As noted in this article, the first Natural Swimming Pool (NSP) has been approved in the U.S. for public use (such pools were already allowed in the residential market as there are no regulations requiring specific disinfection in residential pools -- only restrictions on registered pesticides and their claims).
Bathing Water EC Requirements
I thought I would compare the regulatory water quality requirements in the European Directive 2006/7/EC against that of the field test requirements of EPA DIS/TSS-12. Note that the European Directive does not apply to swimming pools and spas so one cannot call the bathing area that is created a swimming pool. It is generically described as "bathing water".
The bathing water is classified as "poor", "sufficient", "good", or "excellent". By 2015, all bathing waters must be at least "sufficient" in quality and any bathing water that is "poor" for 5 consecutive years will have a permanent bathing prohibition or advice against bathing. The following criteria are for inland waters. Coastal waters and transitional waters have criteria roughly half the levels indicated below.
....................................................... Excellent . Good . Sufficient
Intestinal enterococci (cfu/100ml) ..... 200 ....... 400 ....... 330
Escherichia coli (cfu/100ml) .............. 500 ..... 1000 ....... 900
The criteria for Excellent and Good are at a level of 95%-percentile evaluation while Sufficient is at a level of 90%-percentile evaluation.
Up to 15% of the samples, or one per bathing season, whichever is greater, may be ignored and not counted in the above. These are treated as "short-term pollution" and steps must be taken to warn or prohibit bathing during such times.
By comparison, the EPA DIS/TSS-12 criteria are that the standard plate count (for all heterotrphic bacteria) not exceed 200 cfu/100ml and that coliform bacteria (which would include E.coli) not exceed 2.2 cfu/100ml and that entercoccal organisms not exceed 2.2 cfu/100ml. 85% of the samples must meet this criteria.
Because of the differing criteria for % of samples that must meet a standard, a direct comparison cannot be made, but using rough standard deviation values the EPA criteria is roughly 50-100 times more stringent.
So the discussion would be as to whether the looser European criteria provides sufficient safety. Clearly it is a criteria that prevents uncontrolled bacterial growth. Neither the European nor the EPA criteria are specific to other bacteria (though the EPA does have total plate count) nor to viruses or protozoa (or their oocysts). In the U.S., the primary concern for disease transmission in chlorinated pools is that of the protozoan oocyst Cryptosporidium parvum. It is not clear how the natural pools will handle this pathogen without supplemental coagulation/filtration or supplemental inactivation.
German FLL Requirements
I also compare against the German Forschungsgesellschaft Landschaftsentwicklung Landschaftsbau e.V. (FLL) Recommendations for the Planning, Construction Servicing and Operation of Public Swimming and Bathing Pond Facilities. There are no regulations for E.coli for private ponds, but for public swimming ponds the limit is 100 CFU/100ml while for Enterococci it is 50 CFU/100ml and for Pseudomonas aeruginosa it is 10 CFU/100ml.
Here again, the EPA criteria for the field test for pesticide registration is roughly 20-45 times more stringent. Again, the question is what level of criteria is appropriate for sufficient safety.
Comparing German FLL to EPA DIS/TSS-12
The EPA document Implementation Guidance for Ambient Water Quality Criteria for Bacteria describes the risk level in terms of illness rates over a swimming season. The German FLL requirement for a maximum Escherichia coli level of 100 CFU/100ml corresponds to a risk level of (log10(100/17.742))/1.064 = 0.7% or 7 ill swimmers per 1000 per season while the requirement for a maximum Enterococci level of 50 CFU/100ml corresponds to a risk level of (log10(50/4.656))/1.064 = 1.0% or 10 ill swimmers per 1000 per season. In practice the geometric mean would be kept below these maximum limits, but even if I assume that these limits are 95% confidence levels, then this is equivalent to a geometric mean for E.coli of 100/10^(1.65*0.4) = 21.9 so a risk level of (log10(21.9/17.742))/1.064 = 0.086% or 0.8 ill swimmers per 1000 per season while for Enterococci this is equivalent to a geometric mean of 50/10^(1.65*0.4) = 10.9 so a risk level of (log10(10.9/4.656))/1.064 = 0.35% or 3 ill swimmers per 1000 per season.
The EPA laboratory criteria effectively require a complete kill (6-log reduction) of Escherichia coli in 30 seconds or less and of Enterococcus faecium in 2 minutes or less. The field test requirement requires at least 85% of the samples to have less than 2.2 CFU/100ml. This corresponds to a geometric mean of 2.2/10^(1.03*0.4) = 0.85 CFU/100ml and an illness rate of virtually zero (the essentially zero illness level is at a geometric mean of 17.7 CFU/100ml for Escherichia coli and 4.7 CFU/100ml for Enterococcus faecium). Real data from CDC surveillance summaries shows that this is largely the case where most illness in disinfected water is due to the protozoan oocyst Cryptosporidium parvum because it is so highly chlorine resistant.