Albemarle Terroir

I recently created a simple wine quality index to compare the vintage notes of the Chateau Montelena to daily weather records from St. Helena, CA.       In the post,  I used the variables of primary importance from Ashenfelter et al. 1995.  Since I did not have market pricing data to estimate the value of regression coefficients, I created simple weights from normalized weather data.   The normalized values were:

1. Accumulated Winter Season Rainfall
2. Average Growing Season Temperature
3. Assumulated Harvest Season Rainfall

The index result in a time series displayed below:

A Wince Quality Index for Northern Napa Valley

In this post,  I add a new component to the index and compare the results.

As an example,  the 1972 notes state “Record heat in July caused considerable damage”.

I’ve seen other reports of heat damamge or heat injury.  I recently created a quick post on heat damage as well.   So,  it seems only logical to add to the index a term that measures the likelyhood of heat damage.

The eqaution for the Quality index is:

I modified the equation to include a term that counts the number of days during the growing season that a daily maximum temperature of 95 oF is exceeded.   As before,  this time series is normalized, and only exceedances greather than average are applied.   The equation then takes the form:

The “Temperature Excedance” variable is a normalized time series of a count of the number of times daily maximum temperature execeeds 95 oF during the growing season.

The Wine Quaility Index based on the temperature exceedance (a measure of the potential for  heat injury) and the original Quality Index is displayed below.

A Heat Index based Wine Quality Index for Northern Napa Valley

In general there are no drastic changes to the quality index when heat index is added.  This is not unexpected since it’s wieght is only -0.25.   But, it may “fine tune” a few of the values.

The New Zealand Winegrowers have announced the statistics and outlook on quality on the 2009 vintage recently and they are quite positive. As can be expected, weather is believed to have an influence. To quote a quote from the article:

“We enjoyed a very good growing season this year. Some early humidity and weather pressure in February was replaced by a superb March and April. This meant our growers and wineries were able to pick the grapes at optimal ripeness.”

I’ve tried to track down weather records from the Marlbourough Region in New Zealand (where their famous Sauvignon Blanc is produced), but can’t find any readily available (i.e. free) data.

Since I bring up New Zealand Sauvignon Blanc,  I’ll mention that I  recently had a chance to taste a Virginia Sauvignon Blanc along with a New Zealand Sauvignon Blanc.

The Virginian was a 2008 Barboursville Sauvignon Blanc and the New Zealand was a 2008 Brancott Marlborough Sauvignon Blanc.    Unfortunately (or fortunately, depending on how I look at it) I tasted both of these in conjunction with a dinner I had prepared, which is to say,  I did not get a chance to do a formal tasting but did consciously compare the two.     The Barboursville was definitely good and contained some excellent flavors and aroma, but the New Zealand wine presented itself with complex flavors that ran  circles inside my mouth.     It was very good.    At some point I’ll have to try two or three compared to each other in a more formal manner.

I came across an article describing the effects of recent heat in Padthaway (Australia) wine quality.

The last sentence is quite interesting:  “and the heat damage to the grapes means 2009 will not be one of the great years for wine quality.”   This and other recent readings suggest that a wine quality index must include some sort of indication of excessive heat days.     More of this will be discussed in a follow up article to “Chateau Montelena, Climate, and Wine Quality.”

I came across an interesting page at the web site of the Chateau Montelena, located in Calistoga, CA.   Thier Vintage Summary page discusses the weather and climate conditions of each vintage year and provides a few notes regarding each vintage.    Clearly I had to find some weather data and apply a bit of analysis to it.

The California Department of Water Resources maintains the California Irrigation Management Information System, which includes a network of weather stations around the state.   It includes a weather station located in St Helena, which is in the Northern Napa Valley,  just south of Calistoga.    The data provided by CIMIS includes daily maximum and minimum temperaures and daily precipitation values.     There are a few details that I have glossed over currently regarding the data set, but as I continue my analysis over the coming months, I’ll address those issues.

Following the lead of the Ashenfelter et al., 1995 paper and earlier analysis of Bordeaux Wine Quality and Climate,  have constructed a simple index to predict the quality of St. Helena  and Northern Nappa Valley grown wine.

The index was created using the following steps:

1) Download CIMIS data for Daily Maximum Temperature, Daily Minimum Temperature, and Daily Precipitation

2) Calculate seasonal averages for the above values.  The seasonal averages are:

A)  The daily average temperature during the growing season between May and August.

B) The Winter season precipitation.  Winter in this case is the months between October to March, inclusive.

C)  Harvest season precipitation, which is the total accumulated precipitation during August and September

3) I then normalized each data based on averages for each of the above variables from 1952 to 2008.

4) Applied a weighting such that:

A) The daily average temperature during the growing season, has  a weight of one.   The warmer the average the temperature, the better the quality.

B) The Winter accumulated precipitation is given a weight of +0.5 .  More winter precipitation than average positively contributes to wine quality.

C) The Harvest season accumulated precipitation is given a weight of -0.5.    Above average Precipitation during harvest season is detrimental to wine quality.

A bar chart representation of the index is shown below.

A Northern Napa Valley Wine Quality Index

I compare this to a few of the years from the Chateau Montelena Winergy page (look at the Blue bars,  red comes later):

Best Index Year:  According to the index, the1996 was the  best year for wine quality had an index value of 2.9.

According to Chateau Montelena,  1996 is described as:

Hot. Average rainfall. Mild winter, untimely spring rains and intense summer heat resulted in low vineyard yields. Ripening weather in September-October was excellent. Lightest crop size since 1988, very high extraction and very concentrated juice.

So,  based on the inputs my index, the siginficantly wetter than average winter, a warmer than normal growing peroid, and significantly less rain during harvest time all positively contributed to the quality.

Worst Index Year:  1989.  Index value: -2.33.

Chateau Montelena says:   Cool Variable. Record winter rains. Crop down 20% from 1982. Cool spring gave way to a heat spell in mid-July that caused damage. Early crush with a moderate sized crop. Good maturity, clean, balanced fruit.

Chateau Montelena says:

Hot. First drought year. Fruit had high sugars, low acids. Vineyards stressed. Early August – September crush.

Very low winter precipitation,   lower than average temperature,  and greater than normal harvest rain fall all contribute to the low index value.   Interestingly,  the data contradicts the “Hot” assertion indicating the growing season average temperatures were not “Hot”, but were in fact,  0.35 standard deviations below normal.

Summary:

Using the guidance of the Ashenfelter et al.  1995 paper a simple index is contstructed to indicate the quality of wine produced by Chateau Montelena based on CIMIS daily weather data.         A few things come to mind:

1. The vintage discussion must be taken with a grain of salt.   Since the purpose of thier web page is to educate web visitors and, in the end, sell wine it is interesting to note that apparenetly they never have poor quality wine.    This objective analysis provides a framework to analyze the quaility of Chateau Montelena wine from a year to year basis.
2. Thier 1996 description says:  ” intense summer heat resulted in low vineyard yields.”  While low yields do not necessarily indicate poor quaility,  statements like this are made several times.   An upcoming post will discuss the index with the addition of a “threshold variable” added.  Namely, the number of days each growing season in which temperature exceeds 95 degrees F.   The more days in a season, the lower the quality.
3. The 2000s have been a generally excellent decade for wine quality.
4. I’d like to track down some indicate of market prices for Chateau Montelena wines to allow me to conduct a direct regression of Napa Valley weather against wine quality, as measured by the market.

The Daily Progress, the local Charlottesville, VA daily newspaper, published an article today entitled “Welcome Weather for Wineries.”   The article discusses how the weather so far in 2009 has been supportive of high quality, high yield grapes.

A quick update to follow up on my suggestion that the weather station at Bordeaux-Merignac airport moved sometime in 1987 or 1988. 

For a quick analysis I accessed GSOD  data from a nearby weather station at the Cazuax, France airport.    Cazaux is approximately 15 miles to the southwest of Bordeaux, close to the Atlantic Ocean.  I calcuated the same DEGREES variable as calculated for Bordeaux and compared the two time series.  Click the image below  to see a larger version. 

Temperatures during the growing season in Bordeaux and Cazaux, France

The chart indicates that up through 1987 the temperatures between the two stations are quite similar but then Bordeaux becomes consistiently warmer.    To illustrate this a bit better,  I have also plotted a time series the difference in DEGREES between the two stations.  

The Difference in DEGREES between Bordeaux and Cazaux, France

In addition to the time series I have graphed the average difference between 1973 and 1987 and from 1988 to to 2008.   The average difference from 1973 to 1987 is -0.35  degrees while from 1988 to 2008 it’s 0.25 degrees.   I believe this is compelling evidence indicating the station moved, causing a 0.6 degree average temperature shift.      In addition to this “quick and dirty” analysis,  I’m working on a more sophisticated regression based analysis. 

To put this in perspective,  the standard devation of Bordeaux DEGREES between 1988 and 2008 is 0.52 degrees.  So, the station move imparted a climate signal equal to one standard deviation of the year to year  temperature variability.   

“Working Paper #4“, an update to the orginal Ashenfelter et al., 1995 paper published online by the American Association of Wine Economists,  states “Indeed, the prevalence of such warm weather in the summer in the last two decades no doubt accounts, in part,  for the deeply held convictions that many Europeans hold that global warming is already upon us.”      I believe some of the statements (and the regressions) in the paper need to be revaluated in light of the possibility of the Bordeaux station move.      The good news is that I’m working it.     I am NOT suggesting the results of the paper are incorrect,  but am merely saying that the statistics could be improved upon.     A referenced in a earlier post,  if you are going to speculate on the price of Bordeaux wine,  you better make sure your model is as accurate as possible.

I have prepared a comparison of weather data for Bordeaux-Merignac airport in a effort to reproduce the results of Ashenfelter et al., 1995 based on weather data I’ve procured.      Here are some preliminary thoughts and results:

Goal

My goal for conducting this exercise is to find weather data sources that could be used to monitor Bordeaux and other wine regions (i.e. Virginia) during the season to speculate the likelihood of high or low quality outcomes.     There are data sources that could be used, but I’m not convinced yet that the temperature data at Bordeaux is representative.  I think the station was moved sometime near 1987,  but need to prove it.

Weather Data Sets

1) Weather Data from the original paper, as provided by Liquid Assets – Data from 1952 to 1988.   I contacted Dr. Ashefelter to inquire about the original source of data for the paper.   Apparently, the data used in the first publication were from “a french journal”, in which the data was transcribed from the journal.   For the  updated paper,  the data came from “a dutch web site.”.   I presume the Dutch web site is, in fact, KNMI.

2) Global Historical Climatology Network (GHCN)  data, as provided by NCDC – Data from 1952 to 1999

3) Global Summary of the Day (GSOD) data, as provided by NCDC – Data from 1972 to 2008

The GHCN and GSOD data can be obtained as the daily weather recordings and then averaged by month to reproduce the variables presented by Ashenfelter.   Please note that that difference between the GHCN and GSOD data is not that data is coming from different weather stations.   It’s simply that the data is reported from the same weather station under potentially different reporting standards and made available via different data sets.   Therefore, when using observational data,  and especially meteorological obsevation data,  there is one critical assumption that everyone must start with.  Assume the data is wrong.   Observing the weather is complicated stuff and while a spreadsheet might contain data,  you usually have to be very careful to ensure it’s correct.

In  the paper,  Ashenfelter uses three meteorological variables.   They are:

WRAIN – Winter rainfall.  Accumulated precipitation averaged from October to March prior to the growing year.

DEGREES – Average temperature from April to September of each year.    This represents the temperature during the growing year.

HRAIN – Harvest Rain,  the accumulated precipitation in August and September of each year.

In the charts below,  I have labeled the new data with the same base name (i.e. WRAIN) but appended the source (i.e. GHCN).

The Data – WRAIN

The chart below shows the three time series for the WRAIN (winter rain fall).

Winter Rainfall in Bordeaux, France.

A few points:

1) The Ashenfelter and GHCN data are largely similar in variation but have different mean values.

2) The GSOD and GHCN data are quite similar in the years that they overlap, but interestingly both GSOD and GHCN data significantly differnt means during this time.

The Data – DEGREES

The chart below shows the time series for the three DEGREES time series.

Average Temperature During the Bordeaux Growing Season

Notes:

1) As stated in an earlier post,  I believe there is information not reflected in this graph.  Namely, the station was moved sometime near 1987.

The Data – HRAIN

The chart below depicts the three Harvest time Rainfall for Bordeaux, France.

Harvest Rainfall in Bordeaux, France

Notes:

1)  Where is HRAIN?   It’s covered by HRAIN GHCN.   They are identical for about 20 years.

2) Even the GSOD data is quite similar during the early time period in which all data overlaps.

A Revisit of WRAIN

Let’s take a look at WRAIN again.  There are 13 years where HRAIN, HRAIN GHCN, and HRAIN GSOD overlap.    For each of the time series I have, I calculated the means Winter Rainfall for each time series during the 13 years of overlap and removed it.   So,  I now have “zero centered” time series showing the year to year deviations.    It is shown below.

Click on Image to Enlarge

In general,  the three data sets show the same variability (even if their standard deviations are not identical).   Less than normal rain is reflected in all three.  Greater than normal precipitation is reflected in all three.   Fortunately,   the excellent quality vintages occur in non-normal situations.    From this analysis, I can gain some confidence in using GSOD precipitation data for more recent years (given that the GHCN data availability ends in 1999).

I conclude the following from my analysis.

1) Assume the data is wrong.    Blindly applying the regression coefficients from the Ashenfelter et al, 1985 paper could be a dangerous endeavor without understanding how data from different sources compare.   I showed above, for instance, the means of winter rain are different across three data sources but the variability is at least similar.   You might choose to speculate the quality of a wine is underestimated by the market based on the equations.  But, is it the quality that is underestimated or the weather data that is causing that under estimation.

2)  Assume the data is wrong. I am still unsettled about the temperature time series during the growing season.   I’ll be working up an analysis to see what can be said about it.

I recently wrote about Ashenfelter et al., 1995.   In an effort to reproduce the results of the paper, I tracked down Global Historical Climate Network data for Bordeaux-Merignac and compared it to the weather reported in Ashenfelter’s paper. The problem with the GHCN data for Bordeaux is that it ends in 1999.   There happens to be another source of data provided by the NCDC, called the Global Summary of Day, and it too contains Bordeaux daily weather data up to today.

56 Years of Bordeaux Temperature Data

The graph shows three time series of 3 sources of data:

1) Ashentemp – Temperature data averaged between April and September for each year shown.  This is the data as reported by the original paper

2) GHCN Temp – Average temperature between April and September for each year shown from the Global Historical Climate Network.   The average is the average of “daily average temperate” for each day in the year.   Daily Average temperature is the average of the reported daily maximum and daily minimum temperature.

3) GSOD Temp – Average temperature between April and September for each year shown from the Global Summary of the Day.

When I look at this graph one thought comes to mind.     Something happened in 1987 that drastically changed the climate as reported by the Bordeaux-Merignac weather station.     Much has been written about potential impacts of climate on the wine industry, but I don’t believe the signal shown is related to climate change.    The change is too sudden.   Up to 1987, the average temperature is approximately 16.5 oC, where as after 1987 is about 18.5 oC.   The graph indicates a less than gradual change.
Even more interesting,  the standard deviation of temperature from 1952 to 1988 is 0.63 oC.    So, the change in temperature of 2 oC is much greater than the natural variability of climate.

What happened?   I believe the station was moved .   There was a time when I was a participant in the Weather Derivatives industry, in which a contract pays out based on what happens with weather at specific airport locations.   That industry has put significant effort into documenting known and unknown station moves.    Believe it or not,  but moving a weather station from one side of an airport to another can cause significant changes in the reported weather and therefore its climatology.

I’ll develop an analysis of this station, but I don’t know yet if I have enough data to make a statement about a station move.  There are probably records of what happened to the station, but they are held by Meteo France, meaning I’ll never get them.    If the station was moved and it explains this temperature change,  the results of the Ashenfelter paper need to be reevaluated relative to the change in the reported climate.   In general,  I expect the variability of temperature to remain essentially unchanged, but the mean temperatures may change.   This is important because the perceived quality of post-1987 vintages could be misrepresented (think 1990) after recalibrating the model.

Ah,  much analysis to come.   Who would have though a blog about Albemarle County Wine and it Terroir would have spend so much blog space discussing Bordeaux.   Unfortunately, I need to sort this issue out as the paper provides an excellent framework to evaluate the quality of wines in Virginia and the rest of the world.

If we go all the way back to my first post on Albemarle Terroir, I quoted a description of Terroir written by George Taber.

He says:  “Climatology, geology, and history all fuse in the French concept of terroir.  There’s no exact English translation for the word that combines all the factors that go into making outstanding wine.  Terroir is founded on the conviction that there is a perfect place for making wine, where the soil and the weather and the knowledge of the ages combine to produce truly great vintages. ”

The Ashenfelter et al, 1995 paper that I wrote about found that the quality of a Bordeaux vintage is strongly related to the temperature during the growing season and the rainfall just before and during the harvest season.    As stated above, one of the factors of Terroir is climate.    And, terroir is what can a a wine so fantastic.

But the Ashenfelter paper points to something a bit more sublime.   Variability in the quality of Bordeaux vintages is caused by changes in year to year climate.     Because the climate changes from year to year, the quality changes from year to year.   Since terroir is a function of all three factors above,  terroir is not constant!  (Let’s not talk about Climate Change just yet).

Granted,  there can be some wineries in places you would not normally think of having wineries due to their climate, but the climate component of Terroir becomes very interesting in the face of Ashenfelters paper.    In a side bar,  he quotes A. Haraszthy in a 1862 report to the California Legislature that “California climate is eminently adapted to the culture of grape-vines. . . California, having an even temperature, is warm and with rains in summer.”

There have been long debates about whether California’s Napa Valley has better Terroir than Bordeaux.    It may or may not.   But, given the quote above,  you might be able to say it has a better climate component of Terroir.

Over the weekend a friend told me about an article relating climate in Bordeaux, France to Bordeaux wine quality.    The article is “Bordeaux Wine Vintage Qaulity and the Weather”, published by the American Statistical Association in 1995 and authored by Orley Ashenfelter, David Ashmore,  Robert Lalonde.    The basic premise of the paper is that understanding the weather conditions in Bordeaux is an excellent indicator of the price of top growth wineries from Bordeaux (and therefore quality — i.e. there is an efficient market for wine).        

What a paper!    The paper really got the creative juices flowing and helped consolidate my  long term goal for this blog.  Namely, to monitor climate in Virginia and use it as a discussion piece for the likely quality of Virginia, Albemarle County, and Charlottesville Wine as a year develops.  

From what I can tell surfing around the web, the paper created quite a ruckus at the time it was published.  If weather is a major determinant of quality, for what do we need all the wine expert comentators?       It turns out that Dr. Ashenfelter created a publication called Liquid Assets,  which discusses wine as an asset.   On that site, he has updated the orginal paper with more recent results. 

 More importantly for this post,  the site also provides tabular data from the article.    As there was a time in my former life when I was a scientist,  I can’t help but try to 

1) Recreate the results of paper (accomplished easy enough in MS Excel)

2) Try to extend the results based on the data available,

3) Apply the results to other regions gowing wine (Napa,  Washington, Australia, and, oh yeah, Virginia). 

The article relates weather data to Bordeaux wine prices.    So,  let’s get some weather data.    It turns out that the National Climatic Data Center makes the Global Historical Climate Network – Daily Data avaiable for download.    

Based on the data from Bordeaux,  a question becomes immediately obvious.   What surface temperature data was used in the paper?

A temperature comparision to Ashenfelter et al., 1995 and GHCN Daily Data

In the time series graph,  I compare the time series of  April to September  averaged data for each year  from the Bordeaux weather station available from the data set.   As can be seen, the time series are very similar but not exactly the same. There is a slight warm bias in the GHCN data.  The red line is the temperature data from the Ashenfelter et al. paper whereas the blue line depicts the temperatures directly from the GHCN daily data.    I’d be interested in knowing the exact source of data used by Ashenfelter et al. (in the vein of scientific reproducibility).   What could cause the difference in the time series?

1) Quality Control — I have applied no quality control to the data I extracted from the GHCN data set.  However, NOAA has extensively QC’d the data. 

2) Trend correction –  There are many efforts to correct temeprature and related time series for trends attributed to climate change  that have appeard in data over the last 50-75 years.   The fact that the blue (GHCN data) tends to gradually rise above the red line (the papers data) suggests to me that it may, in fact, be “trend corrected” data.   

3) A different station is used –  I have not yet extensively researched the availability of weather station data from Bordeaux.  It could be that the GHCN daily station (which is the airport data in Bordeaux) is different that the data actually used by Ashenfelter et al.        For those who don’t know about European Weather Data, unlike in the United States,  weather data is generally restricted and difficult to obtain.     I don’t hold out hope that if a different station has been use,  I’ll be able to get it.

By tracking down the weather data from Bordeaux, I’m hoping to be able to extend the results a bit.  Maybe the early season temperatures are more important than late season.   Maybe a heating degree day type formulation improves the model.     The issue above does not prevent me from looking at these questions, but it does complicate the issue a bit.         I’ll be searching for answers however.