Space IntelligenceSpace Intelligence
  • About us
  • Our Technologies
  • Clients
  • Blog
  • Contact
  • Services
  • Why Us
  • Consulting

Blog

  • Home
  • Blog
  • Blog
  • Forest loss maps “systematically underestimate deforestation”

Forest loss maps “systematically underestimate deforestation”

  • Categories Blog
  • Date 10th January 2018

A paper co-authored by Space Intelligence’s Chief Scientific Officer, Dr Edward Mitchard has shown how large scale forest mapping products can underestimate the level of deforestation occurring.

The team, led by Dr David Milodowski of the University of Edinburgh’s School of GeoSciences, compared three forest loss products over Brazil: Global Forest Watch (30m resolution); PRODES (60 m resolution) and FORMA. They compared these products with the results of their own analysis. This involved taking the multi-year differences in classified RapidEye satellite images. The RapidEye satellite has a spatial resolution of 5m, and as such was used as the ‘truth’ dataset.

The team found that broadly the general spatial patterns of change were consistent across the products. However, the marked differences arose when taking into account the size of the forest disturbance being mapped. Smaller clearances of less than 2 hectares were less well mapped by the wall-to-wall products from GFW and PRODES. This suggests that large scale mapping using these products, especially in areas where forest change is dominated by small scale clearances, will result in significant negative bias, or underestimation of deforestation rates.

In order to account for these differences in ability to measure small scale deforestation, the authors suggest that analysts apply a correction factor of +25% for the GFW product, and +50% for the PRODES product, in order that the results can be more directly comparable with those achieved through higher resolution mapping at 5m.  In areas where deforestation is dominated by large scale clearances, the authors suggest that a smaller correction factor of 5% for GFW and 15% for PRODES should be applied. Yet they counsel caution in the uncritical application of such correction factors, since these results are preliminary.

Dr Milodowski said:

“Our study shows that the performance of satellite-based deforestation detection depends on the style of deforestation occurring in a given area: small disturbances and degradation are much harder to detect than large clearings. In turn, these reflect the underlying drivers of deforestation. This also makes it challenging to generalise correction factors across areas with different forest characteristics and different drivers of deforestation, but in regions where small scale clearance and degradation dominate underestimates are likely to be particularly significant.”

The authors conclude by noting that that the differences between products are not errors as such, rather they reflect differences in forest definitions and aims of different monitoring systems. At Space Intelligence we are able to advise on the best solution for your monitoring requirements, and have provided some more information on the differences between our own forest monitoring product and the GFW product here: Why Use Space Intelligence to Monitor Forests?

Reference

Milodowski, D.T., Mitchard, E.T.A. & Williams, M. 2017. Forest loss maps from regional satellite monitoring systematically underestimate deforestation in two rapidly changing parts of the AmazonEnvironmental Research Letters, https://doi.org/10.1088/1748-9326/aa7e1e

 

 

Tag:accuracy, acre, brazil, carbon, climate change, deforestation, forest monitoring forest degradation, GFW, Global Forest Watch, monitoring, mrv, prodes, rhondonia, satellite forests

  • Share:
1spac3murray1

Previous post

Space Intelligence scientist use satellites to map world 's largest peat deposit, in central Africa
10th January 2018

Next post

Distinguishing intact forest from degraded forest in Cameroon
19th February 2018

You may also like

Rapid forest loss around Medco Papua Hijau mill in Papua, the beginnings of traceability (Papua palm oil production Part 1)
4 May, 2018

Deforestation for palm oil production has again received international attention in light of allegations that plantation linked to major household names has been encroaching on protected peatland area. We thought this would be a good opportunity to look at the …

Unilever palm oil mill locations and comparisons to forest change
26 April, 2018

Earlier this year, Unilever decided to release the locations of all of its mills used for palm oil production. The locations of 1431 mills were disclosed. This is very useful information for scientists as these locations can be compared to …

Unilever releases locations of palm oil mills in a bid for supply chain transparency
12 April, 2018

In February 2018, Unilever released details about its palm oil suppliers and mills in a bid to makes its supply chain more transparent. In what they call a ‘radical step’, they are one of very few companies to make this …

Leave A Reply Cancel reply

You must be logged in to post a comment.

Archives

  • May 2018
  • April 2018
  • March 2018
  • February 2018
  • January 2018
  • July 2017
  • June 2017

+44 (0)7951 028647

enquiries@space-intelligence.com

Space Intelligence
Edinburgh Centre for Carbon Innovation
High School Yards
Edinburgh EH1 1LZ,UK

Latest News

Sierra Leone surveys
07Jul2017
Space Intelligence scientist use satellites to map world ‘s largest peat deposit, in central Africa
10Jan2018
Forest loss maps “systematically underestimate deforestation”
10Jan2018

Links

  • Our Technologies
  • Clients & Examples
  • About us
  • Contact & Order
  • Services
  • Why Us
  • Consulting

Copyright Space Intelligence | Website Design by Primal Space