide Material Flow Accounts; [EW-MFA] (https://hagstofa.is/talnaefni/umhverfi/efnisflaedi/efnisflaedireikningar/)).

! [Figure 1: Basic idea of flow in EW-MFA] (./images/ew_mfa_basic.png)

These accounts define that a product flows in and out of the economy as shown in the figure above. Domestic Extraction (DE) covers all raw material productions. Material flows back into nature after a flow path, which is here represented as Domestic Process Output (DPO). Material can also enter the economy as an Import (IMP) or Export (EXP).

EW-MFA defines materials primarily according to which part of the environment the product comes from, or the main topic category. It is a classification system that differs from classification systems commonly used in economic accounts, such as the CPA 2.1 system or the CN system which is more based on which economic class manufactures or uses the product and is therefore better suited to understanding the productivity and performance of businesses than on the environment.

Key topics in EW-MFA include:

• MF1-Biomatter: Biomaterials, where we expect to find all food products, but here also are other biomes
• MF2-Metals (gross ores): Metals
• MF3-Non-metallic minerals: Non-metals
• MF4-Fossil energy materials/carriers: Fuels
• MF5-Other products: Other mixed materials
• MF6-Waste for final treatment and disposal: Waste for processing - here it is assumed that waste materials can be valuable and “stored” in dumpsters for later processing.

The biological class is then branched down into further categories:

• MF11-Crops (excluding fodder crops)
• MF12-Crop residues (used) and fodder crops
• MF13-Wood
• MF14-Wild fish catch, aquatic plants and animals, hunting and gathering
• MF15-Live animals and animal products (excluding wild fish, aquatic plants and animals, hunted and gathered animals)
• MF16-Products mainly from biomass

Each category is then broken down into further categories. The EW-MFA classes fit quite well directly with the description of foods that are inherent. Here, however, some changes are made to go around one restriction in EW-MFA. The categories used here are:

• F1 - Total for all food categories
• F11 - Vegetables and plant products
• F12 - Crop residues and grazed biomass
• F13 - Wood
• F14 - Fish and aquatic products
• F15 - Meat and other farm animal products
• F16 - Product from mixed biomass classes

the sub-classes are not shown here

Possible link to food flow and other statistics

The national accounts account for private consumption of food and beverages. This consumption is recorded as a flow of capital from households and institutions, but such consumption would relate to consumption rather than to food supply. There are also a number of statistics derived from the turnover and balance sheets of food producers, where it is possible to distinguish between the advantages of the main inputs of production and the profits on turnover. These accounts do not, however, provide a good analysis between profits in fishing and profits in processing, as many fishing vessels are operated directly by fish producers and production is sometimes in the hands of fishing companies. These figures should, however, be closer to the related quantities entering the accounts here.

Data structure design

In the first edition of this experimental statistics, data are processed into the following data structure:

! [Data structure in í nodes-flow representation] (images/flowDesign.png)

The picture shows how the amount of material (tonnes) flows (by arrow) from one node to another. The definitions of the nodes are:

• DE: Domestic extraction
• DF: Animal husbandry (Agriculture). These may be farm animals or aquaculture
• IMP: Imports
• DP: Domestic production from domestic raw materials
• IP: Domestic production from imported raw materials
• DPO: Water, gas, harvest wastes or animal carcasses not used for anything else
• F: Food
• M: Non-food
• EXP: Export

The substance flowing between the nodes is indicated by

• Material category
• State of manufacture

In the next iteration of the model, a flow path from DE to DF will be created to account for grazing and feed production. An attempt will also be made to create a flow path between IMP and DF. However, these changes do not affect information on food production.

Here, too, a node for stocks (STO) could be created as a possible outflow to ensure that the mass into the system from DE and IMP is the same as the mass of material out of the system via EXP and STO

In the future, an attempt will be made to define industries in the material flow. This should improve the validation of the figures based on operational information.

Modelling Design

The mathematical model is built on three basic steps:

1. Conversion of lighting to material category and production stage. Here the description of the material depends on the data source. Information may originate from several different data sources. This step therefore needs to ensure that quantities are not double-counted. In some cases, the data sources do not agree on quantities, in which case a realistic estimate of which source is to be used has to be made. In most cases, the data source with a more accurate degradation of information is used.
2. Filtration steps. In this step, we try to determine how much of the product is intended for human consumption. Products such as cereals may be intended for seed potatoes or for compound foodstuff. No attempt was made to assess whether the harvest of barley and rapeseed oil in Iceland is suitable for food production. In the commercial base, the immigrant industry was considered to assess the purpose of imports. This allocation was offset over a three-year period in order to offset volatility in import volumes. Sales figures for dairy products were also considered to assess how much milk is used in the production of cheese, cream, butter, Skyr and other similar products.
3. Calculation steps. The quantities of material intended for food production are multiplied by a transformation matrix dividing the substance between DPO, M and F nodes. The account also takes care of changing the production level and material category if this happens.

Computation design

The first iteration of the project does indicate which industry is the recipient of the material. There is also a limited analysis of whether production varies in efficiency by month/season

Details of processing by data source

Food in Import and Export

Sorting customs numbers down to topics is part of the project when the EW-MFA account is created. Each customs number has a parallel MF class and CP 2.1 class and the EW-MFA system provides a description of how to group materials by production stages into three stages: Finished product (SM-FIN), large-scale container (SM-SFIN) or raw material (SM-RAW). A total of 6000 customs numbers are marked as biomaterials in the customs tariff. The number of categories varies slightly depending on the year.

The filtering uses analysis of how much of the product is imported by industry, agriculture, food businesses, or wholesalers/retailers. The filtration returns about 2500 products directly to the food category, while another 200 products get a division between M and F while the rest is not considered food and goes to the Milk.

The EW-MFA system defines the category MF16 for beverages (alcoholic and non-alcoholic), soup mixes, doughs and other materials composed of a variety of biomaterials. This definition is rather unfortunate, as this category has no direct connection to material flow from the environment. As a result, the customs data is the only source of this category in the first iteration of the project.

Import quantities for 2020 after the distribution of nodes

mfa_cd	F	IP	M
MF111	34247.348	2459.024	18934.818
MF1110	4548.394	1223.188
MF112	2336.067		149.350
MF113	8717.857	13.469	924.964
MF114	359.658
MF115	514.977
MF116	1085.831	2535.787	42.945
MF117	18921.164	59.316	9.516
MF118	29612.189		969.971
MF119	2.926		4.976
MF1211			2338.195
MF1212			26166.358
MF1221			59310.819
MF131			123092.071
MF132			17460.045
MF1411	1117.422	1878.852
MF1412	40.977	11.380	35320.137
MF1413	992.962	2827.013
MF1414			62684.336
MF1415			0.686
MF1421	15027.406	1935.022
MF1422	31.192	0.038	7.616
MF1423			3.542
MF151			10.338
MF152	2858.596	290.658	68.505
MF153	1119.677		12.090
MF154			316.314
MF16	57480.974	6848.876	38797.543

Production from imported raw materials

Customs categories entering as raw material or semi-finished material (e.g. Bulk-bags) are exit_node designated as IP. Proportions are then found for what proportion of the material ends up as food, non-food or waste. This means, for example, that 100 tonnes of unground grain ends up as 68 tonnes of grain for human consumption (F), while part of the imported grain is actually animal feed and is placed in the (M) end node. This does not allow for large losses of the material during milling.

Information in customs data is not as detailed as landing figures at the Directorate of Fisheries, so conversion constants for raw fish in food are based on the best utilization in filleting, which is about 55%. During filleting and drying trimmings, about 38% of the weight of the fish goes as water to the DPO nod, while the rest is placed in the Milk.

Split mass for 2020 from IP

mfa_cd	DPO	F	M
MF111	983.6096	1475.4144
MF1110		1223.1880
MF113		13.4690
MF116		253.5787	2282.208
MF117		59.3160
MF1411		1878.8520
MF1412			11.380
MF1413		2827.0130
MF1421	0.4896	1934.5134	0.019
MF1422		0.0380
MF152	4.6023	286.0557
MF16	8.6947	6832.5053	7.676

Animals in slaughter

Information is taken from the agricultural production base. In this base, the slaughter weight (carcass weight) of the animal is recorded. The identity of the slaughter authorisation holder and the veterinary quality classification of the carcass is also recorded. This information is used to assess whether the carcass went into food production or disposal. Food production is then assessed based on the dressing percentage retrieved from the Food and Agriculture Organization (FAO) database. These figures include the proportion of inclusions, blood, bones and fraction of the animal removed during the slaughter.

Unlike other places, blood and fat are less used in Iceland than in many other countries. A small amount of blood is used in blood sausage, but in Iceland it is not permitted to use blood as fertilizer or as a complementary feed. This object is therefore placed in the DPO node.

Iceland exports some of the skins, but the quality of the leather is not considered very good, so this material often ends up as a material that goes into landfill. Here, skins and leather are placed in the MF154 category and placed in the M nod. The difference between the known slaughter weight and the food and material product that has been specified is then placed in the DPO node. This ignores the potential production of diesel from fat and other uses that are under development.

Weight of material and division for 2020 from DF node

mfa_cd	tpe	DP	EXP	M
MF151	CATTLE_BEEF	7755
MF151	HORSE	1777
MF151	PIGS	9596
MF151	POULTRY	12626
MF151	SHEEP_TOTAL	18236
MF151	01012100		353
MF151	01012901		316
MF151	01012909		25
MF151	01061909		0
MF151	01069000		0
MF154	ULL			616

Weights for 2020 for material in DP which came fro the DF node

mfa_cd	tpe	DPO	F	M
MF152	CATTLE_BEEF	4884	2790
MF152	HORSE	1312	437
MF152	PIGS	4594	4761
MF152	POULTRY	4191	8435
MF152	SHEEP_TOTAL	10682	7180
MF154	CATTLE_BEEF			82
MF154	HORSE			28
MF154	PIGS			240
MF154	SHEEP_TOTAL			374

Eggs and Milk from DF

Information is taken from the agricultural production base. Here, eggs are generally expected to go directly to the food category. Milk, however, goes partly into cheese production, which removes a large proportion of water (whey) from the milk.

In general, 10 kg of milk can be used to produce 1.4 kg of cheese and 8.6 kg of cheese whey. The cheese whey can be used to produce whey cheese, but in this production it can be assumed that about 9 kg of water will go out per 10 kg whey. Assuming that butter is about 95% fat, about 100 kg of milk is needed to produce 4.3 kg of cream. True, this production is processed by skimming the cream out of the milk (which is then skimmed milk) and a part of the cream (36% fat content) is then used to produce cream. Information on the sale of dairy products was obtained from two large stores in Reykjavík to assess the division of sales of dairy products. Weight figures were then used to calculate the “switch-matrixu” from milk to these products, by assuming that it is possible that for some of the water (whey) from the milk, it will go to the DPO node in the production. This method is somewhat unchecked and it would be better to get information from the producers themselves in order to get closer to the actual division of raw materials into products.

Weight of material and division for 2020 for eggs and milk in the DP node

mfa_cd	tpe	DPO	F	M
MF153	EGG	341	3834	85
MF153	MJOLK	46648	108846

Aquaculture

Aquaculture is the fastest growing agriculture in Iceland. Most of the product is exported and aquaculture is the only exporter of live fish from Iceland. Analysis of the product, however, is unreliable from the commercial base, but figures from the operation and economic overview of aquaculture are useful here for assessing fish production. Statistics Iceland has good data on production in aquaculture from 2010 onwards. Here, this product is placed in class F141, which has then become a different definition from the MF141 class of EW-MFA. Apart from being from a different source from other fish statistics, the same factors are used in the collection of food for this product as used for fish processing.

Fish processing

Fish processing can be roughly taken as a extraction of water and bones from the fish. The amount of water that is removed depends on the processing method and the type of fish, but it also plays into the fishing season here, the closer the spawning of the fish is, the looser it is and the utilization becomes worse.

In the Directorate of Fisheries’ database, the following information is accessible

• Year and month when the fish were landed
• Location of landing
• Fishing area of the fish
• Disposition of the catch - this is the use intended for the fish when it is received by the buyer. This isn’t necessarily a real processing route that’s gone at the end, but it still gives a pretty good picture of what happens to the catch.
• Fishing gear
• Characteristics of a boat that landed the catch - this is not necessarily the same as the vessel that caught the fish as the fish can be transhipped
• Type of fish

It is also recorded whether the boat landed by-products, such as liver, crushing, saddles, etc. The species of fish is not necessarily correct for this product.

All in all, there are about 14000 different combinations of species, fishing gear, locations, fishing area and disposal. These are obviously way too many combinations, so key combinations are put together based on information obtained from MAST. A Random Forrest sorting algorithm is then used to find the “most likely” utilization factor. This model is then reviewed and trained until the calculated output is comparable to the export figures when this information is known and domestic consumption of the product is known to be negligible.

This approach to fixed offers modeling that is slightly more complex than for other products

Flow of information in fish modelling

The model has three decision points

1. Losses occurring during the slaughter of the fish (evisceration object) (\(\gamma(i)\)) The value of this constant is generally around 91 % of the weight of the fish, but can be 100 % for non-gutted fish (e.g. capelin)
2. Filet-yield (\(k_{ij}\)). This is a matrix value that specifies how much food products the processing in question is expected to yield in export figures. Thus, the value is high when the fish is disposed of “iced during flight”, as this means that the fish leaves the country unprocessed. The measure “in progress”, however, means the production of skinless boned fillets, unless no fish of this species have been exported in this form.
3. Trimming utilization (\(h_{ij}\)). This matrix value estimates how much flesh is on a bone garden and in thin ends up as bruising or canned. This is generally a relatively small part, as they are fleshy parts of the fish and the constant is partly attacked by \(k_{ij}\)

Constants for most common types can be found in the following data sources

1. Fiskhandbók 2. kafli
2. fiskhandbók 3. kafli
3. Skýrsla MATÍS á bættri nýtingu
4. Ferskfiskhandbók MATÍS
5. Fiskbók MATÍS

Quantity of landed fish

This value is calculated using the gutting fraction of the fish:

\[ m_\textrm{landed} = \gamma_j \cdot m_\textrm{catch} \] Common values for this are:

• Blood: 5%
• Guts: 10%
• Head: 21%

According to Fiskistofa the gutting fraction of common species are

• Cod, Haddoc, Whitling: \(\gamma\)=0,84
• Ling, Blueling, Pike: \(\gamma\) = 0,8
• Sole, Skate, Halibut: \(\gamma\)=0,92
• Herring, Capelin, Mackerel: \(\gamma\) = 1

This assumes that the fish is brought to shore with head-on (or heads not discarded to the sea). If the heads are absent, the gutting fraction is reducted by up to 20%. This means that the value to DPO is:

\[ m_{\textrm{discard, ocean}} = (1-\gamma_j) \cdot m_\textrm{catch} \]

Filet yield by processing

The species and disposition of fish are used here as a proper assessment of the utilization of the fish. On top of that, the determination may need to vary by season, while the third measure is the fishing area of the fish (distance from land). In a few cases, the fish undergo canning or salting, which may mean that the actual weight of the product is higher than the calculated flesh weight. The quantity of food produced shall be:

\[ \begin{array}{rcl} m_\textrm{food} &=& k_{ij} \cdot m_\textrm{landed}\\ &=& k_{ij} \gamma_j \cdot m_\textrm{catch} \end{array} \]

\(k_{ij}\) ranges from 0 (not usable food fish) to 1 (whole-frozen for export). In reality, only about 40% of the total weight of fish is meat that is eaten. However, it is a tradition in many areas to have heads and spine on display on the dinner table, so if consumers happen to wonder if the fish are enjoying the company. Producers in Iceland, however, are usually more interested in finishing the fish into cartons (skinless, boneless or red, boneless) as such a product is more valuable per unit transported as well as producing valuables from the heads, bone garden, skin and viscera of the fish.

The utilization of fish also depends only on the size and flesh density of the fish. Larger fish seem to have slightly better flesh utilization and fish from cold seas have denser flesh.

Cutoff utilization

This processing often generates value from the products, such as fish puree, fish soup and similar food. In addition, gelatin, skin and bone gardens are useful after drying, but this product is not considered food in this analysis.

The quantity here is:

\[ \begin{array}{rcl} m_\textrm{reprocess} &=& m_\textrm{landed} - m_\textrm{food}\\ &=& \gamma_j \cdot m_\textrm{catch} - k_{ij} \gamma_j \cdot m_\textrm{catch} \\ &=& (1-k_{ij})\gamma_j \cdot m_\textrm{catch} \end{array} \]

Here it is assumed that cutoff yield is better in land processing than when the product is processed at sea, as factory trawlers have somewhat less space to store goods of less value if otherwise caught. Here we also assume that heads, bone gardens and skin will not be food despite processing and end up as material (M). Material production is therefore:

\[ \begin{array}{rcl} m_\textrm{non-food} &=& h_{ij} \cdot m_\textrm{reprocess} \\ &=& h_{ij} (1-k_{ij}) \gamma_j \cdot m_\textrm{catch} \end{array} \] The value for \(h_{ij}\) is here nearly the same value as the weight ratio of protein and fat in the trimmings of the fish. The water content of this trimming is quite low (about 15%) so shrinkage is quite low. The amount of material removed is therefore calculated as:

\[ \begin{array}{rcl} m_\textrm{DPO re-processing} &=& m_\textrm{reprocessed} - m_\textrm{non-food}\\ &=& \left((1-k_{ij}) \gamma_j \cdot m_\textrm{catch} \right) - \left( h_{ij} (1-k_{ij}) \gamma_j \cdot m_\textrm{catch}\right)\\ &=& (1-k_{ij})(1-h_{ij}) \gamma_j \cdot m_\textrm{catch} \end{array} \]

The total amount spent in the DPO has therefore become

\[ \begin{array}{rcl} m_\textrm{DPO} &=& m_\textrm{DPO ocean}+ m_\textrm{DPO re-processing}\\ &=& \left((1-\gamma_j) + (1-k_{ij})(1-h_{ij}) \gamma_j \right) \cdot m_\textrm{catch}\\ \textrm{or}\\ m_\textrm{DPO} &=& m_\textrm{catch} - (m_\textrm{food}+m_\textrm{non-food})\\ &=& \left(1 - k_{ij} \gamma_j + h_{ij} (1-k_{ij}) \gamma_j \right)\cdot m_\textrm{catch}\\ &=& \left(1 - \gamma_j(k_{ij} - h_{ij}k_{ij} + h_{ij}) \right)\cdot m_\textrm{catch} \end{array} \]

These sizes are finally all combined in order to get the final division of the fish into food (F), product (M) or waste (DPO)

The final result is dataset with 101260 lines that have information in a much more detailed format than is necessary for people with moderate curiosity

Part of information for 2020 (10 lines)

year	mfa_cd	is_mfa_data	entry_node	exit_node	entry_sm	exit_sm	value_tons	source_descr	source_line
2020	MF141	TRUE	DE	DPO	SM_RAW	WASTE	817.35312	DE|THORSKUR|FRYSTIHUS:FLOKUN|DPO	model.fish.processing|1478
2020	MF141	TRUE	DE	DPO	SM_RAW	WASTE	16195.11696	DE|THORSKUR|FRYSTIHUS:FRYSTING|DPO	model.fish.processing|1478
2020	MF141	TRUE	DE	DPO	SM_RAW	WASTE	108.90080	DE|THORSKUR|FRYSTIHUS:HARDFISKUR|DPO	model.fish.processing|1478
2020	MF141	TRUE	DE	DPO	SM_RAW	WASTE	4.00640	DE|THORSKUR|FRYSTIHUS:HERSLA|DPO	model.fish.processing|1478
2020	MF141	TRUE	DE	DPO	SM_RAW	WASTE	233.94416	DE|THORSKUR|FRYSTIHUS:INNANLANDSNEYSLA|DPO	model.fish.processing|1478
2020	MF141	TRUE	DE	DPO	SM_RAW	WASTE	11127.44208	DE|THORSKUR|FRYSTIHUS:ISAD_I_FLUG|DPO	model.fish.processing|1478
2020	MF141	TRUE	DE	DPO	SM_RAW	WASTE	0.00272	DE|THORSKUR|FRYSTIHUS:REYKING|DPO	model.fish.processing|1478
2020	MF141	TRUE	DE	DPO	SM_RAW	WASTE	6087.36336	DE|THORSKUR|FRYSTIHUS:SJOFRYSTING|DPO	model.fish.processing|1478
2020	MF141	TRUE	DE	DPO	SM_RAW	WASTE	6206.19408	DE|THORSKUR|FRYSTIHUS:SOLTUN|DPO	model.fish.processing|1478
2020	MF141	TRUE	DE	DPO	SM_RAW	WASTE	870.98560	DE|THORSKUR|FRYSTIHUS:UNNID_I_GAM|DPO	model.fish.processing|1478

This data is somewhat turf-read, so aggregates are calculated by omitting information about the invoice line itself (source_desrc):

Calculated totals (10 lines)

year	mfa_cd	entry_node	exit_node	entry_sm	exit_sm	value_tons
2020	MF111	DP	F	SM_RAW	SM_FIN	7287.0000
2020	MF111	F	EXP	SM_FIN	SM_FIN	89.0570
2020	MF111	IMP	F	SM_FIN	SM_FIN	34202.1200
2020	MF111	IMP	F	SM_SFIN	SM_SFIN	45.2280
2020	MF111	IMP	IP	SM_RAW	SM_RAW	2459.0240
2020	MF111	IMP	M	SM_FIN	SM_FIN	645.8530
2020	MF111	IMP	M	SM_RAW	SM_RAW	18288.9650
2020	MF111	IP	DPO	SM_RAW	WASTE	983.6096
2020	MF111	IP	F	SM_RAW	SM_FIN	1475.4144
2020	MF111	M	EXP	SM_FIN	SM_FIN	8.4160

This returns 5153 lines. This is where entry_sm and exit_sm definition gets in the way. This is primarily useful information in the processing of the data, omitted to the final version of the data.

Totals without SM designation (10 lines)

year	mfa_cd	entry_node	exit_node	path	value_tons
2020	MF111	DP	F	DP-F	7287.0000
2020	MF111	F	EXP	F-EXP	89.0570
2020	MF111	IMP	F	IMP-F	34247.3480
2020	MF111	IMP	IP	IMP-IP	2459.0240
2020	MF111	IMP	M	IMP-M	18934.8180
2020	MF111	IP	DPO	IP-DPO	983.6096
2020	MF111	IP	F	IP-F	1475.4144
2020	MF111	M	EXP	M-EXP	243.0380
2020	MF1110	F	EXP	F-EXP	123.7630
2020	MF1110	IMP	F	IMP-F	4548.3940

One year from this data, which is 3416 lines for the year 2021 are shown in the Sankey graphs above.