How Blue Gold works it out
Every figure on both maps, step by step. Sources and their dates are on the Data page.
Rafts
400to a hectare of favourable water
12 × 12 ft bamboo rafts of Kappaphycus alvarezii, the seaweed nearly all Indian farms grow. ICAR-CMFRI, 2023.
Harvest
30.4 tdry, a hectare a year
After keeping back seed for the next crop.
Income
₹18.9 lakha hectare a year, net
At ₹95 a kilo dried, once rafts, seed, labour and capital are paid. Work for about 9 families.
Carbon
22.2 tCO₂ taken up a year
Sold as food or carrageenan, as nearly all of it is today, none of it stays out of the air for long.
Where a parcel can go
the gold band on the map, and only water inside it counts
The band
ICAR-CMFRI surveyed India's coast and named the sites where the water suits seaweed: calm, shallow sea close to shore. The band is sea within 1 km of the coastline and 3 km of a surveyed site, with land cut out and pieces under 5 ha dropped.
What it adds up to
1,12,140 hain 108 stretches of water around 264 sites
More than the 23,970 ha CMFRI itself counted as potential, because the band marks where you can draw, not how much of it is good. What you draw is clipped to the band; the part outside turns orange and is not counted.
How exactly
The low-tide line is not mapped for all of India, so the high-water line from OpenStreetMap is used, which keeps the band on the shoreward, safer side. Land is cut out with OpenStreetMap's land polygons, accurate to about 10 m. Water far from any site may be fine, but nobody has checked it.
One hectare, step by step
every figure on the map is this chain, times the hectares you drew, scaled by the yield slider
The chain
Prices and costs are CMFRI's, in 2023 rupees. The map rounds rafts down to whole ones; this chain does not.
When it pays
Net income, ₹ lakh a hectare a year, by price and by yield against CMFRI's reference farm (outlined). At ₹95 a kilo a farm breaks even at 35% of the reference yield; at ₹50 it needs 66%.
Three carbon numbers, never added
the honest answer depends on what happens to the harvest
Taken up
22.2 tCO₂ a hectare a year
The carbon built into the crop. Real and measurable. This is the figure usually quoted as "sequestration".
Stored
The part still out of the air after 100 years. It depends entirely on the end use.
Avoided
noneclaimed
Emissions saved when seaweed replaces something else. Blue Gold gives none, because it would need a named substitution to count.
Biochar: 35% of the seaweed's carbon ends up in the char, and 65% of that lasts a century (IPCC 2019, low-temperature char). Compost is an order-of-magnitude figure.
What it can't tell you
- Whether your water suits seaweed this year. A site still needs checking: salinity 28–38 ppt, 26–31 °C, water clear to 2–6 m, shelter from heavy waves.
- Whether you may farm there. Permissions come from the state fisheries department; near the Gulf of Mannar's reefs, only native species within the biosphere reserve's rules.
- Disease, cyclones, theft or grazing fish. The reference farm is a good year on a good site.
- Other species or methods. The model is Kappaphycus on bamboo rafts; monoline and tube-net farms are counted in raft equivalents.
The map and this page run the same model; change a parameter and both change.
The rule, in order
Is a boundary published for the site?
if not Not placed preciselyPlanted 4 years ago or more, and the mangrove share has risen by 20 points or more since the year before planting?
if so Canopy gainedDid the record already read 75 % or more of the boundary as mangrove in the year before planting? Asked only of boundaries of 10 squares or more.
if so Already mangrove before plantingIs a planting year reported?
if not Planting not yet reportedWas it planted 4 years ago or more?
if not Too young to tell if so No change seen yet
What is counted
The share of 30 m squares touching the boundary that the yearly record reads as mangrove. Where no planting year is given, the latest year stands in for step 3.
What each status means, and does not
Means Mangrove share inside the boundary rose by 20 points or more after the planting year.
Does not mean That the planted saplings are the trees now seen. Mangroves can also spread on their own.
Means The record already read 75 % or more of the boundary as mangrove before the planting year.
Does not mean That nothing was planted, or that anything is wrong. Planting can be into gaps and thin stands, and a boundary may be drawn round a whole creek.
Means 4 years or more after planting, the share has not risen by 20 points.
Does not mean That the planting failed. Sparse or short stands can go unread at 30 m, and slow coasts take longer.
Means Planted under 4 years ago.
Does not mean Anything at all. The record is not read for these sites.
Means The department lists the land and its boundary; no planting year is given.
Does not mean That planting is late. The record simply has no year to count from.
Means The place is named by village, landscape or tract, with no boundary.
Does not mean That nothing is there. Without a boundary the record cannot be read for it.
Checked by eye
before any site was drawn
Looked at
12sites, before and after
Every site the rule speaks for, and every borderline one, in 10 m pictures from before planting and from 2026, with greenness measured inside the boundary.
Canopy before planting
5of 6 confirmed
At Ghansoli and Koparkhairane the pictures show closed canopy across each boundary in 2019, a year before planting. At Gorai about half the boundary was open.
Rule tightened
75 %was 50 %
Three boundaries the record read as 50 to 62 % mangrove could not be confirmed as standing mangrove in the pictures. They no longer carry that status.
Gain
2of 2 confirmed
At Roadpali greenness inside both boundaries rose sharply between 2023 and 2026. One of the two was already half green before planting.
Counting the coast
and where our count parts from the Forest Survey's
The record
41years, at 30 m
Global Mangrove Watch marks each 30 m square mangrove or not, every year from 1985 to 2025. We add the squares up with each square's own area. No correction is applied for the record's stated error.
Stands off the shore
Gujarat, 2023, sq km. Much of its mangrove stands on tidal flats and islands outside the drawn district outlines, so each square goes to its nearest district within 25 km.
Kachchh: the two counts move opposite ways
The line is the satellite record: 191 sq km in 2013, 419 in 2023. The open dots are the Forest Survey: 780, then 708. Ours is lower and rising; theirs is higher and falling. Both are shown.
A neighbour's shore
157sq km in Kachchh's border creeks
Squares nearer Bangladesh's, Pakistan's or Sri Lanka's land than ours are left out. In the creeks at Kachchh's western edge the line itself is contested, so this part of the count is the least certain. Country outlines: geoBoundaries.
Carbon: the working
one sum, done two ways
The sum
District by district, then added. With the IPCC's defaults in the middle box it comes to 198.
Where the soil was dug
58 cores behind the soil map; 6 reach a metre. Elsewhere it infers.
Dry coast, wet coast
Dry under 1,000 mm of rain a year, by ten years of rain.
IPCC defaults used
- carbon in dry matter 45.1 %
- roots to shoots 0.29 dry · 0.49 wet
- mature stand, t per ha 92 dry · 192 wet
- growth, t per ha a year 3.3 dry · 9.9 wet
- dead wood and litter 11.4 t C per ha
- soil, default 286 t C per ha
- soil gain 1.62 t C per ha a year
The planting sum
- tonnes hectares × share that takes × (trees and roots + soil), over the years chosen
- soil gain 3.5 t C a hectare a year on former mangrove ground; 1.6 on new ground
- to plant ₹3.28 lakh a hectare: the median of 63 Maharashtra sites (₹1.94 to ₹3.62 lakh for the middle half)
- would fetch tonnes × $13 to $35 × ₹96.61 a dollar (9 Oct 2026)
- where 30 m squares the yearly record once read as mangrove and no longer does, that a 10 m land-cover map of 2021 reads as open land; water, buildings and standing trees are left out
Not in the sum: upkeep, verification, land, and how much reaches the village.
Where the rule is weak
- Small sites. 22 of the 104 boundaries cover fewer than 25 squares; one misread square moves their share by four points or more.
- Slow coasts. On arid flats canopy can stay sparse for eight years or more and go unread.
- Loose boundaries. A boundary drawn round a whole creek carries the creek's old mangrove into the share.
- One State. Only Maharashtra publishes boundaries; the record can speak for 10 sites so far.