Cumulus Support

archae86 wrote:Lately I've been toying with the idea of adding a UV and a Solar sensor. As this would run about US $350 plus shipping and a bit of setup effort, I am rather curious as to what I'd get before making a final purchase decision. To that end I'd appreciate links to both Wunderground Cumulus generated data and to direct Cumulus generated web sites for Davis installations with solar, UV, or both.

At the moment, from a UK perspective, the solar sensor isn't too expensive - I imported mine from the US - but the UV sensor is very expensive. I have bought a solar sensor (I imported it from the US), but have yet to buy a UV sensor. I have read elsewhere that Davis are developing a cheaper UV sensor.

My WU data is here: http://www.wunderground.com/weatherstat ... =IOrkneys2
and my own web site (using the solar readings to estimate hours of sunshine) is here: http://sanday.org.uk/weather/

According with the Davis supplied documentation,

Each sensor is calibrated against a Yankee Environmental Systems’ Ultraviolet Pyranometer, model UVB-1, in natural summer daylight.

This makes every sensor a bit expensive because can exist a high rate of discarded sensors that may not accomplish the calibration specs.

http://davisnet.com/product_documents/w ... 490_SS.pdf

http://www.yesinc.com/products/data/uvb1/index.html

Hope this helps your choice.

Happy New Year

steve wrote:My WU data is here: http://www.wunderground.com/weatherstat ... =IOrkneys2
and my own web site (using the solar readings to estimate hours of sunshine) is here: http://sanday.org.uk/weather/

Thanks. I'm surprised and pleased to see that Weather Underground makes a within day variation graph from the data you submit to them from your solar sensor using Cumulus. Comparing, for example, the June 22, 2011 to the November 22, 2011 curves gives a really dramatic indication of the winter/summer difference in the Orkneys even with sun shining.

I see that Cumulus does not, however, supply a graph to the trends page it makes for your own web site from the solar data, so I'll assume not from the UV data either.

I'm nothing like a good programmer, but have knocked together odds and ends and made tiny to moderate modifications in the code of others for something like forty years in dozens of languages. Would you guess I might be able to fiddle the existing Cumulus bits to add generation of solar and UV graphs to the Trends page of the generated website? Or would that require going farther under the hood than a rank amateur could likely succeed?

archae86 wrote:I see that Cumulus does not, however, supply a graph to the trends page it makes for your own web site from the solar data, so I'll assume not from the UV data either.

That's right; it's on the very long list of 'things to do'. You can generate graphs for solar data in 'Select-A-Graph', but there is currently no automated mechanism.

I'm nothing like a good programmer, but have knocked together odds and ends and made tiny to moderate modifications in the code of others for something like forty years in dozens of languages. Would you guess I might be able to fiddle the existing Cumulus bits to add generation of solar and UV graphs to the Trends page of the generated website? Or would that require going farther under the hood than a rank amateur could likely succeed?

There are libraries to generate graphs using PHP etc, so I'm sure this must be possible. See http://cumulus.tnetweather.com/projects/jpgraphgraphs for example.

steve wrote:You can generate graphs for solar data in 'Select-A-Graph', but there is currently no automated mechanism.

I'm not in the current habit of using Select-A-Graph and had completely missed that it would allow me to see the solar data from my installation.

Since it appears that the Wunderground upload works well, and that is my actual primary mechanism for sharing data (almost no one looks at my Cumulus-generated website, at least according to the hit counter I have installed there), and Select-A-Graph will let me see my own data, I plan to go ahead with the purchase, and probably won't try to brew my own add-on.

Thanks for the helpful responses.

I did in fact buy solar and UV sensors, and the Davis sensor shelf from Rainmanweather. The installation instructions were clear enough but some tolerances were a bit snug. In particular getting the last (of five) cables in through the port in the bottom corner of the SIM was quite a bit of bother. Much less serious, but still a little awkward, was that fitting the top of the rain gauge back on, which has never posed a problem, was somewhat tight.

Happily all came up and worked right away, though I did not find all the places to enable solar and UV reporting in Cumulus plus Weatherlink at first try. (I stop Cumulus twice a week and run Weatherlink just to download the data--which gives me two copies of the data, and the ability to look at the past in either). It also took me a while to add appropriate modifications to the Cumulus template files for the various pages of the website Cumulus generates for me. I currently have at least a little solar data on the index (a.k.a now), today, and yesterday pages of this web site:
Lynx Stoll weather by Cumulus

Weather Underground posting worked fine as soon as I enabled the Cumulus solar and UV checkboxes in the Configuration|Internet|Weather Underground sections.
Lynx Stoll Wunderground page

The first two days my sensors were in operation were very nearly cloudless, aside from some thin late afternoon cirrus. This allowed a look at the cloud-free light curve, and comparison to the "max solar" curve provided by Cumulus for my location and these dates.

1. Near winter solstice, the sun rises in my location far enough south to be blocked by my next-door neighbor's house, so there is an abrupt kink in the curve about 8:47 a.m., which would otherwise kink about 8:15 a.m. when it would clear the Sandia mountain range.

2. Any time clouds, the neighbor's house, or mountains were not in the way, the solar sensor recorded considerably more light than the Cumulus max solar curve. While this could be a mis-calibrated sensor, other Davis Vantage Pro2 owner data persuades me that my sensors are not high outliers. Possibly the curve was built not expecting that air could be as transmissive as it is here. Or possibly there is an error somewhere. Here is a screenshot of the data for the solar radiation sensor at my station here in Albuquerque, NM for January 11, 2012 with the solar max curve reference.

3. The UV index data for my sensor has resolution steps of 0.1, but as it tapers off at the end of day, it steps right from 0.5 to 0.0 A review of the Wunderground postings for some Davis Vantage Pro 2 Plus stations near here and also near Phoenix suggests this is their usual behavior, and would also be present in the rising curve in the morning were it not that the abrupt transition as the sun clears my neighbor's house makes the initial step even larger. (just to be clear, I think this 0.5 cutoff is a Davis, not a Cumulus, matter)

Cumulus thoughts and possible upgrade opportunities:

1. If you follow me, you may wish to upgrade to the latest release of Cumulus before starting to tamper with things, as there are optional sections for solar data in the templates. But you may wish to consult the HTML tag section of the Cumulus help and consider adding more that the template files provide.

2. I'd like to think that the evaportranspiration data could be helpful in controlling irrigation or watering, but the HTML tags mentioned in the current (1.9.2) release Cumulus help only seem to provide ET since beginning of current day. Were one provided for ET since, for example, the beginning of the current year, then in principle a neighbor could write down the rain year to date and ET year to date each time he watered his plants, then in considering whether and how much to water again, could just subtract the currently displayed year-to-date values from the ones written down at last watering (most likely with a bias factor multiplied by the displayed ET to allow for differing plant type, wind sensor offset, ...)

This is not really a barrier for my personal use, as the Davis console displays both Rain year to date and ET year to date at the press of a couple of buttons, but much of the fun of posting weather data is the hope it could be useful to someone else. I'd like consideration to be given to supporting in a future Cumulus release HTML tags for Yesterday's ET, and year-to-date ET.

I suspect some past summary statistic would be good to have for UV and solar radiation data as well, but it is less obvious which would be most appropriate or helpful. To me, personally, it would make sense to have the integration across the observing day of both Solar radiation and UV, but apparently these are not traditional measures to report in weather circles. So I can't reasonably ask for this.

archae86 wrote:2. Any time clouds, the neighbor's house, or mountains were not in the way, the solar sensor recorded considerably more light than the Cumulus max solar curve. While this could be a mis-calibrated sensor, other Davis Vantage Pro2 owner data persuades me that my sensors are not high outliers. Possibly the curve was built not expecting that air could be as transmissive as it is here.

This is where the 'transmission factor' comes in; you need to adjust it until your curve fits under the theoretical maximum most of the time.

2. I'd like to think that the evaportranspiration data could be helpful in controlling irrigation or watering, but the HTML tags mentioned in the current (1.9.2) release Cumulus help only seem to provide ET since beginning of current day. Were one provided for ET since, for example, the beginning of the current year, then in principle a neighbor could write down the rain year to date and ET year to date each time he watered his plants, then in considering whether and how much to water again, could just subtract the currently displayed year-to-date values from the ones written down at last watering (most likely with a bias factor multiplied by the displayed ET to allow for differing plant type, wind sensor offset, ...)

This is not really a barrier for my personal use, as the Davis console displays both Rain year to date and ET year to date at the press of a couple of buttons, but much of the fun of posting weather data is the hope it could be useful to someone else. I'd like consideration to be given to supporting in a future Cumulus release HTML tags for Yesterday's ET, and year-to-date ET.

Unfortunately ET is problematic. The Davis DLL returns strange values from time to time which results in the figures going negative or very large (mine is currently showing -5.13 mm for today). As I result, it's not practical to do anything sensible with the ET figure from Cumulus. I don't know where the problem is, I'm assuming it's the DLL rather than the console firmware. I was hoping that version 1.90 of the firmware would fix the problem as the release notes mention fixing an issue with ET, but I have 1.90 installed and it hasn't helped.

I suspect some past summary statistic would be good to have for UV and solar radiation data as well, but it is less obvious which would be most appropriate or helpful. To me, personally, it would make sense to have the integration across the observing day of both Solar radiation and UV, but apparently these are not traditional measures to report in weather circles. So I can't reasonably ask for this.

I do hope to do more with solar data at some point in the future.

Thanks for the comments; it's always good to get reasoned, constructive feedback.