measured soil moisture - clarification

[DrasticAction]

hi Michael.

I have a question about the behaviour of mcsSprinklers when using measured soil moisture as the control strategy, which I have not been able to determine by observation.

Initially I am working on using the csv file interface mechanism to provide the feed of soil moisture data (being the simplest approach for me).

It has just occurred to me that this could create a risk, in that the system will not be fail-safe against breakdowns in the sensor feed. If the soil moisture values in the csv file are not updated, presumably mcsSprinklers will continue to trust the (unchanging) moisture level, and never enter a watering cycle.

My question is: would this be overcome by using xAP to feed the sensor data? More specifically, in the absence of a new xAP packet to update the soil moisture value, will mcsSprinklers continue to apply the soil moisture loss calculated using the ET method?

If this is so, it provides a strong incentive for me to do the extra work and develop an xAP interface for my home-grown sensor network (which is proving to be quite a challenge for my meagre VB.NET skills). I am concerned about the potential impact of a sensor failure in the CSV scenario.

Other than applying a "Maximum Days without Watering" setting, is there any other way to detect or protect against a breakdown in the soil sensor feed? My installation is standalone, so I don't have HS to apply external rules or generate alerts, unless I build somethng into the sensor network.

[Michael McSharry]

There is monitoring for internet based data for weather and forecast with email notification if update are not received. There is not monitoring for specific sensor values no matter what the delivery mechanism is.

An advantage of a distributed architecture with the xAP communication is that it easly supports independent application and monitoring functions. If the assurance level needs to be high for a specific functions then there shoulld be a monitor for that function. In your example you consider the moisture reading to be an important measure so you should have something looking that has a model of an expected soil moisture pattern and if that pattern within some tolerance is not being received then an alarm raised.

Modeling of sensor behaviors is not always easy and could easily result in false reports. For this reason you may find it better to monitor the output of the irrigation control than trying to model the inputs. In essence the simple model provide by mcsSprinklers about maximum dry duration tries to achieve this objective in an operational rather than an alerting manner.

Smarter algorithms can be developed that provide a better monitor and with the xAP interface the input or output data will be available to feed that monitor.