Our friends of the HarvestChoice team at IFPRI have been busy. Hot on the heels of MAPPR, comes news that their Spatial Production Allocation Model (SPAM) will be used to produce a Global Yield Gap Atlas (GYGA), which will “reveal the ‘gap’ between the current average yields of farms and their maximum production potential.” Sounds very useful. We have blogged about SPAM before. I was particularly intrigued by this statement in the IFPRI post on the subject, though:
At a recent GYGA meeting in Naivasha, Kenya, Atlas collaborators—which include Jawoo Koo of IFPRI—comparatively reviewed two major crop distribution maps and announced that they would use the ones produced by an IFPRI model—the Spatial Production Allocation Model (SPAM)—as a basis for the Atlas.
It would be interesting to know what the other lot of crop distribution maps were, the ones that were found wanting. One of our earlier posts does try to get to grips with the taxonomy of crop mapping, not particularly comprehensively, it has to be said. So was it perhaps CIAT’s Crop Atlas of the World? ((Which as you can see if you click on that link does not exist as a separate thing any more, but the components of which you can find on the Harvest Choice data catalogue.)) Or was it the dataset of Monfreda et al. (2008), “Farming the planet: 2. Geographic distribution of crop areas, yields, physiological types, and net primary production in the year 2000”? Or something else that we missed at the time? Maybe HarvestChoice/IFPRI are too modest to say, but it would still be good to know the basis for GYGA’s decision.
The other source was possibly this one:
http://www.geo.uni-frankfurt.de/ipg/ag/dl/forschung/MIRCA/
Which is similar to the Monfreda maps. I would also be curious to know the criteria. I found Monfreda and SPAM were ‘better’ in different places when I was reviewing the legumes but still there are large areas of confusion between, for example, common beans and cowpeas.
Luigi and Andy,
Here’s my response based on my personal experience at (and after) the workshop.
Andy’s correct (as usual) – the other dataset under consideration was indeed MIRCA2000 (that is based around a structured disaggregation of the Monfreda et al. harvested area for each crop into rainfed and irrigated areas).
The current phase of the GYGA project covers major cereal crops in 12 developing countries (10 in SSA + 2 in SA), and the Naivasha workshop had participants (selected crop experts/agronomists) from all of the countries. The workshop organizer distributed MIRCA’s crop distribution maps to each country expert and asked for feedback on whether the maps adequately represented where the crops are (or aren’t) in terms of harvested area. Reviewing all the feedback comments and comparing the results against the SPAM data, the organizers established that “… In nearly all cases, the SPAM maps were more consistent with the feedback we received from the GYGA agronomists at the workshop.”
However, it should be noted that SPAM maps were assessed to be more reliable based on the participants’ subjective feedback at the workshop and for the *specific* purposes of this project, and that the same participants were also able to provide useful suggestions as to where SPAM maps have room for improvement.
(Full disclosure: HarvestChoice has a collaborative MOU with GYGA with respect to data, tools and output applications.)
I hope that helps answer your questions.
Jawoo Koo