This past spring, the months of April and May in Beijing were reported as the cleanest April / May in a decade. (Sources for April: China Daily and Beijing EPB, see also my previous post; sources for May: Xinhua and Beijing EPB.)

Independent of the quantitative results, these reports seem to confirm what a lot of people have been mentioning to me, that this past spring has seemed surprisingly clean.

Let’s evaluate the truth in all of this. The following graph shows average API*, average PM10 concentration, and number of Blue Sky Days for the period April-May from 2005 through 2009:

From these numbers, the results are pretty clear: the period April-May 2009 in Beijing was indeed significantly better in terms of air quality than the same period in any of the previous four years. (I could have looked farther back, but I decided only to look at five years total for this analysis.)

Here are some comparisons of 2009 vs 2005-2008 averages:

I think it’s probably fair to say that the air quality this April and May was 30-40% better than the average air quality during the same period over the previous four years.

As usual, we should celebrate the progress while being mindful of the significant improvements still required. My calculated average PM10 concentration for this period, 117 ug/m^3, is still well above China’s annual target (100 ug/m^3) and well well above the WHO’s ideal target for developed nations (20 ug/m^3). It is also well above my estimate for the average PM10 concentration during last year’s two-month Olympic period (79 ug/m^3).

Related:
Summary of Beijing’s 2009 first quarter air quality
Summary of Beijing’s 2008 air quality
Update on fall air quality in Beijing

*I don’t really like averaging API, because it can lead to some misleading results (further discussion in this post), but despite that I still think it has value as an indicator here.

I’m a couple of weeks late on this (still playing catch up), but I thought this recent story was worth noting.

On May 2nd, China Daily reported that “Beijing has ‘cleanest month’ in 9 years,” writing, “the city experienced its best month of air quality since 2000 with 23 blue-sky days in April.”

The excellent blog Daily Dose of Air Pollution highlighted that this claim is dubious, noting at least three other months (August and September, 2008, and August 2006) in which Beijing had higher numbers of Blue Sky Days and lower average APIs than April 2009.

I think I’ve identified the source of confusion. The official Beijing EPB announcement (Chinese), titled 4月本市空气质量创2000年以来同期最好水平, states specifically that April 2009 was the best April since 2000, not the best month overall. It seems the China Daily (or the Beijing EPB spokesperson during the press conference) misrepresented the real announcement.

Two follow up points:

1) While acknowledging progress, we should also simultaneously not get too excited over the “clean” air. The Beijing EPB claims that the average PM concentration during this month was 120 ug/m3 (主要污染物可吸入颗粒物月均浓度为每立方米0.12 毫克), which is still well above China’s national air quality target (100 ug/m3) and six times higher than the WHO recommended guideline (20 ug/m3). (Comparison of international standards in this post.) Although it is critically important in China to note progress, we must not wrap ourselves so much in cheers of success that we become blinded to the significant challenges and work still ahead.

2) The China Daily article describes in more detail than I have ever seen how the economic slowdown may have contributed to improved air quality, writing:

Besides strict environmental protection measures, experts think the global economic slowdown might be playing a positive role in environmental protection.

Zhu Tong, an environment professor with Peking University, told China Daily on Friday that heavy industry has decreased production in many polluting factories, which benefits the air.

“Most companies in heavy industry are seeing fewer orders. The output of the Shougang Group this year so far equals the same period during the Olympics,” said Wang Dawei, head of the air quality control division of the Beijing municipal environmental protection bureau.

In the first season this year, the added value for ferrous metal and chemistry manufacturing in the capital was 3.36 billion yuan ($490 million) and 1.85 billion yuan, a year-on-year decrease of 18.1 percent and 17.9 percent respectively.

If the improved air quality is indeed due to the slowdown, then this means there is even less cause for celebration.

Yesterday, Xinhua reported that Beijing achieved above an 80% “blue sky” rate in the first half of this year:

BEIJING, March 31 (Xinhua) — Beijing saw 73 “blue sky days”, 81.1 percent of the total, in the first three months of 2009, Beijing authorities said here on Tuesday.

The city experienced six more blue sky days than in the first quarter of last year, and 24.3 days more than the average of the last decade, said an official of the Beijing Municipal Environmental Protection Bureau.

Experts at the bureau said active cold airflows had helped particulate matter to disperse.

Efforts to reduce pollution from heating systems had paid off with January having the most blue sky days since 2000, said the official.

For those who prefer original sources, here is the notice from the Beijing EPB (in Chinese).

Let’s take a closer look at the data and examine what this means in terms of air quality.

First of all, as a quick introduction for new readers, China defines “Blue Sky Days” as days for which the Air Pollution Index (API) is 100 or below. For a detailed description of exactly what the API is, see this post. Past Beijing API data may be queried from MEP’s datacenter, but unfortunately only in Chinese.

For this post, I queried the data for 1/1/09 to 3/31/09, and ran some quick analyses to see what interesting things I could find.

Part 1: Checking Beijing’s EPB’s Numbers

The Beijing EPB claims “截至3月31日，今年累计73个达标天，占监测天数的81.1%。其中一级7天，二级66天，三级15天，四级1天，五级1天。” The second sentence says that in this time period there were 7 days of Grade 1 (API 0-50) air, 66 days of Grade 2 (API 51-100) air, 15 days of Grade 3 (API 101-200) air, and 1 day each at Grade 4 (API 201-300) and Grade 5 (API 301+).

However, by my count, there were 8 days of Grade 1 (one more than reported) 64 days of Grade 2 (two less than reported), then 15, 1, and 1 days of Grades 3, 4, and 5, respectively (identical to that reported). It seems pretty basic to me that you would want your publicly reported data summary to match your public database, so I can’t imagine what’s going on here. This is especially true because, in this case, the data indicates Beijing did better than they claimed.

The data for one day, 2/19/09, is missing from the database. If we assume that 2/19 was a Blue Sky Day, though, then at least we do indeed have 73 Blue Sky Days for the quarter.

Part 2: Converting to Pollutant Concentrations

Because API is a unitless index, in order to evaluate air quality we have to convert back to pollutant concentrations. This is a bit tricky, because the API is only reported according to whichever pollutant had the highest daily concentration, meaning that we do not have daily raw data for every pollutant. Still, we can make a rough approximation by assuming that PM10 is the limiting pollutant on all days. (Of the 89 data points, 72 (81%) were reported with PM10 highest, 9 were reported with SO2 highest, while 8 had no pollutant data because no pollutants are listed for Grade 1 air quality days.)

In any case, given the above assumption, this graph shows calculated daily PM10 concentrations for this quarter:

These data yield a quarterly average PM10 concentration of 124 ug/m^3. This is very similar to what I calculated as last year’s annual average.

124 ug/m^3 is still well above China’s annual target (100 ug/m^3) and well well above the WHO’s ideal target for developed nations (20 ug/m^3).

Appendix

For those of you following along at home, here is an Excel formula for converting API to PM10 concentration (in this example, the API would be in cell E9):

=IF(E9<51,e9,if(e9<201,(e9-25)/0.5,if(e9<301,(e9+300)/1.429,if(e9<401,(e9+225)/1.25,e9+100))))

This is based on equations presented at the bottom of this post.