I want to build an automatic garden/lawn watering system based on a small computer. The system is connected to the Internet and is able to learn the current weather conditions and the short-term forecast. What I need for this project is the ability to determine the soil moisture content using only the readily available weather data.

One may argue that a set of humidity sensors buried into the ground would be the best choice, however the cost of such sensors is prohibitive for a small gardening project (around $300 each). Custom made sensors are cheap and easy to build, but the accuracy is low and it degrades over time.

I found a Chinese study but that is of limited help due to its large scale. Does anyone have a clue about how to calculate the moisture content?

  • Since this is not an answer to your question, I make it just a comment. Is this for learning purposes or just for fun for having it? The reason I ask this, is because nature doesn't need sensors or weather forecasts. If you want to keep your soil moist enough during seasons, you can better use wood chips as mulch. The Back to Eden film shows this and explains why this is the most natural way to keep your soil moist. But if you want to learn how to build such systems, go ahead. Should be a lot of fun! – Aschwin May 3 '14 at 9:23
  • @Aschwin, I am looking for a way to water my garden, but given my propensity for gadgets, I would like to make this a little project. So it's a fun project with a real use. – vbocan May 5 '14 at 18:39

Do searches and read up on information on Evapotranspiration. Weather stations around the USA record local ET data. You'll want to water enough to restore the water lost to ET. This is a decent article to give you some info http://www.rainbird.com/landscape/resources/articles/Irrigation-Scheduling.htm

On my site I also have some lawn watering tips which includes some info on ET.

Soil probes should give you more accurate info on your current soil conditions and would be easier to program because you won't need to find and interpret your local ET data (which may differ slightly from your actual conditions). It drastically reduces the number of variables you need to account for. Did a quick search for arduino soil moisture sensors and prices ranged from $2 to $8.


Exactly what Bamboo said. Often it happens that rain is forecasted for a city, and it rains in one part of the city but skips the other side of town altogether. How could you account for this without some sort of sensor? You'd have no way of knowing whether rain actually fell in your garden. Also, if you are just working based on weather data, you aren't accounting for your soil. How much of the new moisture does your soil retain after a rain? Does it all wash down a hill before it has time to soak in? Does it drain away in sandy soil before the plants can soak it up? Or does the soil hold it a very long time? The weather patterns won't tell you any of that.

It seems to me you'd be better off with a less-than-accurate homemade soil sensor.


I can't see how its possible without sensors, frankly. A weather prediction is of limited value on its own, clearly, but what can't be accounted for are the various microclimates present within any single garden - some parts will be hotter and dryer than other parts, depending on various factors such as wind/sun exposure, humus content of the soil, rain shadow, and so on, so even if you could work out a general guide for watering for the whole area, there would still be inconsistencies. I'll be interested to read other answers.

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.