My Professor friend was recently awarded by the Ministry of Environment & Forests towards his contributions for saving expenditures on environmental monitoring incurred by the Pollution Control Boards (PCBs), industries and Urban Local Bodies (ULBs). The Professor was top in the news, on TV and print media. The exact details or his precise contributions were not known – but there was praise for his ingenuity, especially about the tool that he developed using “random numbers” and a practical approach that he followed. Millions of rupees of the Government were expected to be saved. I was therefore very keen to meet him to learn about this innovation.
When I went to his office, Professor was on the phone, busy as usual. “No, I am fully booked” he was saying. “Talk to me again next month and we will fix a slot to speak a month later”. He turned to me while slamming the phone down and said “too many requests to speak my friend. Everybody wants to know about my random number algorithm on environmental monitoring!” His table was littered with notes, scribbled with numbers, some graphs and couple of cups of tea with used tea bags hanging out. Clearly he had a number of visitors that morning.
I requested the Professor to explain his innovation and apologized at the outset as he must have done this part of explanation so many times earlier to several like me! The Professor however didn’t mind. “No problems Prasad, l will do this for you. I know you well and you will I am sure explain my contributions to others as well – perhaps better than me”. He winked. He then walked across to a white board in his room, cleaned it up and started explaining his theory of random numbers as applied to environmental monitoring.
His basic argument was as follows.
We monitor environment (air quality, water quality, noise etc.) to understand the state of environment, establish trends, detect violations and communicate results to public, decision makers, implementers.
We however conduct environmental monitoring at pretty low frequencies. We also operate relatively sparse density of monitoring stations. On Ganga basin for instance we sample water quality at less than 100 stations (over its 3000 kms of length) and at each station sample water quality only once a month.
“Do you think this kind of data is adequate to understand behaviour of water quality in a river, understand water quality trends or detect violations over the prescribed standards? This data is simply not sufficient to show improvements (if any) in water quality due to the Ganga Action Plan – II. I told the Prime Minister that he is going to have a tough time to prove that investments he is making have actually improved water quality in Ganga” He paused. “We should be having many more stations and much higher sampling frequency if we are serious about environmental monitoring. Either do the job well or just don’t do it. Folks don’t understand statistics”. His voice was now a bit raised.
The monitoring stations we have located are often not well sited. That leads to a bias. Do you know that the so called ambient air quality monitoring station sited in city of Pune was at Nal Chowk – a busy traffic junction – which actually reported air quality of pedestrian exposure and not ambient conditions. Data from this station declared Pune as one of the 10 most polluted cities in India and the city was put under scan of the Bhurelal Committee. And Professor was right as I had visited the monitoring location and had advised the Pune Municipal Corporation and Maharashtra Pollution Control Board to relocate the station.
The Professor continued to speak
“And then look at the vagaries on how the sample taken, preserved and transported and taken to the laboratories. The QA/QC related situation is of concern. You know the BOD test we do in the labs is so unreliable when less than 3 mg/l so I often wonder reliability of our river BOD data. Recently, I was provided with results of synchronous monitoring of industrial effluent at a Common Effluent Treatment Plant. This monitoring was carried out by 3 reputed laboratories and by an automatic water quality monitoring instrument. The arithmetic means of all the 4 were different and the Z-test hypothesis failed across all the parameters ! Isn’t this frustrating? See a snap shot of results.” He passed to me one the sheets like on his table.
Again what do we do with the data collected? We simply report means, max/mins,, time series graphs (that we call as “trends”) and report compliance (over poor frequency data!). Much more can be done.
I nodded. “Yes Professor, we all know this reality, sad it is but then where is your random number innovation?”
“Aha” the Professor said. He lit a cigarette and put his spectacles on.
“I realized that there is NO great advantage of doing actual environmental monitoring. The kind of monitoring we do is as good as generating random numbers! The monitored data is neither representative nor reliable. So why bother? Why should we build and operate such an expensive environmental monitoring infrastructure? Why spend money?”
So I thought of generating random numbers. But cleverly!
All I do here is use past collected data. Prepare quality control chart and then toss random numbers in the range of one sigma, two, three and four sigma – sigma being the standard deviation. The random numbers I generate follow an intelligent algorithm that almost mimics the past data behaviour (to bring in credibility) but with some deviations as needed or desired. See Figure below. So the dots in the figure are my synthetic environmental monitoring data. In this case BOD. Sorry, I won’t tell you how I generate these random numbers – a patent has been already filed. So hold on till then.
Look at the advantages. If you use my algorithm, you don’t need to spend moneys on collecting and analysing the actual sample, no need of laboratories, no need to employ staff and you directly get the environmental monitoring data sitting in the office! The data I generate is not far away from “reality” and cannot be easily questioned. Importantly, the randomness in the data can be controlled depending on what you want to prove.. Whether environmental status is improving or deteriorating. When I explained to the Minister this feature of the algorithm, he was so pleased!
I was shocked! “This is Professor nothing but data manipulation” I said with all my courage, emotions and a concern.
The Professor smiled. “Think deeply Prasad. In either case, does this matter? In both cases, i.e. actual monitoring and monitoring using my random number algorithm, we don’t take decisions on environmental improvement anyways! In India, monitoring is not linked to management. And don’t forget how much money and effort we save using my algorithm”
I left Professor’s office both depressed and disturbed. One thing that intrigued me however was how come the Professor did not think about loss of jobs at PCBs, redundancy created at the field and regional laboratories and loss of the business of equipment manufactures & chemical suppliers etc. The Professor certainly missed to look into the social costs (or benefits?) of environmental monitoring- the Committee at the Ministry should have thought of this downside of his random number innovation before bestowing the Award!!
Follow me if you like my blog
Environmental Monitoring has many spin off benefits. It generates employment. Ensures assured income to environmental laboratories. Gives opportunity to experts to debate on significance of increase from SO2 level 10.2 to 12.4, opportunity to Courts to close industry as BOD 33 is 10 % above limits and so on. Scientists, consultants and regulators are happy when more parameters are included in standards and more baseline data collection is directed. I only hope some day we will have disclaimers accompanying our conclusions and recommendations based on environmental quality monitoring.
Systematic data collection, it’s analysis and using it to take sensible decisions is not in our blood. Many of us regularly check our weight, blood pressure and blood sugar. How many of us have used this data to manage our life style and improve our scores on these parameters?
Dear Santosh
Your example of health “data” and our “lack of seriousness on actioning”is synonymous to “data on damage to the environment” and our callousness in taking a timely action or response.
In both cases, monitoring serves as early warning and if not attended to then lead to irreversible damage.
Regards
Professor is pointing towards a placebo – a fake treatment that can produce a real response sometimes. I find this very interesting and worth a shot because we don’t have all the data ‘yet’. In the mean time, the government can continue positioning more stations, hoping that it won’t get hooked to the placebo.
Innovative thought though !