Saturday, May 02, 2015

Interpreting the Water Footprint of Food

I've read some discussions of the LA Times Water Footprint of Food interactive feature, mainly expressing surprise about the comparatively large water footprint of pulses, dried legumes.

When you look at these plots, a beef burger looks only slightly indulgent compared to a falafel (chickpeas).  What's the harm in a burger?  Or the Atkins diet in general?
Water footprints of protein sources calculated by the
I scratched my head when I saw that graphic as I understood that beef is a highly inefficient source of protein compared to plant sources (but several magnitudes more efficient than tuna!).  How was this calculated?  Is the methodology valid?

This weather and climate data consultant dug deeper.

The folks at The Water Footprint Network explain their methodology here.  The LA Times article said, "Below you can see the U.S. water footprints of selected foods as measured by gallon of water per gram of protein produced or per calorie."

So is this country-specific data?  How can you compute the water-intensity of the US chickpea crop when there is nearly no commercial chickpea production in the US?

Later, in the LA Times coverage, they quote Mekonnen and Hoekstra:
"The average water footprint per calorie for beef is 20 times larger than for cereals and starchy roots," they note, referring to global averages, not U.S.-specific figures. "The water footprint per gram of protein for milk, eggs and chicken meat is 1.5 times larger than for pulses," a group of legumes that includes peas, beans and lentils.
Perhaps the LA Times reported GLOBAL (not US) water intensity in US units of gallons? The article wasn't clear.

I searched and found that the EU compiled and mapped some statistics they downloaded from McGill University.

Global acreage used for chickpea production.

Global production of chickpeas in tons per square km.
If the LA Times reported US statistics, then the chickpea data is highly suspect.  It's the classic "statistics of small numbers" problem.  The smaller the sample size, the more variable and unreliable the statistic.

Suppose the LA Times did report the GLOBAL water footprint, then it is important to look at the hydrologic cycle of the areas where chickpeas are farmed.

Luckily, I work at a weather and climate data archive and have access to stuff like this classic paper about the terrestrial seasonal water cycle by Willmott and Rowe.  See and download the Willmott and Rowe data.  I give you permission to play with your food data.

First page of Willmott and Rowe.

Recall that chickpeas are mainly grown in India, the middle east and sub-Saharan Africa.  Hmm, look at the evaporation in those regions in the early summer.  If most of the chickpeas are grown in India, and ground-level evaporation is exceptionally high there (and moderately high in other regions where chickpeas are also grown), then the global water footprint for chickpeas will be high.

Evaporation climatology 1950-1979.  Notice the extremely high evaporation in chickpea-producing areas of India and central America during monsoon season. 
Did the LA Times report rely on a small and unreliable dataset (US chickpea production) or report global statistics and label them as US statistics? I don't know. Either way, their reporting is extremely suspect.

People need protein.  Plant-based proteins such as chickpeas are largely grown and consumed in regions where the resources (land, water, labor) cannot support animal sources of protein.

The water and carbon intensity of crops vary greatly by location.  That's why I don't eat an entirely locavore diet.  Our family enjoys CSA boxes grown with reclaimed water from the Irvine waste treatment plant.  But, we occasionally eat lamb chops imported from New Zealand, where the sheep are raised on rainfall-watered pasture.

Yes, sheep meat has a large water footprint.  But New Zealand has abundant rainfall and doesn't need to artificially irrigate their pastures.  If the lamb is frozen and then shipped via ultra-efficient container ships to a harbor < 15 miles from our home, then the carbon footprint of that lamb chop is much, much lower than a beef burger from the Central Valley of CA.

Early summer evaporation in the American west.  Note the hot spot in the Fresno area.
I took a closer look at the evaporation in California.  Download the data and the Panoply data viewer and play with the data yourself.  Scroll through the months.

The Willmott and Rowe data is based solely on 1950-1979 ground-based station data.  Later datasets rely on satellite data, but WR gives monthly averages, which is important because row crops are not grown year-round.

It's important to note that the evaporation measured by the ground stations are influenced by temperature, winds and water availability.  Water availability depends on both natural sources, e.g. precipitation and surface stream flow, and irrigation.

See that bright yellow hot spot near Fresno, California?  100 kg per square meter is 100 mm or ~4" of water.  Look at Fresno's climatology.  That's nearly all irrigation with water from elsewhere or  groundwater.

This is old data.  The numbers today, with millions of acres planted with permanent tree crops that can't be fallowed during droughts, would be even more scary.

This is why California's central valley is sinking.  This is a slow-motion environmental catastrophe.  It has to stop now.


  1. SanMarie13:19

    That chick peas would be a bigger environmental issue than beef simply makes no sense. Thus it is so helpful that you have dug deeper into the data. Even using their sources, however, chick peas look better (on this factor of the total concern) for calories-vs. protein-than beef. So firstly, calories are necessary as well as protein is, and secondly, might their findings be partly based on the protein efficiency of beef vs. plant protein? In that case a fair comparison would have to include the available protein utilization increase provided by smart protein combining, which is to say, largely traditional plant protein combining, as in rice with pulses, for example.

    1. In the 1980s, I read a report that said that it takes 23 pounds of plant protein (soybeans) to produce 1 pound of beef protein. I was an undergrad then and didn't delve deeply into the methodology. I do remember a hierarchy that pork took half as much resources of beef and chicken was half as much as pork.

      You are right, that combining rice and pulses provides a complete protein on a plate. I grew up with soy and enjoy it in many forms. Between soy and the eggs and milk I consume, I don't worry too much about combining plant proteins.

  2. Anonymous06:54


    It's things like this LA Times article that confuse the bejeebers out of most of us regarding climate change and the water situation in the southwest, and the world in general. Then add in the rebuttals from all the commercial/political interests.....And top it all off with the the reality that most of us have NEVER had education in critical thinking & statistics interpretation (aka "teach to the test" in schools now). And apathy, at least until it hits consumer pocketbooks, like when gas hit $5-6 (or more) a gallon a few years back. We can live without burgers, but when the faucet is turned on an nothing drinkable, or nothing at all, comes out, that will be disastrous.