Chester County PA - Analytical Battle of Actual vs. Altered Climate Data

[TheClimateChanger]

On 6/8/2025 at 6:35 AM, chubbs said:

Your in denial mode. These two charts fit together well. The station moves at Coatesville and West Chester produce spurious cooling if they are left in the raw data. That's why NOAA matches raw data without station moves and you (COOP station avg) don't. You have Chester County as warm as the Philadelphia airport before the station moves. How silly is that.

@chubbs - this is all a bunch of subterfuge. In the vast majority of cases, NOAA systematically reduces the reported temperatures from the observed temperatures on a county-by-county basis, often in ways that make little logical sense. Chester County may very well be the only county in Pennsylvania where the opposite is true. So if there is a conspiracy, it's in the opposite direction as the one presupposed with this analysis.

 

 

 

[TheClimateChanger]

Let's look at May 2025:

xMacis reveals three NOAA stations in Chester County, with a mean temperature of 61.5F. So it's true that the reported mean temperature is higher than the mean of these three stations. But the opposite is true in just about every other county in Pennsylvania.

[bdgwx]

And at the global level the net effect of all adjustments actually reduces the overall warming trend.

[Haufather 2017]

[chubbs]

21 hours ago, TheClimateChanger said:

@chubbs - this is all a bunch of subterfuge. In the vast majority of cases, NOAA systematically reduces the reported temperatures from the observed temperatures on a county-by-county basis, often in ways that make little logical sense. Chester County may very well be the only county in Pennsylvania where the opposite is true. So if there is a conspiracy, it's in the opposite direction as the one presupposed with this analysis.

 

 

 

The Chester County deep-dive has shown that a county station average is a terrible way to evaluate NOAA. The available stations aren't designed to produce county averages by simple averaging and the stations change with time. The average shelf life of a coop station isn't very long. Even stations with long-term records have station moves and equipment changes.

The NOAA method is well proven for getting the right answer from a constantly changing station network. If the stations didn't change then simpler methods would also work; but, as we have seen in Chester County the simpler methods breakdown when there are large changes in the station network with time.

 

[TheClimateChanger]

On 6/28/2025 at 5:48 AM, chubbs said:

The Chester County deep-dive has shown that a county station average is a terrible way to evaluate NOAA. The available stations aren't designed to produce county averages by simple averaging and the stations change with time. The average shelf life of a coop station isn't very long. Even stations with long-term records have station moves and equipment changes.

The NOAA method is well proven for getting the right answer from a constantly changing station network. If the stations didn't change then simpler methods would also work; but, as we have seen in Chester County the simpler methods breakdown when there are large changes in the station network with time.

 

Taking a look at the actual data, I was largely able to replicate NOAA's rankings/values from the raw data back to at least 2000. I limited this to 25 years, because it would be a big undertaking to expand significantly beyond that and you run into more data discontinuities as you head back in time.

In the "Occasional Thoughts on CC" thread, user @ChescoWx claims an actual summertime mean of 72.8F for the 2000s, 73.2F for the 2010s, and 73.5F for the 2020s, which it is claimed that NOAA alters to 73.1F, 73.9F, and 74.1F. Limiting the analysis to actual NOAA stations, I calculate a summertime mean of 73.1F, 73.9F, and 74.4F for the three decades, the first two exactly matching the reported and the current decade actually slightly warmer than reported. It appears those values are calculated only by adding in stations that aren't included in the NOAA analysis - mesonets, secondary airports (AWOS sites?), perhaps personal weather stations. I'm not sure what's going on.

In my analysis, I have data for Phoenixville 1E for 2000-2024, West Chester 2 NW for 2000-2016, Coatesville 2 W for 2000-2007, Avondale 2 N for 2006-2024, and Spring City 2 WSW for 2023-2024. Looking at this basket of stations, Phoenixville 1 E looks to be the warmest, followed by West Chester 2 NW, then Coatesville 2 W, then Avondale 2 N, and Spring City 2 WSW. So it looks like the station mix has gotten cooler with time with the Coatesville and West Chester sites gradually swapped out for the cooler Avondale and Spring City sites, so I would expect homogenization to increase the recent temperatures and/or decrease the older temperatures, but the opposite is true, with the older temperatures equal to the true mean and recent temperatures slightly chilled from the true mean. If anything, it looks like the NOAA trend for Chester County is slightly low for the most recent 25 summers?

[ChescoWx]

10 minutes ago, TheClimateChanger said:

 It appears those values are calculated only by adding in stations that aren't included in the NOAA analysis - mesonets, secondary airports (AWOS sites?), perhaps personal weather stations. I'm not sure what's going on.

 

Below are the stations used and dates operational

[ChescoWx]

The Phoenixville and Spring City sites that are now the only remaining NWS COOP stations of course run warmer than all other stations in 24 of the last 25 years. On average a 0.8 degree variation. With those sites running as much as 3 degrees too warm in one year and at least a degree warmer in 15 of those years!! Makes very little sense as these 2 stations are much lower in elevation and near each other in the Northeastern section of Chester County....yet these are the 2 stations that NCEI will use to calculate their county averages.....call me stunned!! LOL! 

[ChescoWx]

Look at how close these 2 stations are in NE Chesco....certainly not representative of the county climate

[ChescoWx]

On 6/28/2025 at 5:48 AM, chubbs said:

The Chester County deep-dive has shown that a county station average is a terrible way to evaluate NOAA. The available stations aren't designed to produce county averages by simple averaging and the stations change with time. The average shelf life of a coop station isn't very long. Even stations with long-term records have station moves and equipment changes.

The NOAA method is well proven for getting the right answer from a constantly changing station network. If the stations didn't change then simpler methods would also work; but, as we have seen in Chester County the simpler methods breakdown when there are large changes in the station network with time.

 

fixed it for you - The NOAA method is well proven for getting the DESIRED answer

[TheClimateChanger]

2 hours ago, ChescoWx said:

Look at how close these 2 stations are in NE Chesco....certainly not representative of the county climate

Looks like there's a cooler Avondale site in the mix as well. It's also misleading to call Spring City site unusually warm. It looks to be the coolest NOAA site since 2000, at least in the month of July based on my analysis. Despite its closeness to Phoenixville, it is substantially cooler than Phoenixville. Indeed, it is even cooler than Avondale, which was cooler by 1+ degree during the overlap with the old Coatesville and West Chester sites. Like I said, it looks like the current station mix is the coldest since 2000, so you would think recent years would be adjusted upwards from the calculated average observed July mean, but it is actually a little lower than the mean of Avondale, Phoenixville, and Spring City [at least based on what you presented as NCEI's figure for the 2020s].

Remember when you add a huge slew of new stations that aren't included in the NOAA analysis, that's essentially the same as saying those locations were equal to the county mean for all prior years in which those sites did not exist. If those stations are instead below the mean of the existing NOAA sites, then that is not a valid form of averaging.

[ChescoWx]

So I have finally completed the analysis of the Ghost Data stations and compared their inputs vs. the actual averages across the County. NOAA has employed techniques like interpolation to estimate temperatures in what they normally would say are data-sparse regions and in other cases this data may be resampled from another grid.  Of note there are seven (7) ghost stations employed across Chester County PA with this estimated temperature data in the period of record from 1895 through 2013. All of these were during the ghost years reporting only precipitation data. Some eventually did begin actual temperature reporting after the ghost years. The Ghost Stations and years of estimated data are Phoenixville (1895-1914) / West Grove (1928-1962) / Coatesville (1948-1955) / Glenmoore (1957-2005) / Honey Brook (1957-2013) / West Chester 2W (1979-1981) and Devault (1988-2004). Overall there are 189 years of ghosted data used to "fill in data gaps". Of note this ghost data was warmer than the actual county records in 101 of the 106 years of ghosted data between 1895 and the "shuttering "of the last remaining ghost station at Honey Brook in 2013.. The largest warming adjustments were unsurprisingly made during the coldest decades of the 1960's and 1970's. Below is an analysis of the impact on average temperature trend lines between the actual raw data and ghost data in the years these adjustments were applied. Some more analysis on this data will follow. 

[chubbs]

1 hour ago, ChescoWx said:

So I have finally completed the analysis of the Ghost Data stations and compared their inputs vs. the actual averages across the County. NOAA has employed techniques like interpolation to estimate temperatures in what they normally would say are data-sparse regions and in other cases this data may be resampled from another grid.  Of note there are seven (7) ghost stations employed across Chester County PA with this estimated temperature data in the period of record from 1895 through 2013. All of these were during the ghost years reporting only precipitation data. Some eventually did begin actual temperature reporting after the ghost years. The Ghost Stations and years of estimated data are Phoenixville (1895-1914) / West Grove (1928-1962) / Coatesville (1948-1955) / Glenmoore (1957-2005) / Honey Brook (1957-2013) / West Chester 2W (1979-1981) and Devault (1988-2004). Overall there are 189 years of ghosted data used to "fill in data gaps". Of note this ghost data was warmer than the actual county records in 101 of the 106 years of ghosted data between 1895 and the "shuttering "of the last remaining ghost station at Honey Brook in 2013.. The largest warming adjustments were unsurprisingly made during the coldest decades of the 1960's and 1970's. Below is an analysis of the impact on average temperature trend lines between the actual raw data and ghost data in the years these adjustments were applied. Some more analysis on this data will follow. 

Don't know what you are talking about. Can you provide a link to the so called "ghost data".

[ChescoWx]

1 hour ago, chubbs said:

Don't know what you are talking about. Can you provide a link to the so called "ghost data".

https://mesonet.agron.iastate.edu/request/coop/fe.phtml?network=PACLIMATE

One example of many below....this is a precipitation only station - so those Tmax and Tmins are ghosted estimated data......

[ChescoWx]

Looks like another temperature Ghost Station with the Coatesville 1E precipitation only station....generating high and low daily temps without the observer doing so! Thought ghost data ended back in 2014....looks like it continued right through the end of 2022....

 

[chubbs]

3 hours ago, ChescoWx said:

https://mesonet.agron.iastate.edu/request/coop/fe.phtml?network=PACLIMATE

One example of many below....this is a precipitation only station - so those Tmax and Tmins are ghosted estimated data......

As I thought "ghost data" is a big nothing burger. A figment of the denier imagination. The estimates are made by IEM not NOAA.  Below is  IEM's rationale for providing estimates. 

Why do this?
Previously, the IEM has only provided raw observations with limited quality control checks in place. Quality control is hard! Many times, users are simply looking for something "close" and perhaps not as perfect as high quality sensor observations can be. Producing a gridded analysis is one way to produce a dataset from point observations which can be sampled as a means of spatial interpolation.

https://mesonet.agron.iastate.edu/iemre/

 

 

[ChescoWx]

7 minutes ago, chubbs said:

As I thought "ghost data" is a big nothing burger. A figment of the denier imagination. The estimates are made by IEM not NOAA.  Below is  IEM's rationale for providing estimates. 

Why do this?
Previously, the IEM has only provided raw observations with limited quality control checks in place. Quality control is hard! Many times, users are simply looking for something "close" and perhaps not as perfect as high quality sensor observations can be. Producing a gridded analysis is one way to produce a dataset from point observations which can be sampled as a means of spatial interpolation.

https://mesonet.agron.iastate.edu/iemre/

 

 

Of course in reality IEM serves as a data provider to the NWS and others  I am looking to see if some of this data is incorporated into how NCEI arrives at their adjusted data....to tweak the actual real climate data.

[chubbs]

1 minute ago, ChescoWx said:

yep the old gridded analysis LOL!! I am looking to see if some of this data is incorporated into how NCEI arrives at their adjusted data....to tweak our data.

Don't forget your tin hat.

[ChescoWx]

1 hour ago, ChescoWx said:

yep the old gridded analysis LOL!! I am looking to see if some of this data is incorporated into how NCEI arrives at their adjusted data....to tweak the actual real climate data.

 

50 minutes ago, chubbs said:

Don't forget your tin hat.

LOL!!!! The truth is out there Charlie! It is so much fun peeling back the onion skin on these kind of data puzzles! Where and how did these alternate facts get added???? The raw facts as always will win out - stay tuned! The amount of interest in this across the media platforms is amazing!

[chubbs]

Found a MADIS site in Coatesville that is close to the WWII location of the Coatesville COOP. Per google map shot below, the area around the current station is  less built up, with larger lot sizes, than the old COOP site which I have posted previously. Per chart below, the current Coatesville station hit 100 3 days in a row during the recent heat wave, topping out at 102.2. During the heat wave, high temperatures at the station ran roughly 1F warmer than the Philadelphia airport. Interestingly, the Coatesville COOP also ran roughly 1F warmer than the Philadelphia airport in June and July during WW2. Indicating that the current MADIS station is a good match to the Coatesville COOP before it's move to a cooler location in 1946+47.

Warmer July highs than Philadelphia were not unusual Coatesville before the 1946 and 1947 moves to a cooler, more rural, location. Shows how warm the Chesco COOPs were back in the day and the mismatch between older and modern stations in Chester County.  Another datapoint that supports scientific methods to remove station moves and other network changes that contaminate raw temperature data. 

[ChescoWx]

2 hours ago, chubbs said:

Found a MADIS site in Coatesville that is close to the WWII location of the Coatesville COOP. Per google map shot below, the area around the current station is  less built up, with larger lot sizes, than the old COOP site which I have posted previously. Per chart below, the current Coatesville station hit 100 3 days in a row during the recent heat wave, topping out at 102.2. During the heat wave, high temperatures at the station ran roughly 1F warmer than the Philadelphia airport. Interestingly, the Coatesville COOP also ran roughly 1F warmer than the Philadelphia airport in June and July during WW2. Indicating that the current MADIS station is a good match to the Coatesville COOP before it's move to a cooler location in 1946+47.

Warmer July highs than Philadelphia were not unusual Coatesville before the 1946 and 1947 moves to a cooler, more rural, location. Shows how warm the Chesco COOPs were back in the day and the mismatch between older and modern stations in Chester County.  Another datapoint that supports scientific methods to remove station moves and other network changes that contaminate raw temperature data. 

 

 

 

That station is a personal weather station that uses The Acu-rite 5 in 1 personal station. A low cost model without fan aspiration - and the QC check failed of course over the last 2 weeks with that heat wave. It is recommend that station get a radiation shield to correct it's errors!! See the analysis below.  

[ChescoWx]

Now my station over the same period....

[chubbs]

1 hour ago, ChescoWx said:

That station is a personal weather station that uses The Acu-rite 5 in 1 personal station. A low cost model without fan aspiration - and the QC check failed of course over the last 2 weeks with that heat wave. It is recommend that station get a radiation shield to correct it's errors!! See the analysis below.  

Thanks for the info. The old Coatesville COOP didn't have fan aspiration either. None of the older stations did. Another reason to remove station changes from the raw data.

That is an interesting site. The average daytime error on your own station over the past 7 days, 3.9F, is almost as high as the Coatesville station but in the other direction.  The main message here is that every station is different. You can't throw a changing station population into a spreadsheet, take a simple average,  and expect to get the right answer.

[ChescoWx]

Last 26 weeks at East Nantmeal - good QC of course

 

[ChescoWx]

During the last month the radiation shield problem at that Coatesville station was enhanced by that heat wave....stations with strong shields like the VP 2 Pro that I have always used will be superior during heat waves. Below is the 28 day comparisons including the heat wave.

[FPizz]

18 minutes ago, ChescoWx said:

During the last month the radiation shield problem at that Coatesville station was enhanced by that heat wave....stations with strong shields like the VP 2 Pro that I have always used will be superior during heat waves. Below is the 28 day comparisons including the heat wave.

Can this tool be used to check personal sites or only ones that are registered or something? 

[TheClimateChanger]

1 hour ago, chubbs said:

Thanks for the info. The old Coatesville COOP didn't have fan aspiration either. None of the older stations did. Another reason to remove station changes from the raw data.

That is an interesting site. The average daytime error on your own station over the past 7 days, 3.9F, is almost as high as the Coatesville station but in the other direction.  The main message here is that every station is different. You can't throw a changing station population into a spreadsheet, take a simple average,  and expect to get the right answer.

Makes sense. It's not well sited, so it probably has a seasonal bias. MADIS does not have a good way to describe a cool bias, because typically the bias is warm. Funny too, his average error is higher over the past 28 days, even though only the Coatesville site is flagged due to the high standard deviation.

[ChescoWx]

I thought it would be of interest to do some deep analysis into my own weather station data that has been operational in East Nantmeal since November 2003 vs. The only NOAA Climate Reference Network (CRN) station here in Chester County which is the PA Avondale 2N station. These CRN stations are a highly specialized weather station designed to provide long-term, high-quality measurements of climate variables like temperature and precipitation. In essence, CRN stations serve as a high-precision benchmark for monitoring the nation's climate, ensuring the availability of reliable data for understanding and addressing climate change. I ran an analysis of East Nantmeal vs. Avondale 2N over the last 18 years that we have both been recording the weather. I have to say I am pretty pleased to see that our observations could not get much closer than this – exactly the same average temperature over 18 years of data!! I have only used a Davis Vantage Pro 2 station with the fan aspirated shield over the entire 22 years I have been recording data.

[ChescoWx]

47 minutes ago, FPizz said:

Can this tool be used to check personal sites or only ones that are registered or something? 

It is for personal sites also - below is my station performance over the last 28 days....green check means good siting and readings...

[FPizz]

2 hours ago, ChescoWx said:

It is for personal sites also - below is my station performance over the last 28 days....green check means good siting and readings...

Awesome thanks, I'll have to see if I can figure out mine.

[TheClimateChanger]

A little deeper dive in the Chester County temperatures over the past 18 years. Maybe @chubbshas an explanation.

I calculate a warming trend of 12.64F/century at the U.S. CRN station located at Avondale 2N, a warming trend of 12.13F/century at PHL Airport, a warming trend of 11.14F/century at @ChescoWx's East Nantmeal location. Yet NCEI only reports a warming of 7.99F/century.  According to data supplied by @ChescoWx, other stations he analyzes (varying sites) show a warming of 6.91F/century over this timeframe when averaged together. It looks to me like NCEI is underdoing recent warming, when the gold standard CRN station is showing so much warming, supported by PHL's data and @ChescoWx's high quality backyard data. I suspect actual warming could reach 15-20F/century over the coming decades before plateauing.