Frédéric Achard · Global Change Biology 2014 · Remote sensing longitudinal sample survey and biomass modeling · n=4000 sample units (10 × 10 km)

Determination of tropical deforestation rates and related carbon losses from 1990 to 2010

Cited 578 times in the scientific literature.

Level 3 - non-randomized controlled study

Level assigned by design analogy (non-clinical): longitudinal observational sampling of remote sensing data across two decades.

OpenAlex W2094738700 · doi:10.1111/gcb.12605 · record verified 2026-08-30

What was done

The authors estimated tropical forest cover changes (deforestation and regrowth) across two decades (1990–2000 and 2000–2010) using a systematic sample of 4,000 units (10 × 10 km each). Forest cover was interpreted from satellite imagery at 30 × 30 m spatial resolution. These area change estimates were integrated with pan-tropical biomass maps to calculate gross carbon losses and carbon removals from forest regrowth.

What was found

Gross tropical forest loss was 8.0 million ha yr⁻¹ in the 1990s and 7.6 million ha yr⁻¹ in the 2000s (annual rate of 0.49%), representing no statistically significant difference between decades. Humid forests accounted for 64% of total forest cover in 2010 and 54% of net forest loss in the 2000s. Carbon losses from forest and other wooded land were estimated at 887 MtC yr⁻¹ (range: 646–1,238) in the 1990s and 880 MtC yr⁻¹ (range: 602–1,237) in the 2000s, with humid regions contributing two-thirds. Carbon removals from regrowth were estimated at 115 MtC yr⁻¹ (range: 61–168) in the 1990s and 97 MtC yr⁻¹ (range: 53–141) in the 2000s, which is 5 to 15 times lower than prior published estimates. Deforestation carbon losses equaled approximately 10% of global fossil fuel and cement emissions during 2000–2010.

Why it matters

This study provides a consistent, empirical satellite-based baseline showing that tropical deforestation rates and associated carbon emissions did not decline between the 1990s and 2000s. It also indicates that carbon sequestration from tropical forest regrowth is substantially lower than previously estimated.

Limits

The study relies on statistical sampling rather than continuous wall-to-wall mapping. The forest area change estimates conflict significantly with national survey data. Biomass maps used to convert area change to carbon fluxes carry inherent calibration and spatial uncertainties, and sub-pixel forest degradation below the 30 × 30 m resolution was not directly quantified.

Cited by