Documentary photograph illustrating coffee Agroforestry Carbon
← Back to blog

Coffee Agroforestry Biomass: Why the Equation, the Diameter Height and the Plot Size Decide the Number

Sustainability•13 November 2026•TREEO Indonesia•5 min read

Shaded coffee looks like a simple measurement job. Count the shade trees, count the coffee bushes, apply equations. In practice three choices decide the final number: which equation you apply, where on the stem you measured the diameter, and how big your plot is compared with the satellite pixel. Each can move the result more than the field work itself.

Published figures worth remembering

1.6-2.8 tC/ha

coffee shrubs in Peruvian shaded coffee

27.5-57.5 tC/ha

shade trees in the same plots

0.23-0.33 to 0.67-0.75

R² of lidar-to-biomass fits as plots grew from 80 to 1,000 m² at two dry-forest sites

Two pools, two equations

A study in Nicaragua set the standard design: separate allometry for the shade trees and for the coffee. Shade trees were modelled on stem diameter at breast height, with adjusted R² between 0.71 and 0.97. Coffee bushes were modelled on stem diameter at 15 cm and height, with a correlation of 0.84. The two pools are very different in size. Coffee bushes carried 0.005 to 2.8 kg of above-ground biomass per plant, against 3.5 to 386 kg per shade tree. Wood specific gravity also differed: 0.52 for the trees and 0.82 for coffee.

In Peruvian coffee agroforestry the pattern held at stand level. Coffee shrubs stored 1.6 to 2.8 tonnes of carbon per hectare, shade trees 27.5 to 57.5, and tree roots 7 to 13.7. Calculated from the paper's table, coffee was roughly 5% of above-ground biomass carbon. For design, this means the shade-tree equation matters most for the total. Our post on measuring carbon in coffee agroforestry covers why the shade trees drive the project.

Choosing, not averaging

Equations built for forest trees do not suit a pruned shrub. For Robusta in Indonesia, one option is the Arifin 2001 equation, 0.281 times diameter to the power 2.06, listed in the ICRAF manual for regularly pruned coffee. A 2026 study in south-western Uganda fitted separate above- and below-ground Robusta equations on 60 plants, with 40 used to calibrate and 20 to evaluate. Its best above-ground model, built on girth at the base, reached R² 0.772.

For shade trees, the pantropical equation of Chave and colleagues (2014) uses wood density, diameter and an environmental stress term, and advises building a local height-diameter relationship to limit bias. Indonesia has no shortage of alternatives: one compilation gathered 2,458 equations from 168 destructive studies, and noted that they are skewed by region and species.

The risk is to run five equations and report a range or an average. Equations fitted on different plants, pools and diameter definitions are not five estimates of one quantity. A more defensible routine is to assign one equation per pool by fit domain, check it against a small destructive sample, and propagate its error. The BIOMASS package in R supports the error propagation. Chave and colleagues (2004) found that the choice of allometric model is the most important source of error in this kind of estimate, and a 2026 study looks at how the choice of allometric reference changes Sentinel-2 biomass estimates.

The diameter height trap

An equation is only valid for the diameter it was fitted on. The Nicaraguan coffee model uses diameter at 15 cm. The Ugandan Robusta model uses girth at the base. The Arifin equation appears in a manual that defines its diameter generically as "DBH", which cannot literally apply to a coffee bush, so the measuring height for coffee is not stated there. Collar diameters, 15 cm diameters and 30 cm diameters are not interchangeable.

Practical rules:

- Record the height at which each diameter was taken. - Match the equation to that height, or measure a paired set on a sample of bushes and convert. - Record stems per bush. Multi-stem bushes need per-stem measurement and summing, or an equation fitted on the same unit. - Treat diameters outside the range an equation was fitted on as extrapolation.

Plot size against pixel size

A 20 m by 20 m plot covers 4 Sentinel-2 pixels. A circular plot of 7 m radius covers about 154 m², which is under 2 pixels. Published work shows why that matters. In a study of tropical dry forest in Yucatán, the R² of lidar-to-biomass fits rose from 0.33 to 0.75 at one site (0.23 to 0.67 at the other) as plot size grew from 80 to 1,000 m², and GPS errors of 2 to 10 m degraded the fit further. A study of border effects tested plots of 10, 20, 50 and 100 m and found that 40% of the uncertainty at 20 m came from crowns crossing the edge, and none at 100 m. A review of field biomass used to calibrate remote sensing reports typical overall uncertainty of about 25%, with a range of 16 to 53%.

On slopes there is one more correction. Distances measured along the ground are longer than their horizontal projection. Per-hectare values should use the horizontal area of the plot, not the nominal one.

What satellites can and cannot see under shade

The sources reviewed for this article include no study that separates coffee from shade biomass at 10 m. A LiDAR study in coffee agroforestry used a mixed model with a height slope of 2.400 and found that one farm's slope was 3.065, 27.7% higher. Coefficients of variation were often above 100%. The study, from northern Peru, concludes that biomass models need farm-level calibration.

The reference matters as well. In a forest study using Sentinel-1 and Sentinel-2, models trained on field plots reached R² of 0.79 to 0.84, while models trained on a GEDI-derived biomass product reached 0.36 to 0.47. Coffee mapping is also uneven. A Dak Lak study reported 89.4% overall accuracy for coffee against non-coffee but 72.5% for a nine-class legend. A 2026 coffee-systems mapping study reported 89.9% overall accuracy. A Landsat study distinguishing shade coffee reported 82.1% user's and 80.0% producer's accuracy.

Frequently Asked Questions

There is no single answer in the sources. The Indonesian manual equation and the 2026 Ugandan equations are candidates, and each is tied to a diameter definition. Pick one per pool and validate it on a small destructive sample.

Turn uncertainty into a plan

Sampling design and growth assumptions decide how much measured carbon survives conservative deduction. TREEO Carbon Simulator models sequestration scenarios and biomass growth before the field season.

Related Articles

Coffee Agroforestry Biomass: Equation, Diameter, Plot Size | TREEO Indonesia