You certainly need to have a time series for each sensor. However, it is important to understand that the soil water migrates in the direction towards lower humidity/soil water content. This is normally at the surface in dry weather in areas where irrigation is necessary to grow crops. The migrating water carries chemicals in solution, and these are left behind in the soil when the water evaporates. Accordingly, measuring soil salinity/ electrical conductivity is a reliable method of tracing moisture movement in these soils. The measurements are more accurate than most soil moisture sensors (what type are you using?) and the sensors can be hooked up to a reliable multi-channel data logger and the data downloaded when convenient.  By taking measurements every 30 minutes and graphing the results, you can readily see the effect of diurnal temperature changes, precipitation events, etc. on the distribution f soil moisture.

An additional benefit of this method is that it will show when the soil salinity becomes a problem as it almost always does in irrigation projects. The only way to void this is by using drainage ditches or pipes.  The salinity measurements will aid in the design of successful planning of the necessary drainage to avoid and control the build up of salinity in the soil.

Stuart

Stuart,

Thanks for your answer. As I'm just joining this project I don't think my team have considered the salt effect you have mentioned. I'll ensure that problem has been solved before continuing working with the data. 

The five of them are all dielectric sensors (Sentek "Soil Moisture Sensors"). I do have the times series for each of them, including measurements made every 15 minutes and recorded through almost two years, for the most complete dataset I have access. Additionally, I have datasets from different fields where sets of 4 sensors of the same type and at the same depths have been deployed.

I'll consider the option you said, but for now I'm afraid I have to choose among those that are already in use and decide which of them to keep on the field in order to reduce maintenance cost and still have a "good enough" approach to the whole soil humidity level in the root zone.

Yes, take a good look at the variations from one string of sensors to the next at the time scale you are interested in and pick some typical ones if possible.  If they are all different, pick some examples covering the extremes as well as average ones. However your employers have to realize that to get good results, they cannot cut back too far and expect no consequences.  Good luck!

Stuart.

Yes ,I am also agree with Mr. Stuart but one of the major thing is that, as plant grows its root water abstraction depending upon plants feeder root. As plant grows tap roots grows deeper and feeder roots remain near the surface (In case of agronomy crop). Growth of the feeder root will also vary depending upon the  irrigation methods. Like in Drip irrigation feeder roots remains near the surface and most of the water abstraction takes pace near the surface so ultimately soil moisture variation takes place near the surface. Therefore my suggestion is to take close look on which type of crops on the field and its root growth and then decide which sensors should be aggregated.

In order to answer your question better, it would be good to mention whether your irrigation planning model considers a depth-resolved soil moisture profile. I guess not because the question of how to aggregate is not really necessary in this case. Therefore, I assume that you are using a "bucket" model to represent plant available water. In this case, I would say that the bucket needs to represent the root zone. The depth of the root zone should be a function of  crop type, phenology, and perhaps the soil profile. Hope this helps.

I would also like to comment on the use of electrical conductivity (EC) for water content determination as suggested above. Although promising and interesting, it is actually not easy to translate EC in water content in a quantitative manner as needed here when both water content and the solute concentration are changing. Therefore, dielectric sensors are often a better choice if point or profile information would be sufficient. The dielectric permittivity is strongly correlated to water content, and this correlation is relatively unaffected by EC if working in non-saline soils and high enough frequency (>100 MHz), such as with Time Domain Reflectometry. For other sensors, you may want to check whether changing EC with constant water content affects the sensor reading.

Best,

Sander

If you want to run a model for predicting moisture distribution temporally and spatially within root zone ,you need some parameters rather than measured data. The measured data can be used for validating the running model. To represent the irrigation model, spatio-temporal information is required. So, the sensors locations are spatial information. At the same time you need a time series information . The location of the sensors are perfect .

The depth you install your sensors should take into consideration the depths which are to be explored by the rooting system of the crop under consideration in Kenya for annual crops we routinely observe Soil Moisture upto to 1 metre for most annual crops. Thus sensor are installed at 5cm 10cm 20 cm, 50 and 100cm.

If the crop is a depth rooted crop observe in addition to the above 150 and 200 cm i.e upto 2 meters. Case of tree crops and areas where the plant may benefit from water table capillary rise the installations of these sensor may be varied appropriately.