Announcement

Collapse
No announcement yet.

New OpenSprinkler Plugin

Collapse
X
 
  • Filter
  • Time
  • Show
Clear All
new posts

    #61
    Originally posted by frankc View Post
    But, I am not using the qpf figure. I need to fix that. Maybe reduce the amount of irrigation by that qpf figure?
    That's the key question, and I would also like to know the answer too. As a starting point, it's why I thought it worthwhile to investigate how accurate the forecasts actually are. For instance, early today WU was predicting 0.25" total precipitation for today, but we actually had 1.48". I'd say that's a pretty large variance between forecast and actual. If that's typical of forecasts, then maybe a good strategy is to irrigate the bare minimum (and only if it's absolutely needed) to get through the next day or two to see how the rainfall actually pans out. Then once all the facts of how much rain actually precipitated are known, you could irrigate more to supplement, if needed. That's the sort of strategy that I'm currently thinking might waste the least amount of water, but I'd be very interested in a brainstorm on other possible strategies, of which you have already mentioned one. So, if anyone can suggest any other possibly worthwhile strategies, please do jump in and say it.
    Last edited by NeverDie; October 11, 2014, 11:03 PM.

    Comment


      #62
      I agree. Error on the side of water conservation.

      And, yes, the stuff I pasted came from here:

      http://api.wunderground.com/api/---your_key_here ---/forecast/q/KS/SHAWNEE_MISSION.json

      You don't need a key to get it, but, without the key, the URL format is a little different. But the keys are free and give you 500 hits a day.

      That is the place Python goes to to get the forecast. And I've looked at it several times in the past and the 0% probability and then significant amount of precip happens often. I am sure it makes sense to people who understand what it actually means.

      Comment


        #63
        Thanks for the link.

        Looks as though the forecast for tomorrow there has already shifted: "day":12, "month":10, "year":2014, "yday":284, "hour":19, "min":"00", "sec":0, "isdst":"1", "monthname":"October", "monthname_short":"Oct", "weekday_short":"Sun", "weekday":"Sunday", "ampm":"PM", "tz_short":"CDT", "tz_long":"America/Chicago" }, "period":2, "high": { "fahrenheit":"64", "celsius":"18" }, "low": { "fahrenheit":"55", "celsius":"13" }, "conditions":"Overcast", "icon":"cloudy", "icon_url":"http://icons.wxug.com/i/c/k/cloudy.gif", "skyicon":"", "pop":10, "qpf_allday": { "in": 0.61, "mm": 15 }, "qpf_day": { "in": 0.00, "mm": 0 }, "qpf_night": { "in": 0.61, "mm": 15 }

        Anyone please do jump in to correct me, but I'm guessing the interpretation of the revised forecast might be: There's a 10% chance of rainfall >= 0.01" tomorrow. If it does rain, then the expected amount of precipitation is 0.61", all at night and none during the day.

        Come to think of it, maybe the earlier pop of zero that you reported wasn't really zero in the sense of 0.0000. Maybe it was a float that was greater than zero but less than 0.5, and it got rounded to the nearest integer (zero) for reporting purposes.
        Last edited by NeverDie; October 11, 2014, 11:52 PM.

        Comment


          #64
          Anyhow, the pop and qpf for Sunday may make more intuitive sense when you realize that lots of rain is almost a sure thing in your area on Monday based on the pop and qpf you reported for Monday, but that some of that rain might start on Sunday night--probably late on Sunday night, if at all

          Comment


            #65
            Originally posted by NeverDie View Post
            Come to think of it, maybe the earlier pop of zero that you reported wasn't really zero in the sense of 0.0000. Maybe it was a float that was greater than zero but less than 0.5, and it got rounded to the nearest integer (zero) for reporting purposes.
            But you notice that the pop for the next day was 90. So had the probability been, say .6%, the pop would have been displayed as 1. And it would seem that if someone calculated that the actual precip (qpf) would be worthy of mention, the conclusion that the pop would be at least 1% seems reasonable.

            (I don't get it) Anyway, that is why I had chosen to ignore the qpf and require the pop to be above 40. Not sure how to handle that.

            Comment


              #66
              ET Spreadsheets

              I found some spreadsheets that might have value. I haven't looked at them yet, but, these look real promising.

              http://extension.uidaho.edu/kimberly...-calculations/

              The example that the Python script follows is this:

              http://www.fao.org/docrep/x0490e/x0490e08.htm
              (example 18)
              That is the example the spreadsheets also follow.

              Also:

              http://www.kimberly.uidaho.edu/water...etmain2005.pdf

              appendices for above:

              http://www.kimberly.uidaho.edu/water...i/appendix.pdf
              Last edited by frankc; October 12, 2014, 11:31 AM. Reason: added another link

              Comment


                #67
                Originally posted by frankc View Post
                But you notice that the pop for the next day was 90. So had the probability been, say .6%, the pop would have been displayed as 1. And it would seem that if someone calculated that the actual precip (qpf) would be worthy of mention, the conclusion that the pop would be at least 1% seems reasonable.

                (I don't get it) Anyway, that is why I had chosen to ignore the qpf and require the pop to be above 40. Not sure how to handle that.
                I agree it's better to know with certainty than to have lingering doubts. Therefore, seems like it's worth posting the snapshot you captured to the WU forum and appending your query to it

                Comment


                  #68
                  Originally posted by frankc View Post
                  I found some spreadsheets that might have value. I haven't looked at them yet, but, these look real promising.

                  http://extension.uidaho.edu/kimberly...-calculations/

                  The example that the Python script follows is this:

                  http://www.fao.org/docrep/x0490e/x0490e08.htm
                  (example 18)
                  That is the example the spreadsheets also follow.

                  Also:

                  http://www.kimberly.uidaho.edu/water...etmain2005.pdf

                  appendices for above:

                  http://www.kimberly.uidaho.edu/water...i/appendix.pdf
                  Thanks! The more the merrier.

                  I'm making progress on learning the basics of python, and afterward I'll be circling back to take a deeper pass at both this and the scripts above that I struggled getting to run.

                  Comment


                    #69
                    My HomeSeer (2.5.0.81) execution environment is a 6 year old HomeTroller 2 (XP and .Net 2.0) so loading Python, json, et al is probably not a great option.

                    I have translated the Python code at https://github.com/frankpc3/ET into VB and am using the wunderground XML data files.

                    The code appears to be working but, at this point, I need to validate the VB results against a known Python results.

                    At first blush, it seems that comparing the daily balance files would provide a valid test assuming the values for the 'window' and the 'city' are the same.

                    Would anyone be willing to share this data?

                    After validation, I will be happy to share the VB code with anyone who is interested.

                    Best Regards,

                    Mark

                    Comment


                      #70
                      Welcome, Mark. Good idea.

                      You may be slightly ahead of things, though, as I'm not sure to what degree the python results themselves have been validated, and if they were, what the touchstone was. Maybe frankc knows? So far, to the extent validation checks have been done, it would have been frankc or David who did them. As the code is all fairly recent, I wouldn't at all be surprised if the validation work is still largely pending. If it's yet to be done, then maybe the Texas A&M ET website (earlier link), or perhaps a Davis weather station that's been setup to do Davis ET (e.g. earlier link) might serve as touchstones? Ideally, the FAO people would have published a test suite of inputs and official results from using their chain of formulas (above link).... Maybe they did, and I simply didn't find it on my cursory search. So, you may want to also try taking a look yourself to see what you can find. If none of those, then maybe one of the spreadsheets (above links or another spreadsheets yet to be found)....? There definitely exist ET spreadsheets out there, but finding one that's free and downloadable has been a challenge. Maybe you'll get lucky with Bing, or perhaps a fresh google search will turn up some more.... If you want to help rather than wait, it would plainly expedite progress toward your goal....
                      Last edited by NeverDie; October 13, 2014, 04:14 PM.

                      Comment


                        #71
                        Thanks for the welcome.

                        I've started looking at the 'solar' calculations. I've found noaa calculations here http://www.esrl.noaa.gov/gmd/grad/solcalc/solareqns.PDF which are of the same form as in the Python code but the constants in the Python code appear to have been modified by various factors for convenience (losing all traceability to original sources). I'm working to convince myself the logic in this area will give valid answers when compared to an independent workbook.

                        Comment


                          #72
                          Hello. Yes. We've been working on the Python script. I am pretty sure you would be amazed at how well the Dallas Texas A&M posted results match those of the Python script. I have plotted them in Excel. I will try to post the file here. I will have to zip it first. It has two charts and the actual A&M data and the Python results data.

                          Welcome Mark. I would appreciate seeing your vb code. I need to update my Python script on github I guess. Although it might make more sense to post it here.

                          I use Windows XP with Homeseer and several other applications that run 24/7. So I'm thnking XP isn't a problem. And this isn't a fast computer by any means.

                          This thread was started by Mattyjee for his OpenSprinkler Plugin. A friend is planning to buy one and is planning to use Matyjee's plugin here for it. I hope the Python script will work with it.

                          I'm continuing to improve the script. Today, I wrote some code for it to get actual irrigation time from the Homeseer log. I've been using the script for perhaps 6 days and it is working fine. The problem I've had is I didn't know what Python was two weeks ago.

                          Comment


                            #73
                            Excel file

                            Spreadsheet attached showing how the Python script results compare to Texas A&M online calculations for solar radiation and ET.

                            My first attachment here. Let me know if it didn't work. Or maybe the wrong file. who knows?
                            Attached Files

                            Comment


                              #74
                              Originally posted by frankc View Post
                              Spreadsheet attached showing how the Python script results compare to Texas A&M online calculations for solar radiation and ET.

                              My first attachment here. Let me know if it didn't work. Or maybe the wrong file. who knows?
                              Your spreadsheet and graph download and open fine.

                              Did you pull the input parameters into python from weatherunderground or somehow from the Texas A&M website? The reason I ask is that you've been referring to it as Dallas, whereas the link I had provided you earlier in this thread was their ET readouts from Austin. So, if you're pulling your data from Dallas rather than Austin, that might explain some of the variance.

                              Comment


                                #75
                                <<I use Windows XP with Homeseer and several other applications that run 24/7. So I'm thnking XP isn't a problem. And this isn't a fast computer by any means.>>

                                The HomeTroller 2 is a dedicated diskless/fanless box which shipped as XP SP2. Since I don't use it for anything but HS 2, I have left it unaltered (more or less).

                                Beyond just converting the Python code to VB, I'm pondering how far to take the conversion. There are a number of 'stylistic' things I would want to change, e.g. all the configuration information should be in a INI file and the code should be broken up into logical procedures and functions which are easier to manage.

                                Of course, the further it migrates away from the original Python code, the more difficult it would become to incorporate future updates to the Python code.

                                BTW, the variable totalClearSkyIsolation is probably meant to be totalClearSkyInsolation - Insolation is used in other variable names and is, apparently, a technical term in solar radiation. Also, the variable previousCloudCover is uninitialized on the very first pass in getHistorical data.

                                At the moment, I have a basic 8 zone Rainbird system, and as a first pass I'm looking for the application to provide a go/no go output which I would use to disable the Rainbird - next year of course, winter is just around the corner in NJ :-).

                                Comment

                                Working...
                                X