Tuesday, June 27, 2017

Monthly Precipitation Rankings: Wettest to Driest

In an earlier blog post, I looked at which months (and seasons) were the wettest and driest for thousands of stations across the U.S. In each case, just the wettest and just the driest months were identified. What if we want to know how each month of the year ranks in terms of wettest to driest? The map set below answers that question. 

For each of the twelve months of the year, a rank from wettest to driest was computed for each station in North America. Importantly, month lengths were standardized. This negates differences in the lengths of months. For example, a February that averages 3.00" of precipitation will "outrank" a March that averages 3.01" of precipitation since the per-day average of February is greater. All station data is based on the 1981-2010 time period. For U.S. stations, the values come from the NCEI published normals. For international stations, data come from the GHCN version 2 data set.


Thursday, June 22, 2017

Average Annual High Temperature vs. Climate Normal High Temperature

How hot does it get on the hottest day of the year? Several times in the last few years, I have presented maps and charts that look at the annual curve of normal temperatures. That's all fine and well, but it leaves out what people care about the most – namely, the actual temperatures. So let's look at them, when they occur, and how if differs for the normal curve.

The map in Figure 1 shows the average value for the warmest day of the year as measured for the years 1981-2016. If a station had at least 15 complete seasons (no more than 5 missing days for the May through September time period) it was included in the analysis. The calculation was an arithmetic average – not a median. I won't bother describing the patters, except to note that no attempt was made to account for the drop of temperatures with altitude beyond the actual station values themselves.

Fig 1. The average temperature on the hottest day of the year between 1981 and 2016. A minimum of 15 years required for inclusion. 8,011 stations were used to make the map.

Of interest to me is when the annual hottest temperature occurs. Do they coincide with the summer solstice? Do they lag by days or weeks? Using the same set of stations that was used to determine the value for the hottest temperature, I flagged the date of occurrence (if more than one date in a season tied for the highest temperature, an average of those dates was used.). The map in Figure 2 shows the results. There's a lot of stuff going on withing the map. Notice the early dates for the peak summer maximum temperature in Alaska, southeastern Canada, and the deserts of the southwestern Lower 48! In Alaska, this reflects the time of high sun before the clouds and rainy season commences. In the southwestern Lower 48, it reflects the time of high sun before the clouds and rainy season commences. In the southwestern Lower 48, it reflects the time of high sun before the summer monsoon sets it. Not sure what is going on in southeastern Canada or the region of Iowa and Minnesota. 

The relatively late date of the hottest day of the year in the western Gulf of Mexico coast is a function of the lag of warmer water temperatures advecting inland. Many of the other patterns reflect the onset of sunny/cloudy and wet/dry seasonal patterns.

Fig 2. The average date of the highest annual temperature observed between 1981 and 2016. A minimum of 15 years required for inclusion. 8,011 stations were used to make the map.

From a biological, agricultural, and energy point of view, a single date where the hottest temperature occurs is less important than the annual curve of temperatures. For that, the use of National Center for Environmental Information (NCEI) climate normal is called for. The map in Figure 3 shows the date when the average daily temperature reaches its seasonal inflection point (if more than one day, the median date was used). Once again, a distinct early season peak in temperatures is noted in Alaska and the desert southwest – owing to high sun and before a strong transition to greater cloud coverage and precipitation. Other regions reflect the thermal lag of ocean water temperatures and regional variations in seasonal cloud and rain patterns.

Fig 3. The date where the annual temperature cycle reaches its maximum value using 1981-2010 climate normals. A total of 6,076 stations were used to generate this map. For U.S. stations, it uses NCEI published values. For Canadian stations, values interpolated from GHCN v.4 data.

Of tremendous interest to me is the difference between these two maps. Figure 4 shows how many days earlier or later the average annual hottest day is compared to the normal peak summer temperature. The computation is: [average hottest temperature {minus} normal peak temperature]. Remarkable, there are some dramatic differences between the two. In Alaska, if all variables fall into place, the days near the solstice often produce the warmest temperatures several weeks before the normal temperatures peak. In Texas, the dry air from the Mexican Plateau allow for exceptional heat to move into Texas through the month of June, but is suppressed later in the summer. At the same time, a warming Gulf of Mexico slowly drives up the normal daily temperature through July (thanks to Richard Berler from KGNS in Laredo, Texas, for this analysis). South of Hudson Bay, a similar effect occurs in Ontario and Quebec. To be honest, I have no idea why.

Of equal interest, is the area shaded in red along the lower Ohio River valley. These areas typically see their highest observed temperature of the years two to three-and-a-half weeks after the seasonal temperature inflection point. This is either a result of an early peak in seasonal normal temperatures, a late occurrence for the observed highest temperature, or some combination of both. Based on the patterns in Figure 3, it appears they have an unusually early peak to their seasonal temperatures compared to surrounding areas but their observed dates for the hottest temperatures is in line with surrounding areas. Still, no other portion of the country is this disconnected with positive values.

Fig 4. Difference in days between the average date for he warmest day of the year and the normal date of peak temperatures (actual minus normal). A total of 6,506 stations had both an actual average max temperature and a normal annual match temperature.

June 24 Addition:

Some of the patterns that we see in the maps are artifacts of the time period chosen. In the above sections, only data from 1981 onward was used. If we do not restrict the time frame, patterns shift somewhat. The mosaic below shows what happens when the hottest day of the year analysis is changed. The top row in Figure 5 shows the information depicted in Figure 2 – data from 1981 onward. The bottom row uses all station data regardless of the year. A minimum of 30 years was required for inclusion. 

Fig 5. Comparison between the hottest day of the year when looking at only data since 1981 (top row) and all years (bottom row). In each row, the left map shows the interpolated categories and the right map show the interpolated categories with the station data overlaid. 

Friday, March 3, 2017

Wettest and Driest Months / Seasons

What are the wettest and driest months of the year? Using published normal values, we can answer that question. These maps were generated using the 1981-2010 NCEI climate normals for monthly precipitation within the U.S. In Canada and Mexico (and the rest of the globe [not shown]), the GHCN v.2 monthly precipitation was used.

To standardize the months due to differences in the number of days, I used an average daily precipitation value. In a few instances, this will cause discrepancies. For example, if February averages 3.00" of precipitation and March averages 3.10", I show February as having more precipitation – since their per-day value is higher.

On the mapping side of things, how do you interpolate a month between two stations where one is wettest in December and the other is wettest in January. If you average the month numbers (12 and 1), you end up with June being the wettest month! Obviously this is not correct. I used trigonometric functions to compute these polar-coordinate averages. Still, there are boundary issues when nearby stations have vastly different wettest or driest months.

There are four sets of maps. 

1) Wettest / Driest month of the year (color shaded)
2) Wettest / Driest month of the year (station dots)
3) Wettest / Driest season of the year (color shaded)
4) Wettest / Driest season of the year (station dots)

Note: Seasons are based on calendar months (e.g., winter = December through February).

Wettest / Driest month of the year (color shaded)

Figure 1. Wettest month of the year based on climate normals (color shaded).

Figure 2. Driest month of the year based on climate normals (color shaded).

Wettest / Driest month of the year (station dots)

Figure 3. Wettest month of the year based on climate normals (station dots).

Figure 4. Driest month of the year based on climate normals (station dots).

Wettest / Driest season of the year (color shaded)

Figure 5. Wettest season of the year based on climate normals (color shaded).

Figure 6. Driest season of the year based on climate normals (color shaded).

Wettest / Driest season of the year (station dots)

Figure 7. Wettest season of the year based on climate normals (station dots).

Figure 8. Driest season of the year based on climate normals (station dots).

Thursday, August 4, 2016

Global Köppen Climate Classifications

In our daily lives we experience weather from one day to the next but we also have an expectation of the weather for certain times of the year. This expectation is a our sense of the local climate. For example, someone living in Chicago, IL, knows that temperatures will not reach 70°F in the middle of January; however, their experience tells them that 70°F high temperatures are possible in April, May, and September. On this level, we all have an sense of the climate for the place we live based on our life experiences.

Köppen Climate Classification system:

From a scientific perspective, it is convenient to develop guidelines for describing the climate that are consistent and comparable. The most famous of all climate classification systems is the Köppen Climate Classification system. The Köppen Climate Classification system (http://en.wikipedia.org/wiki/Köppen_climate_classification ) was developed in 1884 by Wladimir Köppen and subsequently refined by others. The beauty of the Köppen system is that the categories are solely based on temperature and precipitation. The distinction between categories is determined by the magnitude and the seasonal distribution of those categories.

Köppen broadly grouped regions into 5 categories using letter designations. They are: 

Tropical (A) – temperature of coolest month 18°C (64.4°F) or higher.

Desert (B) – complicated criteria based on precipitation. This category supersedes all others.

Mesothermal (C) – temperature of warmest month greater than or equal to 10°C (50°F), and temperature of coldest month less than 18°C (64.4°F) but greater than -3°C (26.6°F).

Microthermal (D) – temperature of warmest month greater than or equal to 10°C (50°F), and temperature of coldest month -3°C (26.6°F) or lower.

Polar (E) – temperature of warmest month less than 10°C (50°F).

Each of these major groupings can be subdivided into sub-groupings based on finer-scale temperature and precipitation data. A great description of the criteria for each major and sub-grouping can be found here: http://www.britannica.com/EBchecked/topic/322068/Koppen-climate-classification . The map below is from Wikipedia and shows the major and minor groupings for ll of North America. The legend uses the A through E major groupings and each major grouping has different color shades representing the minor groupings. 

Figure 1. Köppen Climate Classification map of North America from Wikipedia. Categories were derviced from 1951-2000 data obtained from GHCN v.2.

New Climate Data:

The National Center for Environmental Information (NCEI) archives temperature and precipitation for thousands of stations across the globe – not just the U.S. The Köppen Climate Classification system utilizes monthly data for bot temperature and precipitation. The data utilized for this exercise are the Global Historical Climatology Network (GHCN) version 2 precipitation data and the version 4 temperature data. Oddly, temperature and precipitation are stored in separate files. The stations in the two data sets have slightly different coordinates, identification numbers, periods of record, and even station names. This makes the linkage between the two data sets (temperature and precipitation) unusually difficult.

My solution was the start with the temperature data set. All stations that had at least 20 complete years of temperature data between 1981 and 2010 were used. The total number of valid temperature stations was 9,419. Next, I identified all precipitation stations with at least 20 complete years between 1951 and 2010. I expanded the range of years under the assumption that monthly precipitation does not change enough over long periods of time to reflect a changing climate. Since most of the precipitation criteria involve the relationship between wet and dry season, the total precipitation is not overly important. Also, limiting the range of years to 1981-2010 produced too few matches. The total number of precipitation stations was 12.675 using the expanded criteria.

Now that the temperature and precipitation data sets were isolated, I needed to link them together. As noted earlier, the ID numbering system is completely different between the two data sets and no table exists to cross-reference the stations. In addition, the location precision is different between the two data sets so the dots on the map are not perfectly aligned. Therefore, I performed a spatial search. If a precipitation point was within 10 miles of a temperature point, the two records were combined. It's not a perfect arrangement, but it's the best I can do for now.

A total of 3,918 stations have monthly temperature and precipitation data in a suitable form for Köppen analysis. The next map (Figure 2) shows the major grouping for each of the 3,918 stations in the U.S. Unlike the map from Wikipedia, I decided to leave to data in point form. 

Figure 2. Köppen Climate Classification major groupings based on GHCN climate data from NCEI. Click on map to see a larger version.

For large sections of the globe, homogeneity in the rule. However, since Desert (B) classifications supersede other classifications, the distinction between categories in the dry areas can be difficult to put a boundary around. When the map data is changed to show minor categories, this boundaries between categories are too complex to generalize with a choropleth map. The next map (Figure 3) shows all 3,918 stations with their Köppen Climate Classification minor groupings.

Figure 3. Köppen Climate Classification minor groupings based on GHCN climate data from NCEI. Click on map to see a larger version.

Google Earth:

While static maps are an excellent way to communicate spatial information, an even better method is to let uses browse the data for themselves. I put together a .kmz of the data for all 3,918 stations. This is a Google Earth file that can be opened by anyone with the Google Earth software installed on their computer. The file can be downloaded at the following location: 

Once the file is downloaded and opened using Google Earth, the user can navigate to any station to see their classification type and all of their monthly normals from NCDC by clicking on the dot. The final map (Figure 4) shows a screenshot of the stations loaded into Google Earth. The station information for Madrid, Spain, is displayed in the figure.

Figure 4. Google Earth screenshot of the Köppen Climate Classification placemarks file.


Here is the link to the Google Earth file with Köppen Climate Classifications for all 6,405 stations in the U.S. that have 1981-2010 climate normals.   koppen.kmz 

Thursday, June 16, 2016

Daylight-Twilight-Astronomical Maps

Here is my collection of maps showing length of daylight and twilight. They are in no particular order.

Hours of daylight on the winter solstice.

Difference between the hours of daylight on the summer solstice and the winter solstice.

Difference between the hours of daylight on the summer solstice and the winter solstice (Alaska focus).

Summer solstice sunset (local time).

Date of earliest sunrise and latest sunset in summer.

Date of earliest sunset and latest sunrise in winter 2015-2016.

Date of latest sunrise and earliest sunset by latitude.

Hours of daylight on the summer solstice.

Hours of daylight plus civil twilight on the summer solstice.

Average length of daylight plus civil twilight for all days of the year.

Average length of daylight plus civil twilight for all days of the year. Same as Fig. 6 but a N. America perspective and slightly different map categories.

Average length of daylight plus civil twilight for all days of the year. Same as Fig. 6 and Fig 7 but an Alaska-centered perspective.

Average length of daylight for all days of the year.

Average annual hours of 1) daylight and 2) daylight plus civil twilight by latitude in the northern hemisphere.

Number of days per year wit 24 hours of daylight.

Number of days per year wit 24 hours of daylight plus civil twilight.

Hours of daylight on the summer solstice. Northern Hemisphere perspective.