Refreshing my memory about tropical light and temperature

22 November 2016

On a visit to Manila some time ago, I visited a golf course and had a look around with the golf course superintendent. Our discussion turned to seasonal changes in weather and the relative impact on the grass. For example, in the winter, when the temperatures are a little cooler, is it lower temperatures that have a big impact on slowing the growth? Or might it be the light, because the days in winter are a little shorter and the sun is a little lower in the sky?

I calculated what the photosynthetic light would be on a sunny day in each month.

That's how the photosynthetic photon flux density (PPFD) will be throughout the year when there are no clouds. It varies a bit.

For the total light per day we can look at the daily light integral (DLI), and this is how the DLI changes through the year.

The PPFD is how much light reaches the turf every second. Adding together the light from each second, from sunrise until sunset, gives the total amount for the day -- the DLI.

What about temperature? Manila is 13 degrees north of the equator with a tropical climate. The coldest month is January, with an average temperature of 25.6°C. The hottest month is May, with an average temperature of 29.5°C. That's a difference of 3.9°C from the coldest to the hottest month.

The sunny day DLI ranges from about 44 to 59 mol m-2 d-1. So which changes more, the temperature, or the light?

To look at that, I plotted the standard score (z-score) for the DLI and for the temperature across 12 months. The z-score shows how many standard deviations a value is from the average (mean).

If the z-score is less than 0, that means the value for that month is less than the average for the year. If the z-score is more than 0, that means the value for that month is more than the average for the year.

Because the z-scores are standardized, I can compare directly the DLI and the temperature, and how much they are changing in any one month, compared to their average value during the year.

At Manila on sunny days, the temperature varies relatively more than the light in January, February, April, and May. In March, and from June through December, the light varies relatively more than the temperature.

Animated chart of potential and actual PAR at Batesville for the first 214 days of 2016

03 August 2016

The photosynthetically active radiation (PAR) changes at a location by time of year and by cloud cover. This animated chart shows what the average photosynthetic photon flux density (PPFD) would be for every 5 minute interval in clear sky conditions, and also shows what the actual PPFD was. Adding up the PPFD from every second gives the total PAR for a day -- the daily light integral (DLI). This chart shows potential DLI under clear sky conditions, as well as actual DLI.

Is it normal to be cloudy like this?

02 August 2016

On July 17, I was in the Tokyo area with Jim Brosnan. The daily light integral (DLI) in Tokyo on July 17 was 14.2 mol/m2. Jim asked me if it was exceptionally cloudy that day. Not really, I answered. I told him that the such cloudiness was normal.

Now that July 2016 is over, I looked at the DLI for every day in July at Tokyo and also at Batesville, Arkansas. Both are at about 35.7°N latitude, so the day lengths will be identical.

The lowest DLI at Batesville in July was 22.8 mol/m2 on July 29. In Tokyo, there were 10 days in July with a DLI less than 22.8 mol/m2, including 5 days with a DLI less than 10 mol/m2. In that context, the cloudiness on July 17 was not exceptional.

To see more, check out the average hourly PPFD and DLI values for Tokyo in this chart and for Batesville in this one.

Warm-season turfgrass growth rates and competition at 35°N

01 August 2016

Mike Richardson pointed out that the growth rate of zoysia is less than bermuda, so by implication there must be something other than growth rate that allows zoysia to invade bermuda. That is, in the situations when bermuda and zoysia are growing together -- competing -- when zoysia appears to grow faster, Mike suggests it may be a factor such as turf density that allows such a result, because bermuda grows faster than zoysia.

I've outlined a hypothesis about grass growth rates and their required inputs, and have more to write about that later. In that hypothesis, I mention location, and in my recent discussion with Mike about the growth rate I said that there is a variety by climate interaction. By climate, I mean the same as location. I'll use these words interchangeably.

Let me try to explain what I mean by an interaction by climate. I'll use data from Tokyo, and from Batesville (2016 data) and Fort Smith (climatological normals data). These locations are all about 35°N.

Light, temperature, plant water status, and leaf nitrogen content all influence growth. In turfgrass management, light and temperature generally can't be controlled; plant water status and leaf nitrogen content can be modified by turfgrass managers. We can imagine that bermuda and zoysia are growing side by side, or together, and then think of what may happen with modifications to these growth-influencing factors.

On average, this is the part of the climate that can't be controlled, at Fort Smith and at Tokyo, shown in 2-dimensional space.

That's a similar temperature range but different amounts of sunshine. Thus, there is no overlap during the months when warm-season grasses are growing. I focus on light and temperature because the water and the nitrogen can be adjusted by the turf manager.

Temperatures for 2016 are pretty similar through July 30. I express temperature here as the cumulative sum of growing degree days.

Ok, so temperatures are similar. If it were only temperature that influences growth, one would expect the grasses to perform pretty much the same at these locations. If bermuda does have an inherently faster growth rate than zoysia, then in this side-by-side comparison, with the same temperature, then bermuda should grow faster at both locations.

I downloaded the global solar radiation data also and then converted it into photosynthetic radiation units. This is Batesville for the first 7 months of 2016.

This is Tokyo for the first 7 months of 2016.

In 2016, there has been more photosynthetic light at Batesville than at Tokyo.

The DLI was pretty much the same from January to March, but since the start of April Batesville has jumped ahead by about 1,000 moles/m2. In the past 4 months, Tokyo has accumulated about 4,000 mol/m2 and Batesville has accumulated about 5,000 mol/m2. That's a log percentage difference of 22%. The difference has been especially pronounced in June and July -- the hottest months of the year so far.

Imagine growing bermuda and zoysia in 10% shade at the same temperature. Bermuda may grow faster than zoysia. Now imagine 20% shade. Probably the same result. How about 30, 40, and 50% shade? 60% or 70% shade? At some point, the growth rate of zoysia will be greater than the growth rate of bermuda. The bermuda will die in shade under which the zoysia can still produce a turf.

Consider now that there are varying growth rates among bermudagrass varieties, and also among zoysia varieties. That's what I mean by the location (or climate) by variety interaction. Take an inherently faster-growing zoysia, mix it with bermuda, grow it in a climate with high temperatures combined with lower DLI, mow the grass and make sure plenty of water is applied during the dry season, and see which one grows faster. It's not bermuda.

Yes, with a high DLI, plenty of fertilizer, moderate water supply, and high temperatures, bermuda grows faster than zoysia. Here's a photo of the ATC research facility putting green during grow-in. It's easy to tell which plots are zoysia -- those closest to the camera.

But if one thinks of growth as something that happens over years, at a location, with the grasses maintained as turf, then one can find the growth rate of zoysia can be higher than that of bermuda.

I find it useful to look at growth rate in those terms, rather than trying to explain it as a response to density or as competition for some other factor.

Turfgrass and shade: daily light integral (DLI) in Sydney

07 July 2016

The Australian Government Bureau of Meteorology (BOM) provide satellite-derived global solar radiation data. I downloaded 2015 and 2016 data for station number 66120 (Gordon Golf Club). The data are in energy units of megajoules per square meter per day. I multiplied by 2.04 to convert to daily light integral (DLI) units of moles per square meter per day.

This is the DLI in full sun, adjusted for clouds. Any tree or structural shade will result in a lower DLI.

Looking at monthly summaries of DLI, one can see the median and the normal range in each month since January 2015.

I downloaded temperature data from the Sydney Airport (SYD) and used those to calculate an estimated DLI using the Hargreaves equation, as described in Estimating daily light integral in 4 Tennessee cities. I did not make any corrections to the estimate from the Hargreaves equation, and SYD is about 25 km south of Gordon. Still, the uncorrected Hargreaves equation gives a decent estimate of DLI.

Checking my calculations

28 June 2016

I enjoyed reading the recent paper by Hodges et al. on Quantifying a daily light integral (DLI) for establishment of warm-season cultivars on putting greens. They measured the DLI at Starkville for the duration of this experiment, from 13 June to 29 September 2013 and again from 2 June to 27 September 2014. The mean DLI in full sun, on their test area, was 42.3 mol m-2 d-1 when averaged across those dates.

Last year I made some calculations to estimate DLI. You can read about that in Estimating daily light integral in 4 Tennessee cities. I wondered what that calculation method would give for a mean estimated DLI in Starkville. That is, Hodges et al. measured DLI with a quantum light sensor from Spectrum Technologies, and I wanted to check my calculations to see how close the estimated DLI was to the measurement.

The code for the calculations is in the dli_tn repository.

In full sun, Hodges et al. measured an average DLI of 42.3. The mean estimated DLI, using my calculations, for those same dates, was 40.6. Not too far off. To put the error of my estimate into context, that's a difference of 1.7 moles. An hour of midsummer midday sun at that location will have about 7.2 moles of PAR per hour, so 1.7 moles is equivalent to about 15 minutes of midsummer midday sun.

17 May 2016

Shade from trees will often reduce the photosynthetically active radiation (PAR) by about 80%. For example, the area in full sun on this green had a photosynthetic photon flux density (PPFD) of 749 micromoles of photons per square meter per second.

It is possible to make a good estimate of the effect of shade, and to know just how much PAR is reaching the turf, without using a meter. Take this putting green, for example, with part of it in sun and part in tree shade.

The PPFD in sun (with no clouds) can be estimated by knowing the day of the year, the time of the day, the latitude, and the longitude. This Shiny app makes the calculation based on those inputs. The calculated PPFD by that app is pretty close to the measured PPFD. Here are some calculated PPFDs compared with measured PPFDs from sites in full sun, unobstructed by clouds.

If you are in full sun with no clouds, then you can get a good estimate of PPFD from the app.

If there is tree shade, I'd assume that the PPFD in shade is 20% of that in full sun.

If there are clouds, I'd look for my shadow and look for the sun, to get an estimate of how much the clouds are reducing the PAR, as described here.

Grass selection by normal temperature and sunshine hours

18 April 2016

Plotting the normal temperatures and sunshine hours for a location places that location in a particular 2-dimensional space. I demonstrated that in these charts. @turfstuf suggested that a diagonal line might show a break point for classifying warm and cool-season grasses.

The idea is that the top right would be warm-season, the area around the line would be transition zone, and the area to the bottom left would be cool season. That chart looks like this.

I agree that different regions of the chart are indicative of over/under points for different grasses or growing conditions. I wouldn't separate by that diagonal line. Here's the break points I would use.

• mean annual temperature less than 15°C, cool-season
• mean annual temperature from 15 to 20°C, transition zone
• mean annual temperature above 20°, warm-season

For those general breaks, one can estimate the annual mean from the monthly charts, or plot the locations by the mean annual temperature.

Continuing with the breaks, specifically looking at which warm-season grasses will be suitable:

• within warm-season, and more than 6 hours sunshine per day, bermudagrass
• within warm-season, and less than 6 hours sunshine per day, zoysiagrass or other warm season grasses that are tolerant of low light conditions: bermudagrass will struggle
• within transition zone, and less than 6 hours sunshine per day, if warm-season grasses are used, zoysiagrass or other warm season grasses that are tolerant of low light conditions: bermudagrass will struggle

A transition zone location like Atlanta looks like this when those points are marked on the plot.

Two warm-season locations, one where bermuda thrives (Honolulu) and another where bermuda is overgrown by more shade tolerant grasses (Hilo), are shown here.

In the next plots I show some other locations: cool-season, warm-season, and transition zone. The break points I use seem to agree pretty well with grass distribution and performance around the world.

Visualizing climate differences

16 April 2016

Of the factors that influence plant growth, turfgrass managers are able to modify in some way the plant water status and the nitrogen supply to the grass, but they can do little to adjust the temperature and the light. As a consequence, both the grass adaptation to a particular environment, and the management requirements for the grass, will be influenced or controlled by the combination of light and temperature.

I spoke about this at a conference in 2012 and shared this handout. From the start of the handout:

The weather, and specifically the temperature and the amount of sunshine, has a major influence on the growth of grass and therefore on the suitability of certain grasses for certain climates. By plotting the climatological normal weather data with temperature on the horizontal axis (x-axis) and sunshine hours on the vertical axis (y-axis), we can see which locations are similar in these parameters, and thus likely to be suitable for the same grasses, and to similar maintenance practices for grasses. Many locations in East, South, and Southeast Asia are distinguished by relatively low sunshine duration as compared with locations of similar temperature in North America, Oceania, Africa, and Europe. For additional information about the use of these charts, see www.climate.asianturfgrass.com.

The idea is that when temperature and sunshine are the same (or similar) at two or more locations, the growing conditions, and the energy available for grass growth, are the same (or similar). When the temperature and sunshine are different, with no overlap, then the growing conditions are clearly different.

I think this is interesting and informative because such an approach can help to identify places that we might think are similar, but are in fact different, and vice versa. The implications for maintenance requirements, grass selection, and location to location comparisons are also evident from such representations of climate data. I've made some more plots to illustrate this.

I start with Miami. The normal monthly mean temperature is shown on the x-axis and the mean daily sunshine hours for that month are shown on the y-axis. The polygon defined by each of the 12 months of the year expresses what the normal growing environment is like at Miami. Places that are similar to Miami should have overlap in light and temperature with Miami. Places that are different should have little or no overlap.

Moscow, for example, has no overlap with Miami. I don't think anyone would expect it to.

The hottest months of the year at Moscow are cooler (with more sunshine) than the coldest months at Miami. There is no overlap between these locations.

New York City has some overlap with both Miami and Moscow. If I plot New York on this chart, I can see which months at New York are similar to Miami or Moscow.
June in New York is similar in temperature and sunshine to March in Miami, September in New York is almost the same as January in Miami, and July and August in New York are between March and October conditions in Miami. One can also see the seasonal overlap between New York and Moscow conditions.

How about another warm season location like Miami? This plot adds Singapore conditions.

There is no overlap between Singapore and Miami, even though both are warm-season locations. There is more overlap between New York and Miami (about 3 months) than there is between Singapore and Miami (0 months). This has implications for grass selection and management. That is, the grasses the work in Singapore may not do very well in Miami, and vice versa.

Some places are predictably similar. Portland, and Seattle, for example, have almost complete overlap.

Other locations that one might expect to be similar have no overlap at all. I often use Honolulu and Hilo as an example. And sure enough, one finds different grass species growing at these two locations.

This video discusses Hilo and Honolulu.

One can also look at transition zone locations, like Atlanta, where both warm and cool-season grasses are grown.

How does a cool-season location like London compare to Atlanta?

June and July in London are similar to March and November in Atlanta. July and August in London have similar temperatures as October in Atlanta, but with less sunshine.

Melbourne is another transition zone location, with golf course fairways and sports fields usually planted to warm-season grasses, and golf course putting greens usually planted to cool-season grasses.

Melbourne has some overlap with both Atlanta and London.

Hanging around 1750 all summer

17 March 2016

I looked at the photosynthetically active radiation (PAR) in Corvallis and Ithaca for each day of 2015, and there was something strange in the data. I didn't think anyone would notice, but someone caught it right away:

The station -- I've used data from the U.S. Climate Reference Network -- latitudes at Corvallis (44.4°N) and Ithaca (42.4°N) are similar, so I expect the maximum radiation to be similar. In fact, on average during summer I'd expect Corvallis to receive more radiation, and Ithaca less, because of the climate differences between those two locations.

What's going on? Are my calculations wrong? Is it sensor error? Were there forest fires producing smoke over Corvallis through the summer of 2015? Was it exceptionally sunny in Ithaca, and cloudy in Corvallis? I decided to look at more years, getting data from every year since 2007 for comparison. I looked at the daily totals.

Something strange happened with the data for Corvallis in 2015. In all the previous years, there were some days with global solar radiation above 27 megajoules per square meter per day (27 MJ m-2 d-1). In 2015, nothing. At first I'd thought that the Ithaca data were abnormally high. But in looking at the data from 2007 to 2015, it seems that Ithaca is within a normal range of global solar radiation, but Corvallis data are abnormally low.

I counted the days with global solar radiation greater than or equal to 27 MJ m-2.

In the 8 years prior to 2015, Corvallis had 4 years with more than 40 days above 27 MJ m-2, and only 1 year (2011) with less than 20 days reaching that level. Then in 2015, 0 days.

I'm guessing this was an undetected sensor problem or other data error. The photosynthetic photon flux density (PPFD) in Corvallis probably should not be hanging around 1750 all summer. It was in 2015, because of the data I was working with. But after a closer look, those data seem abnormally low. I'd expect the PPFD in Corvallis to be blasting through to 2000 in early summer. Can anyone with a quantum meter in the Willamette Valley confirm this?