Systems and methods for determining moisture content of soil

Abstract

A method for determining moisture content of soil within a measurement zone of a field includes transmitting and / or receiving radio frequency signals between a plurality of spaced apart nodes and receiving, by a controller, signal data about the intensity of radio frequency signals transmitted and / or received from each of the plurality of spaced apart radio nodes. The method transmits, by the controller, the signal data about the intensity or other characteristic of radio frequency signals to a server and the relative signal strength of the radio frequency signals transmitted between each of the plurality of spaced apart nodes. The moisture content of soil for each location between each of the plurality of spaced apart nodes within the measurement zone is then determined.

Description

Systems and Methods for Determining Moisture Content of Soil FIELD OF THE INVENTION

[0001] The present invention relates to systems and methods for determining moisture content. More particularly, the invention relates to systems and methods for remotely determining moisture content of soil in a measuring zone.

[0002] The invention has been developed primarily for use as a system for determining moisture content of soil in a field using a network of radio frequency nodes. While some embodiments will be described herein with particular reference to that application, it will be appreciated that the invention is not limited to such a field of use, and is applicable in broader contexts. BACKGROUND

[0003] The following discussion of the prior art is intended to facilitate an understanding of the invention and to enable the advantages of it to be more fully understood. It should be appreciated, however, that any reference to prior art throughout the specification in no way be considered as an admission that such art is widely known or forms part of common general knowledge in the field.

[0004] The measurement of soil moisture is extremely important to agriculture, yet there are few effective ways to measure soil moisture in an accurate, inexpensive and timely manner. Capacitive soil moisture probes, for example, only sample soil near the probe. Capacitance is hyper local (on the mm scale) and therefore a large number of probes are needed to accurately measure a property. Neutron probes are considered the standard for soil moisture measurement, but this technology is extremely expensive. Remote methods are being developed, but are based on drone or satellite imagery and are infrequent. Satellite data is low resolution in both spatial coverage and temporal coverage (often weeks or months) and does not provide real-time or up-to-date information.

[0005] The present invention seeks to ameliorate the above disadvantages, or provide a useful alternative by providing an inexpensive, scalable and robust soil measurement system and method that can take frequent measurements over large areas with high accuracy. SUMMARY OF THE INVENTION

[0006] According to a first aspect of the invention, a method for determining moisture content of soil within a measurement zone, including the steps of: transmitting and / or receiving radio frequency signals between a plurality of spaced apart nodes; receiving, by acontroller, signal data about the intensity of radio frequency signals transmitted and / or received from each of the plurality of spaced apart radio nodes; transmitting, by the controller, the signal data about the intensity / characteristic of radio frequency signals to a server; determining, the relative signal strength of the radio frequency signals transmitted between each of the plurality of spaced apart nodes; and determining, the moisture content of soil for each location between each of the plurality of spaced apart nodes within the measurement zone.

[0007] Preferably, the method further includes the step of redirecting a received radio signal from at least one of the plurality of spaced apart nodes to at least one of the plurality of spaced apart nodes, thereby to define a local measurement zone.

[0008] According to a second aspect of the invention, there is provided a system for determining the moisture content of soil within a measurement zone, including: a plurality of spaced apart nodes disposed in at least one of: within, at a periphery and adjacent to the measurement zone, each of the plurality of nodes configured to transmit and / or receive radio frequency signals, a controller in communication with each of the plurality of spaced apart radio nodes; and a server in communication with the controller; wherein the controller is configured to receive signal data about the intensity of radio frequency signals transmitted and / or received from each of the plurality of spaced apart radio nodes, and transmit the signal data to the server; and wherein the server is configured to determine the relative signal strength of the radio frequency signals transmitted between each of the plurality of spaced apart nodes, and calculate the moisture content of soil for each location between each of the plurality of spaced apart nodes within the measurement zone.

[0009] Preferably, the system further includes at least one reflector disposed within the measurement zone, to redirect a received radio signal from at least one of the plurality of spaced apart nodes to at least one of the plurality of spaced apart nodes, thereby to define a local measurement zone. More preferably, the plurality of spaced apart nodes form at least a partially connected mesh network.

[0010] Each of the plurality of nodes preferably includes at least a transceiver configured to transmit and / or receive the radio frequency signals and / or transmit the signal data to the controller. In one embodiment, each of the plurality of nodes includes at least a processor, an antenna and a power source.

[0011] In one embodiment, the system includes a plurality of reflectors to define a plurality of local measurement zones

[0012] In another embodiment, the server is configured to output a map representing moisture content of soil within a measurement zone, based upon the determination of the moisture content of soil for each location between each of the plurality of spaced apart nodes.

[0013] Preferably, the server is configured to output a map representing moisture content of soil within the local measurement zone.

[0014] In one embodiment, the radio frequency signals have a frequency in the range of 100kHz to 100MHz. In another embodiment, the radio frequency signals have a frequency in the range of 100kHz to 10MHz.

[0015] In yet another embodiment, the radio frequency signals have a frequency in the range of 150MHz to 1GHz. In one embodiment, the radio frequency signal is 433MHz.

[0016] Preferably, each of the plurality of spaced apart nodes are located in the range of 0 to 10 m above the ground. A distance between any of the plurality of spaced apart nodes is preferably in the range of 0.01km and 10km. Preferably, a distance between any of the plurality of spaced apart nodes and the controller is in the range of 0.01km to 10km.

[0017] In one embodiment, any of the plurality of spaced apart nodes is configured to operate as a controller.

[0018] In another embodiment, the server uses the determined moisture content of the soil between the plurality of spaced apart nodes to output a control signal to an irrigation system. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] Preferred embodiments of the disclosure will now be described, by way of example only, with reference to the accompanying drawings in which:

[0020] FIGURE 1 is a schematic of the system for determining the moisture content of soil within a measurement zone.

[0021] FIGURE 1B is a schematic of the system for determining the moisture content of soil within a measurement zone according to another embodiment, with reflectors.

[0022] FIGURE 2 is a graph of the water content vs relative received RF power according to an embodiment of the invention.

[0023] FIGURE 3 is a schematic diagram of one of the plurality of nodes of the system according to an embodiment of the invention.

[0024] FIGURE 4 is a schematic diagram of the system according to an embodiment of the invention.

[0025] FIGURE 5 is a schematic diagram of the layout of the system according to an embodiment of the invention.

[0026] FIGURE 6 is a graph comparing the 915MHz radio frequency signals and water content near a sensor according to an embodiment of the invention.

[0027] FIGURE 7 is a graph comparing the 433MHz radio frequency signals and water content near a sensor according to an embodiment of the invention.

[0028] FIGURE 8 is a graph comparing the 433MHz radio frequency signals of FIGURE 7 with the water content near a sensor of Figure 6 according to an embodiment of the invention.

[0029] FIGURE 9 is a graph comparing the 433MHz radio frequency signals of FIGURE 7 with the water content near a sensor of FIGURE 6 according to another embodiment of the invention.

[0030] FIGURE 10 is a comparative graph showing the water content as measured by a reference sensor against the radio frequency sensor according to an embodiment of the invention.

[0031] FIGURE 11 shows moisture-RSS relationship in another preferred embodiment in conversion of a series of single moisture measurements into a spatially oriented map.

[0032] FIGURE 12 shows a graph of Tx / Rx path is parallel to the X-axis in the embodiment of FIGURE 11.

[0033] FIGURE 13 shows a graph of Tx / Rx path is parallel to the Y-axis in the embodiment of FIGURE 11.

[0034] FIGURE 14 shows a pixelated error estimation maps over a predetermined block pixelated field of the preferred embodiment of FIGURE 11.

[0035] FIGURE 15 upper shows a 3-D simulated moisture content map of a field as pixelated into blocks shown below for the embodiment of FIGURE 11.

[0036] FIGURE 16 is a graph of calculated poles and errors for the moisture simulation in the embodiment of FIGURE 11.

[0037] FIGURE 17 is a graph of moisture value estimates in the embodiment of FIGURE 11.

[0038] FIGURE 18 is a graph of moisture value estimates in the embodiment of FIGURE 11 with a lower set variance than FIGURE 17.

[0039] FIGURE 19 is a graph of estimation errors in the maps of FIGURE 15.

[0040] FIGURE 20 shows graphs of error estimation of the map of FIGURE 15. DETAILED DESCRIPTION

[0041] Referring initially to FIGURE 1, there is a system for determining moisture content in soil, including a plurality of spaced-apart nodes, configured to transmit and receive radio frequency signals. The spaced apart nodes are located around a field, and form a network, where each node may transmit and receive radio signals at one or more frequencies, and where each node is able to communicate with each and every other node to measure every possible iteration. Each node may include a transmitter, a receiver, or both. Each node may transmit a radio signal and each of the other nodes reports back the signal strength so that all nodes had measured every node pair. Frequencies used could be 2.4 GHz, 915MHz, 900MHz, 400 MHz, 433MHz, 169MHz, 28 MHz or open bands in a given country. Identifying available radio frequencies is important as there are very few frequency ranges that are available internationally. These include 900 MHz, 2.4 GHz, and the citizen bands (CB) around 30 GHz and 400GHz. The radio frequency signals may have a frequency in the range of 100kHz to 100MHz. Alternatively, the radio frequency signals may have a frequency in the range of 100kHz to 10MHz. In some embodiments, the radio frequency signals may be in the range of 150MHz to 1GHz.

[0042] The nodes may also be referred to as stations. It will be readily appreciated that a person skilled in the art may select frequencies and frequencies of bands as required.

[0043] The field may also include a radio reflecting object, such as a reflector, within the field to define local measurement zones to discriminate between different areas of the field. This enables inspection of moisture content in localized areas of soil. The reflector / s may be repositioned within the field to further aid in characterising and / or determining the moisture content of the field. The reflector may include a passive repeater or a passive radio link reflector. The reflector may take the form of a flat metal sheet or as determined by a personskilled in the art. In some embodiments, one or more reflectors may be at least partially buried in the ground or on the soil surface, and placed in such a way as to increase the number of paths measured in the network / across the field to increase the resolution of the measurements across the measurement area.

[0044] After all signal data is collected by the plurality of nodes it is sent to a controller. The controller may also be referred to as a hub. The controller may be a gateway. The controller collects the signal data related to the strength of the radio frequency signals transmitted / received by each of the plurality of nodes. In one embodiment, the controller includes a at least one LoRa module. In some embodiments, the at least one LoRa module is a Sparkfun Pro RF wireless board. The controller may be configured to send the signal data on to a remote server at predetermined time intervals. In some embodiments, the controller may transmit signal data every hour to conserve power.

[0045] The controller sends the signal data on to a remote server where the soil moisture measured over the space between two nodes would then be processed to calculate the soil moisture over the entire field. In some embodiments, the soil moisture is measured over the space between a plurality of nodes, either simultaneously or sequentially. Regarding simultaneous operation we also refer to FIGURE 1B as described further below. Then tomographic inversions may be applied to the measurements. In other embodiments, the determination of soil moisture across a field is processed within the controller. This may include taking an average of the soil moisture measured over the space between two nodes. In some embodiments, the whole-of-field moisture content of a field may be presented as a heat map.

[0046] In some embodiments, the remote server may include a computer. In some embodiments, the remote server may include Raspberry Pi. The remote server may be controlled via a 4G or 5G network.

[0047] Moisture content of the soil in the field is calculated from the change in transmitted signal intensity over the soil, with this intensity known to vary with soil humidity in a predictable manner. In some embodiments, the moisture content of the soil in the field is calculated from the change in transmitted signal intensity in the ground. In other embodiments, moisture content of the soil in the field is calculated from the change in transmitted signal intensity above and below the ground. That is, the skin-depth of the propagated electro-magnetic signal across the measurement area of the field will depend on soil content and other properties. It is important to understand the difference between digital signals, where Received Signal Strength Indication could serve as the power dissipationover distance, versus analogue signal strength meter readings common to CB (citizen band) communications for example.

[0048] FIGURE 1 also shows a simplified version or model where ‘x’ station transmits to every other station, and where possible a reflector creates a localised measurement zone.

[0049] The reflector is used to redirect the radio signal and create a local measurement zone to discriminate between areas of a field. Some fields (or paddocks) are quite large and the use of reflectors to break the paddock up into smaller segments provides information on local conditions, versus a very long distance.

[0050] FIGURE 1B shows another version or model of the invention, in which the plurality of nodes are disposed around a periphery of the measurement zone. Each node can transmit to every other node, and receive from all other nodes simultaneously, creating a mesh network. As shown in FIGURE 1B, the layout may include buried reflectors within the measurement zone to create a plurality of local measurement zones. The reflectors may alternatively be partially buried or placed on the soil surface,

[0051] The invented solution enables measurement over meters of distance (up to tens of meters), can be arranged over many areas of a field while communicating regularly (as often as required), and uses inexpensive equipment.

[0052] The invented solution is not localised, not expensive and does not require much power. Moreover, the invented solution relies on common electronics for both measurement and telemetry. Importantly, the use of a mesh network and the use of Received Signal Strength Indication (RSSI) would greatly simplify data collection as this is a common network measurement and built into most equipment. In other embodiments, the signal intensity may be measured and reported using a relative measure such as dB (decibels). Preferably, the network is set up to enable data analysis to identify issues and resolve them. For example, if one node is compromised, it can be removed from the data analysis without affecting the infrastructure of the network or the results of the other nodes.

[0053] In some embodiments or versions, the server may use the determined moisture content of the soil between the plurality of spaced apart nodes to output a control signal to an external system. The external system may include an irrigation system, where the whole- of-field moisture content of the soil may be used to control the amount of water dispersed by the irrigation system on the field. Additionally, the control signal may be used to control the amount of water dispersed by the irrigation system within localised measuring zones on the field. In some embodiments, aspects or versions of the invention may be included on anirrigation system, such as an irrigator. For example, a plurality of nodes and / or controllers may be included on an irrigator, to take measurements of soil moisture and control said irrigator based on the soil moisture. For example, controlling the irrigator to disperse more water on an area with low soil moisture than on an area with high soil moisture as measured by the system.

[0054] It will be appreciated that the system may be used for a number of other advantageous applications. In other embodiments, the system could be used to determine when the soil is saturated, and could be used, for example, to monitoring before and after natural disasters such as flooding and river levels. Additionally, it could be used to monitor local weather and moisture / rain patterns, or monitor livestock movement. In other applications, it could be used to monitor leaks in pipes underground, for example local saturation in a field / measurement area by a leaking irrigation pipe or supply pipe to a livestock water trough.

[0055] The use of radiofrequency (RF) to detect soil was examined in 1973 by Chadwick (D. G. Chadwick, “Integrated Measurement of Soil Moisture by Use of Radio by Use of Radio Waves,” Utah Water Research Laboratory, 1973.), which showed a positive correlation between the signal strength and the soil moisture in a sensing area.

[0056] The early, limited and primitive study by Chadwick indicated that using RF to sense soil moisture has a potential for the soil sensing industry. The ability to sense moisture in a wide area without direct contact sensing would be economically advantageous as compared to the traditional direct, single point sensing method, which is expensive to use and would also be more costly to implement in a large area.

[0057] Without wishing to be bound by theory, vertically polarised RF waves travelling on the Earth’s surface induce charges on the surface of the soil. These charges induce current flow through the soil surface, which behaves like a leaky capacitor (leaky current). The amount of current “leaking out” represents the amount of RF strength of the vertically polarised wave being attenuated. The amount of induced charge depends on the characteristics of the soil. For example, soil conductivity and dielectric constant, which directly relates to soil moisture.

[0058] As the dielectric of the soil increased, the number of induced charges reduced, and resulted in less power dissipation of the RF signal strength. In other words, as the moisture content of the soil increased the RF signal strength decayed less as compared to when the soil was dry.

[0059] According to the Chadwick reference, the relationship between RF signal strength (power received over expected transmitted power – P) and soil moisture (or water content in soil) could be modelled by the following formula:(1)

[0060] Thus, the value P was the function of antenna separation distance, the electrical conductivity, and the moisture of the soil above which the RF wave was travelling. As the ratio of receiving over transmitting power would always be less than 1, the value P would always be a negative value in dBm (dB scale for milliwatt power). The use of the dB scale is consistent with the current RF power measurement standard. From here on, the value P is called relative RF signal strength.

[0061] A brief modelling graph is shown in FIGURE 2 of the formula of P. It shows the relationship between relative RF signal strength and water content (soil moisture). Here, the frequency was 915MHz, the distance was 1.2 m and the electrical conductivity was 5x10-4S / cm (arbitrarily chosen based on the RF module to be used and the substrate of sand for the field testing).

[0062] The inventors examined the use of radiofrequency in inferring the soil moisture (soil water content) in an area between two RF transceivers. The field testing examined the effectiveness of soil moisture measuring with a radio frequency of 433 MHz as compared to a 915MHz frequency.

[0063] In actual field testing the inventors surprisingly found that the 433MHz sensor prototype produced a data trend that was highly correlated to the soil moisture data trend. A915 MHz sensor prototype was also part of the testing for comparison and the produced data trend did not show any significant correlations for the 915 MHz prototype tested.

[0064] The testing opened a new set of questions and possible experiment designs to better refine the relationship between RF and soil moisture. Future works such as frequency tuning, sensor characterisation, calibration, sensing network setup and so on are some of the possibilities to be explored.

[0065] The equipment used in the field testing was made up of off-the-shelf components with a few hardware modifications to work. The prototype used a 433 MHz. Surprisingly, the data trend provided by the new prototype with the lower frequency was highly correlated to the soil moisture.

[0066] FIGURE 3 shows a diagram of a prototype system. The addition of the SDI-12 sensor (hereinafter referred to as the Steven soil moisture sensor) was included to provide a reference for comparison with the moisture content measured using the signal strength of the radio signals.

[0067] To collect RF strength data, another prototype was made which housed all different LoRa Modules (915 MHz, 433 MHz and 169 MHz) – hereinafter called “the hub”. As the result, the hub had three different antennas and one antenna was also used to send data back to the server device for data storage.

[0068] The 915 MHz module of the hub also communicated with the server by setting the frequency to 910 MHz to avoid collision with the 915 MHz prototype on the field. FIGURE 4 shows an initial server’s system diagram.

[0069] The server device would poll data from the hub every hour to conserve power at the test site. The server itself was powered with the main supply and remotely controlled via 4G network and a VPN network.

[0070] The chosen test site had little to no vegetation and was enclosed within a fenced area, making it ideal for testing to be free of unknown interference such as growing grass and wildlife. The prototypes were also made to operate remotely, independent of a constant supply of mains / grid electrical power as compared to the previous experiment.

[0071] A further set-up of the experiment was as shown in FIGURE 5. The results of FIGURES 6 to 10 are with respect to the FIGURE 5 set-up and as described herein. During the initial setup, the goal was to space the sensors as far as possible given the shared space available for testing and they would be moved back slowly if there were issues with datacollection. The data coming from 433MHz and 915MHz prototypes were robust and left untouched to get a few weeks of interrupted data. However, it was noted that the distances of 433MHz and 915MHz from the hub were not the same, accordingly appropriate scaling of the measurement area / field with the frequency propagated may be required. It will be readily appreciated that experimental set-up of FIGURE 5 may also be done as a single frequency. In FIGURE 5 two frequencies were used for the convenience of testing at the two frequencies in a field as described herein.

[0072] Steven moisture reference sensors were attached to the 433MHz and 915 MHz sensors and were placed near these two sensors respectively to collect soil moisture data as a reference measurement. The sensors were buried about 5 cm deep in the soil to quickly show any changes to the moisture due to watering events (such as rain).

[0073] The data was collected over 3 weeks, with the first few days of dry weather conditions, suitable for baselining and then the following weeks of occasional raining events. This range of weather conditions was very good for determining if there was a good correlated data trend between RF and soil moisture.

[0074] FIGURES 6 and 7 show the result of the field testing. The left black outline box shows the baseline event and the right blue outline box show rain events. The reference Steven moisture sensor results are the lower graph.

[0075] FIGURE 6 is the graph of the data for the 915MHz prototype. It was apparent that the data trend of the RF signal strength was not well correlated to the soil moisture data even though it reacted to a rain event.

[0076] FIGURE 7 is a graph showing the data for the 433MHz sensor and its respective Steven sensor data (located nearby). Unfortunately, the Steven sensor (adjacent the 433MHz sensor) data seemed to be corrupted, which may indicate a faulty operation of that Steven sensor. Interestingly, the RF graph looked very similar to the moisture data from near the 915MHz sensor. As the field test area was relatively small, a rain event would distribute water to the whole area evenly. Thus, the 433MHz sensor data was then compared with the Steven sensor data from FIGURE 6, which is shown in FIGURE 8.

[0077] As can be seen in the graph of Figure 8, the data appeared very correlated, especially at the end data level of RF when it was higher than the baseline event. The baseline noise of the RF was within about 2dBm and this noise level was present throughout the data trend.

[0078] FIGURE 9 shows a graph of the data of the 433MHz sensor from the baseline event to a time period after three rain events at the site. The data trend appeared very consistent and correlated with the reference soil moisture measurements.

[0079] A graph mapping the RF signal to soil moisture is shown in FIGURE 10. This was done by using a mapping method that included removing the offset of the RF graph and then re-scaling the RF data against the soil moisture data, as shown in the below equation:

[0080] However, it is noted that this equation did not account for other effects such as temperature, humidity and area of sensing. It will be readily appreciated that normalizing for such may be done. It is readily apparent that the changes in soil moisture in the field sensed by the invention at 433 MHz correspond with those of the reference Steven sensor.

[0081] In the testing, two radio frequencies – 433 MHz and 915 MHz – were used with the nodes to sense moisture content of soil in a field. The 433 MHz module showed a good correlation with the reference soil sensor data. The data from the 433 MHz sensor also appeared to agree with the theoretical equation used by Chadwick Reference.

[0082] The invention may be further developed and improved by: • Experimenting with the distance between the 433 MHz and the hub: by varying the distance, a better understanding of the limitation of the sensing could be made for a prototype. • Experimenting with the power of RF: the present experiment only had the power setting to be at the lowest (0 dBm gain for both Tx and Rx) whereas the module could have up to 20 dBm of gain for each side (effectively it could reach 40 dBm). This might help with the lower end of the moisture level as the SNR (signal to noise ratio) would be significantly improved to provide a better reading (less noise). Additionally, the higher the RF power, the further away the sensors could be placed. • Experimenting with the distance scaled to wavelength instead of meters: this could help to further verify if 915 MHz sensor would not still work. The current data on the 915 MHz was ranging from -110 dBm to -95 dBm, which was a lot lower than the data of 433 MHz. An experiment with distance in terms of the wavelength of theused frequency would be engineeringly sound. E.g. if 433MHz at a similar wavelength would still be working. • Experimenting with varying humidity and temperature only. • Experimenting with a lower frequency – 169 MHz module could be further developed to be sufficiently operating on the field.

[0083] In another preferred embodiment, the conversion of a series of single moisture measurements are converted into a spatially oriented map. Here, the moisture content is sought of 4.5m×4.5m field blocks and the entire field is 90m×90m; and the Tx / Rx separation distance is 90 meters. Unlike the first embodiment of FIGURES 1 & 1B where each transmitter communicates with each other, each transmitter only communicates with a respective paired antenna as the field is divided into blocks.

[0084] Note that the lower part of Fig.15 shows the moisture-RSS / RSSI relationship between the transmitter and receiver 90 meters apart. Further, it is noted this embodiment employs the inverse of equation (1) to calculate soil moisture. It can be seen that the curves are not linear, but they are monotonically non-decreasing similarly to Fig.1. The inverse function w = h−(P, d) was determined and it is presumed that suitable soil conditions are satisfied but they cannot always be valid in practice.

[0085] GPR is derived from Bayesian inference and offers a non-parametric approach to data approximation. It is a versatile and robust method for regression that provides estimates while tolerating uncertainties and noises. It is presumed that RF sensors typically have weak powers and RSS signals are prone to electronic perturbance and noise. GPR can estimate the soil moisture from the measurements. The Radial basis function kernel (RBF) is used as kernel functions (4) for GPR. The RBF kernel offers smooth, continuous predictions. It is highly flexible, allowing it to map complex relationships and local variations in the data. Additionally, it is robust to noise and can serve as a universal approximator.where ℓ is a design parameter for the kernel function; ∥x – x’∥2 denotes the L2-norm difference between x and x’. The RF sensors are not mobile. Thus, we define the data point (x, y) from the sensors as x = k and y = w, where the moisture w is calculated using the k-th sensor (pair). Given the definition of GPR, we have the equation (5).where σnis a design parameter to model the noise variance; Xtrain= {k1, k2, ...} and Ytrain= {w1,w2}, where (ki,wi) is the i-th data point; Xtest= {1, 2, ...}.

[0086] The GPR parameters are separately designed for sensors parallel to the X-axis (ℓ∗ = 1 and σ∗n = 0.2) and the Y-axis (ℓ∗ = 0.99 and σ∗n = 0.21). The train data points are denoted as red stars in Fig 12. The estimations are denoted as the black dashed line and the real moisture content values are denoted as the blue line. Similarly, the sensors along the Y-axis provide 20 data points. This was repeated 20 times and obtained 400 train data points. The measurement noise had zero mean and 0.0663 standard deviation. It was estimated moisture content values along the Y-axis. The results are shown in 13. The confidence limit is denoted as the blue shades in Figs 12 and 13 where the limit provides a range within which the true soil moisture is expected to lie with a 95% possibility.

[0087] As regards the estimation of Soil Moisture in map form, noise was removed from the RF sensors and the averaged soil moisture obtained along Tx / Rx paths by (5) above. However, this does not directly represent the soil moisture of the field blocks in the lower part of Fig.15. Finite moisture samples in the field were to estimate the soil moisture as follows.

[0088] It was assumed that a moisture sample is denoted as (q,w), where q =< latitude, longitude > is the location of the sample and w is the moisture value. We thus define Xtrain = {q1, q2, ... }, Ytrain = {w1,w2, ... }, Xtest = { q̂1, q̂2, ... }, where (qi,wi) is the i-th moisture sample and q̂j is a location, at which the soil moisture is to be determined. By (5) above, the soil moisture estimation is obtained and hence the estimation of the soil moisture map.

[0089] The performance of noise cancelling and moisture mapping is sensitive to the magnitudes of the train data. Data normalization is used to improve the numerical stability by re-arranging the data on a consistent scale. As a result, in the preferred embodiment there is no need to retune the parameters for any new train data whereby normalization techniques are used for the training and test datasets such that the moisture map estimation aligns with the averaged soil moisture values and the samples.

[0090] GPR is primarily data-driven where a set of parameters apply to the same data. The process of parameter selection involves choosing the kernel function and its parameters and tuning its parameters is a critical step in noise cancelling and moisture mapping. In this preferred embodiment, the performance of the GPR method was dependent on the choices of σn and ℓ. The GPR method estimation errors were minimised and the confidence range maximised so that (7) yields the optimal σ∗n , ℓ∗where N̂ is the number of measurements beyond the GPR estimation limit. Given the test data Xtestand Y∗test, the optimal σ∗nand ℓ∗. The two items in (7) are equally weighted. Presently, Tx / Rx paths is used to divide the field into the grid (see Fig 14, right hand side for example). For example, if there are 2 × Nx sensors along the X-axis and 2 × Ny sensors along the Y-axis. Therefore, Nx + Ny measurements with all the RF sensors are collected. Simultaneously, time, the field is divided into Nx × Ny pixelized blocks. We are interested in estimating the moisture content values Ytest at Xtest ∈ {i|i ≥ 1}, and Ytest = h(P(i, k)) + ^, d(Xtest)), where P(i, k) is the RSS value of i-th sensor at k-th data point, d(Xtest) is its Tx / Rx distance, and ^ is the measurement noise.

[0091] Location estimation errors are calculated in the preferred embodiment and these are represented as EX = [E1, E2, ... , ENx ], EY = [ENx+1, ENx+2, ... , ENx+Ny ]. EX and EY to determine the estimation error. Additional moisture measurements can be gathered from moisture samples in the field, and the data used to train the GPR model. From this, appropriate locations for further sampling.

[0092] A series of numerical examples were used to demonstrate the performance of the proposed approaches. The estimation accuracy was verified in numerical cases under different moisture maps and the estimation errors of the GPR method was used to determine the performance in selecting the locations of moisture samples.

[0093] In the process of producing a simulated soil moisture map, soil moisture is dependent on many environmental factors. In particular, if the soil profile is relatively homogeneous in terms of composition, texture and structure, the soil moisture map at a given time can be represented as a function of space the temporal characteristics of the soil moisture map are disregarded and the GPR method used to provide a reasonable approximation in space. In this case, the soil moisture map could exhibit symmetry around a central point, making a Gaussian-like shape plausible.

[0094] In the preferred embodiment, simulations are used to demonstrate the effectiveness of the GPR-based approximation approach. The sum of Gaussian distributions defined by (12) represents the moisture content in the soil.where ℓ0is a constant; x =< latitude, longitude > is the coordinate of a given location in the field; x1, x2, ... are the coordinates of the predefined poles; each pole represents a normal distribution. The example in Fig 15 is used to illustrate the usage of equation (12). Here, ℓ0 = 200 and N = 107 to generate the moisture map in Fig 15 having a mean and variance of 0.1678 and 0.0071 respectively. The moisture values were normalised such that the minimum was 5% and the maximum was 40%; and the mean moisture value was 16.8%, and the mean-square variance is 0.0071 among moisture values evenly distributed (see lower part of Fig 15; station Rxireceives the signal that is transmitted by Txi, 1 ≤ i ≤ 40).

[0095] In creating simulated moisture maps, the performance of the GPR method is sensitive to the distribution pattern of soil moisture. In the simulation, moisture maps with different poles and variances were randomly generated to obtain the estimation error, as shown in Fig 7. ℓ∗ = 0.5 and σ∗n = 0.01 were used for the GPR method. More poles were used to simulate soil moisture with a larger variance (the −−×−− line in Fig 16), leading to a smaller mean estimation error (the −◦−・ line). The curve in Fig.17 show a higher moisture map variance (moisture variance is 0.24) than that in Fig.18 (moisture variance is 0.012). The GPR parameters are optimized as desired. The GPR estimation approximates the soil moisture more closely in Fig.17 than in Fig.18.

[0096] In respect of sample selection, simulation examples were used to demonstrate the performance of sample selection. The moisture map in Fig.15 is used; ℓ∗= 1 and σ∗n = 0.002 are assumed for the GPR method. The GPR method enabled determine moisture estimation given arbitrary locations in the field. In this embodiment, the accuracy of the GPR method is largely dependent on the amount and quality of train data the soil moisture samples were used as training data for the soil moisture map and the RF sensors to divide the field into pixel blocks. It is expected the more RF sensors and soil moisture samples there are, the more accurate the produced soil moisture map will be. As more soil moisture samples are obtained, the estimation errors reduce gradually.

[0097] Optimized sample selection was used in the simulation of Fig 19 and it can be seen that the optimized outperforms the random sample selection when there is a finite number of samples (20 ∼ 50). For example, with 30 soil moisture samples (see Fig 20 and the first map of Fig.14), the average estimation error and the variance of the optimized sample selection are twice as small as those by random sample selection. Contrary to intuitions, the ideal sample selection (see Fig 14) does not yield better performance in meaningful terms. It is a deterministic approach that might work with a short range of moisture maps. However, it does not have the generalization ability to approximate a widerange of moisture maps. When there are 400 samples, the GPR method (see Fig 14, second map) has a 6.135% moisture estimation error. The GPR-based solution will have a non-zero estimation error even with more measurements.

[0098] The foregoing describes only one embodiment of the present invention and modifications, obvious to those skilled in the art, can be made thereto without departing from the scope of the present invention.

[0099] The term “comprising” (and its grammatical variations) as used herein is used in the inclusive sense of “including” or “having” and not in the exclusive sense of “consisting only of”.

Claims

CLAIMS:

1. A method for determining moisture content of soil within a measurement zone, including the steps of: transmitting and / or receiving radio frequency signals between a plurality of spaced apart nodes; receiving, by a controller, signal data about the intensity of radio frequency signals transmitted and / or received from each of the plurality of spaced apart radio nodes; transmitting, by the controller, the signal data about the intensity / characteristic of radio frequency signals to a server; determining, the relative signal strength of the radio frequency signals transmitted between each of the plurality of spaced apart nodes; and determining, the moisture content of soil for each location between each of the plurality of spaced apart nodes within the measurement zone.

2. The method of claim 1 further including the step of redirecting a received radio signal from at least one of the plurality of spaced apart nodes to at least one of the plurality of spaced apart nodes, thereby to define a local measurement zone.

3. A system for determining the moisture content of soil within a measurement zone, including: a plurality of spaced apart nodes disposed in at least one of: within, at a periphery and adjacent to the measurement zone; each of the plurality of nodes being configured to transmit and / or receive radio frequency signals, a controller in communication with each of the plurality of spaced apart radio nodes; and a server in communication with the controller; wherein the controller is configured to receive signal data about the intensity of radio frequency signals transmitted and / or received from each of the plurality of spaced apart radio nodes, and transmit the signal data to the server; and wherein the server is configured to determine the relative signal strength of the radio frequency signals transmitted between each of the plurality of spaced apart nodes, andcalculate the moisture content of soil for each location between each of the plurality of spaced apart nodes within the measurement zone.

4. A system according to claim 3, wherein the system further includes at least one reflector disposed within the measurement zone, to redirect a received radio signal from at least one of the plurality of spaced apart nodes to at least one of the plurality of spaced apart nodes, thereby to define a local measurement zone.

5. A system according to claim 3 or 4, wherein the plurality of spaced apart nodes form at least a partially connected mesh network.

6. A system according to any one of claims 3 to 5, wherein each of the plurality of nodes includes at least a transceiver configured to transmit and / or receive the radio frequency signals and / or transmit the signal data to the controller.

7. A system according to any one of claims 3 to 6, wherein each of the plurality of nodes includes at least a processor, an antenna and a power source.

8. A system according to any one of claims 3 to 7, further including a plurality of reflectors to define a plurality of local measurement zones.

9. A system according to any one of claims 3 to 8, wherein the server is configured to output a map representing moisture content of soil within a measurement zone, based upon the determination of the moisture content of soil for each location between each of the plurality of spaced apart nodes 10. A system according to any one of claims 3 to 9, wherein the server is configured to output a map representing moisture content of soil within the local measurement zone.

11. A system according to any one of claims 3 to 10, wherein the radio frequency signals have a frequency in the range of 100kHz to 100MHz.

12. A system according to any one of claims 3 to 10, wherein the radio frequency signals have a frequency in the range of 100kHz to 10MHz.

13. A system according to any one of claims 3 to 10, wherein the radio frequency signals have a frequency in the range of 150MHz to 1GHz.

14. A system according to any one of claims 3 to 10, wherein the radio frequency signal is 433MHz.

15. A system according to any one of claims 3 to 14, wherein each of the plurality of spaced apart nodes are located in the range of 0 to 10 m above the ground.

16. A system according to any one of claims 3 to 15, wherein a distance between any of the plurality of spaced apart nodes is in the range of 0.01km and 10km.

17. A system according to any one of claims 3 to 16, wherein a distance between any of the plurality of spaced apart nodes and the controller is in the range of 0.01km to 10km.

18. A system according to any one of claims 3 to 17, wherein any of the plurality of spaced apart nodes is configured to operate as a controller.

19. A system according to any one of claims 3 to 18, wherein the server uses the determined moisture content of the soil between the plurality of spaced apart nodes to output a control signal to an irrigation system.

Images