NZ623007B2 - Method and device for determining greenhouse gas emission from a ruminant - Google Patents

Abstract

method for estimating a greenhouse gas emission from a ruminant (6) in a loose housing environment (1), and in a predetermined time period from T0 to Tdesired, characterized in that the method comprises: - determining a model emission rate function EM(t) for the ruminant, - determining the moments of eating feed by the ruminant, at least during the predetermined time period, as a series of points in time {T1, T2, T3,..., Tn}, - constructing the estimated real emission rate function ER(t) on the basis of the model emission rate function and the moments of eating feed, and - integrating ER(t) from T0 to Tdesired. of eating feed by the ruminant, at least during the predetermined time period, as a series of points in time {T1, T2, T3,..., Tn}, - constructing the estimated real emission rate function ER(t) on the basis of the model emission rate function and the moments of eating feed, and - integrating ER(t) from T0 to Tdesired.

Description

Method and device for determining greenhouse gas emission from a ruminant The invention relates to a method for estimating a greenhouse gas emission from a ruminant in a loose housing environment, and in a predetermined time period from T0 to Tdesired, and to a greenhouse gas emission estimation device for estimating a greenhouse gas on from a ruminant, according to said method.

Enhanced greenhouse gas (GHG) emissions are deemed to be a cause of an enhanced greenhouse effect. Emissions by dairy s are believed to contribute a large part of the enhanced greenhouse gas emissions, not only via manure but also, and ed mainly, via tion of methane. Therefore, it is important to know the emission of such ouse gases in dairy animals. For knowing the emissions is a necessary first step of controlling and reducing such emissions.

In the art, a lot of research has been done to try and determine the GHG on of individual dairy animals in a controlled environment, such as completely sealed cabins, in which the complete emission of the animal could be measured and analyzed. However, in a normal dairy environment, useful models are scarce and mainly ct such as through fatty acid profiles from milk.

It is an object of the present invention to provide a useful model for the determination of greenhouse gas emissions of dairy animals in their normal environment, in particular in a loose housing environment.

Alternatively, it is an object of the invention to provide the public with at least a useful choice.

The invention provides a solution for either of these objects with the method of claim 1, that is characterized in that the method comprises: determining a model emission rate function EM(t) for the ruminant, ining the moments of eating feed by the ruminant, at least during the predetermined time period, as a series of points in time {T1, T2, T3, ..., Tn}, constructing the estimated real emission rate function ER(t) on the basis of the model emission rate function and the moments of eating feed, and integrating ER(t) from T0 to ed. In this way, the various eating moments are taken into account as regards their effect on the on. The invention's insight is to apply a fresh "start", each time that the animal eats, and to determine the total as the sum over a consecutive series of [Followed by page 1a] such fresh starts. The inventors' work indicates that a useful approximation of total on can be obtained with such a model. The inventors have found that each moment [Followed by page 2] [Followed by page 2] of eating starts a new "cycle" of GHG emissions, which has an effect on total emissions in a certain time period. For example, in case there are more eating moments in that time period, there is very likely also an increased output of GHG, and vice versa.

It is noted, that the model emission rate function should be taken to have a peak value some time after the eating moment. This reflects the fact that, when new feed is eaten, the digestion that is to produce GHG will need some time to do so.

This will work to a peak, after which the gradually decreasing amount of feed still to be ed will cause a decrease of the emitted GHG.

Advantageous or otherwise l ments are mentioned in the dependent claims.

In embodiments, the function ER(t) is constructed as ER(t) = EM(t - T1 - C1) for T1St<T2, EM(t - T2 - C2) for T25t<T3, EM(t - Tn-1) for Tn-1St<Tn, and EM(t - Tn - On) for TnSt, n Cl, CZ Cn are constants.

In this way, after each eating moment, a similar shape of the emission curve is constructed, as the emission mechanism, stemming from the particular animal, does not change itself. One could also say that n each set of subsequent eating moments, a new emission curve is set up, but in each case based on the same basic curve shape. The nts serve to adapt the curve parts to specific circumstances or approximations, as will be ned further below.

In a particular embodiment, all constants are zero. In this approximation, each previous curve is simply cut off at a new eating moment, and the curve starts anew from zero emission at that new eating moment. This is a very simple imation, and easily implemented mathematically. Still, however, especially since in practice most curves will show a peak emission very shortly after the eating moment, this is already a useful approximation.

In other embodiments, all constants 01 Cn are determined such that the function ER(t) is a continuous function. This holds in particular for the eating moments, i.e. the transition points. In practice, the emission function will be a continuous function, just like any other physical quantity. Therefore, an imation that will take this into account can be a better approximation. In this embodiment, a us curve part will have a certain value at a certain time after its corresponding starting eating moment. In this ment, it is assumed that the curve part for the subsequent interval, i.e. after the following eating moment, will start at that same value. Note that this still leaves two ilities, one before the peak value, and one after the peak value. It is assumed that the value before the peak is taken, g to a corresponding value for the corresponding constants C1, CZ, The above shows just a number of possible embodiments of the general idea of the invention. Others could be to simply add a new curve to the existing one. In other words, all emission curves started before a new eating moment are allowed to go on indefinitely, while the value for a new curve, starting at the new eating moment, is added for each time after that new eating .

Therefore, in an additional embodiment, the function ER(t) is constructed as ER(t):ZEM(t—Ti) for all eating moments T1, Tn relevant for the desired 1'21 period of time T0 - Tdesired. Note that this encompasses the case that even T1 is after TO, leaving the first time part from T0 to T1 open to some uncertainty. It also encompasses the case of one or more eating moments T1, T2, before TO. The relevant time frame to be taken into account before TO depends on how quickly the function EM(t) falls to zero. If that is a period Tfall, then ably all eating moments up to Tfall before TO should also be taken into account, as the first eating s T1, T2, However, if that time is only a small part of the total period of time T0 - Tdesired, it may safely be ignored.

Note that the step of integrating is deemed to encompass all mathematical equivalents thereof, among others the approximation of summing a number of points times the width of the relevant intervals et cetera. Also note that it is necessary for the most precise tion, to determine all moments of eating.

However, it is possible to miss one or more moments, allowing for a less precise approximation. Also, an estimate between T0 and T1 for TO<T1 is not possible, and is left out. In such a case, some (more general) approximation for the first interval should be made, or the time and eating moments should be selected such that this situation does not occur. Note that it is also possible in the present model to have more than one measurement of GHG emissions between two eating s. Such apparently additional measurements then serve to make the model more precise.

In particular embodiments, the feed is ration, ning roughage such as grass, hay, silage, maize, and optionally containing concentrates. Herein, the concept of "ration" is deemed to se the so-called TMR, or total mixed ration, as well as PMR, or partial mixed ration. All these rations se roughage, while some also comprise concentrates. It is believed by the inventors that these non-"concentrates only" feeds provide the main GHG emissions.

Concentrates only feeds, such as given in for e milking parlours or ted concentrate feeding parlours, are to be excluded. It is believed that "concentrates only" feeds will have an effect on the total GHG emissions, but what that effect is, is not yet known.

In ning the invention above, a part comprises determining a model emission rate function for the animal. In embodiments, determining the model emission rate function comprises measuring a greenhouse gas on rate at a plurality of measurement points in time TMO, ,TMn, providing tive values EMO, , EMn, establishing for each of said measurement points in time TMO, TMn the most recent feeding moment TFO, TFn before the respective measurement point in time, and determining the function ER(t) by fitting a mathematical curve that fits h the sets of values ((TMO-TFO), EMO), ln , ((TMn-TFn), EMn). itself, the mathematical technique of fitting a curve to s points is well-known.

However, according to the present invention, it is understood that all measured points can be related to a single curve, the (model) emission rate function. As said above, this is based on the insight that, at least on average, the GHG mission for a dairy animal will always show the same our after eating. Therefore, it does not matter when the GHG emission (rate) is measured, as long as it is measured.

As a remark to be made here, the inventors believe that the actual amount of feed consumed may have an influence on the production of GHG. However, to the inventors' knowledge, there is not known a clear description of any mathematical connection between amount of feed and GHG emissions.

In an important further development of the invention, the method further comprises repeating the steps of - measuring at an additional point in time Tadd said greenhouse gas emission rate EMadd, and - redetermine the function by adapting and refitting the mathematical curve h one of the last X sets of values of TM and EM, wherein X is a predetermined number, and all sets of values of TM and EM. The former case represents something of a rolling e emission rate function, while the latter represents a more general average function. With this embodiment, it is possible to update the emission rate function to new . This not only allows an improved accuracy, but also allows dynamic adaptation of the function/the curve to changes in the , such as might follow from development of the body of the animal, e.g. maturing or its lactation cycle, or a change in diet, seasonal changes and so on.

In a particular embodiment, in the step of redetermining the on, one or more ermined constraints are applied, comprising predetermined rules with respect to the relative and/or absolute change of one or more coefficients used in the function. In this embodiment, care is taken that the function, i.e. the model emission rate function EM(t), does not change too wildly. A particular approach to applying such constraints is the so-called DLM method, or dynamic linear modeling. For an e, reference is made to EP2154952, relating to a DLM model for feeding dairy animals. More in particular, this example is based on Bayesian rules for adapting the model. Such Bayesian rules may also advantageously be applied as the constraints to be applied in the model of the present invention.

In the present invention, use is made of a standard curve for emission EM(t) for each . This is a first approximation. In a particular further development, the method is refined in that the mathematical curve is selected from a set of standard curves depending on the type or race of animal, and wherein the fitting comprises multiplying the curve by a constant. This allows the following advantage. Although it is assumed that the basic shape of the emission function EM(t) is basically the same for all animals at least of a n type or race, there may be ions within such a type or race of animals. For example within the race Holstein-Friesian, there may be a small or extra large animal. This might simply be the result of age, or of genetic variation in the final ions. heless, this embodiment assumes that the basic curve may then be scaled up or down by a nt. This constant is to be determined from a number of measurements. For example, the first 3 - 10 measurements are used to determine the constant (factor) for scaling the standard curve for Holstein-Friesian cows, selected from the set of standard curves. This can be done by fitting those ements to the rd curve, with the scaling factor is the variable for fitting. The resulting fitting factor, applied to the standard curve, then gives the "adapted standard curve" for that specific animal, which will lead to more accuracy. Of course, other ways of fitting to the standard curve are not excluded, such as a shorter or longer time delay between eating moment and peak of the emission curve.

One of the steps in the method is determining the moments of eating of the feed. In embodiments, determining the s of eating feed ses determining the moments when at least one of the ing conditions is met: new ration, excluding concentrates only feed, is provided to the nt; feed is displaced towards a feed fence at which the ruminant is allowed to feed; the ruminant is provided access to a pasture. Herein, the choice is made that the eating moments are the starting s. It may also be argued that the moments of eating should be the average of the period of time during which the condition is met, or a predetermined time after the first moment the condition is met. Although the ors made a choice for the first moment, this is done more out of practicality than for other specific reasons. For GHG emissions do not start to increase at the very moment of the ng to eat, as they at least require digestion down the digestive tract. It should be noted, therefore, that the abovementioned alternative moments are also deemed to be within the scope of the present invention.

In particular embodiments, determining the moments of eating comprises identifying the ruminant at a feeding place and determining that said ruminant is eating, by means of a eating sensor. In some barn layouts, there are specific individual feeding stations for feeding ration. As these provide dually adapted s to the animals, they are ideally suited to provide specific information on eating moments. By the way, such individual feeding stations are also ideal to determine GHG emission, since there is no interference from other animals. Note that the emission is deemed to be caused by us eating. All this will be ned further below.

In some embodiments, the eating sensor comprises a camera with image processing software for recognizing eating behaviour, a feed weight sensor arranged to determine a weight change in a feeding device, and/or a microphone with sound processing software for recognizing eating behaviour. These have proved useful ways of determining when an animal eats. For example, if the image processing software ishes that the animal repeatedly puts its mouth where the feed is, it is safe to assume that it is eating. If the, commonly known, feed weight sensor establishes a change in the feed weight, the animal must have eaten. As to the microphone with sound processing software, reference is made to EP1301068, which discloses a sensor able to distinguish between ruminating and eating. Thereby, eating actions may be established.

In embodiments, determining the moments of eating comprises determining when the ruminant produces eating sounds by means of a sound sensor, in particular provided on or near the neck and/or mouth, that is ed to detect sounds produced by the ruminant when moving its mouth in particular for chewing. This is a at more particular ment as mentioned before.

Again, reference is made to EP1301068 for more particulars. Also, as already mentioned above, when determining the moments of eating, rumination sounds are to be excluded, as these are not believed to have a major influence on GHG ons.

In particular, it is possible to include detecting whether the ruminant is standing up or lying down, and neglecting moments/periods with sounds produced when lying down, more in particular detecting whether the head of the ruminant is d to the ground, and neglecting moments/periods with sounds ed while the head is not lowered during at least a predetermined time. All this serves to select the s of actual eating, and leaving out other activities.

Herein, it is assumed that it is relatively rare that the animal ruminates while standing up, and to further exclude this situation, which is not impossible, it is assumed that the combination of standing up, ruminating and the head s the ground is really to be ted. In other words, if the animal es eating sounds while standing up and the head is down, this is deemed definitely eating.

In other embodiments, the method comprises that, when determining the moments of eating, only rumination sounds are detected, and the moments of eating are determined to be the moments of a local minimum in rumination . Again, the rumination sensor as disclosed in e.g. EP1301068 may be used to determine periods when the animal ruminates. It was found a good approximation of the eating moments to take the time of minimum rumination activity, inbetween periods of rumination. This embodiment is not applicable in loose housing systems, but in particular also in systems where, at least during part of the day, but also during longer periods before returning to the housing, the cows graze on pastures. On pastures, it is more difficult to determine exact eating moments. However, by using a rumination sensor, it is still le to determine the moments by the method just described. Note that such rumination data may be sent to a central computer, by Bluetooth(tm) or other transmitting devices.

In advantageous embodiments, the method comprises: - providing a herd having a plurality of ruminants - providing one or more ruminant ID devices, at least at one or more ons for feeding the ruminants and at one or more positions for measuring a greenhouse gas emission rate and arranged to identify the ruminant and provide an ID signal - performing the method for each of the ruminants in the herd, and - providing a total greenhouse gas on in the ermined time period as the sum of the greenhouse gas emissions from each of the ruminants.

With this embodiment, the GHG emissions may be determined for a whole herd, instead of just one animal. It is understood that for each animal it is possible to take the same standard curve EM(t) for constructing the individual curve ER(t), but it is of course also possible, and more te, to use individualized curves as described further above. By thus knowing the total GHG emissions of the herd, le effects of diet or othenNise may be studied. This knowledge may be used to see if it is possible to reduce the total GHG emissions.

In the present method, the animal can be any nt animal, such as oes, goats, and so on. In particular however, the ruminant is a cow.

Furthermore, the greenhouse gas may be any gas that is emitted by the animal that has an effect on the enhanced greenhouse effect, such as carbon dioxide and methane. In particular, however, the GHG in the present method is methane, as this is not only a gas with a much bigger relative effect on the enhanced greenhouse , with a global warming potential of 25 times that of carbon e, over a period of 100 years, but it is rmore more ly related to digestion, thereby providing more direct information thereabout. Contrarily, carbon dioxide is also, and maybe predominantly, related to burning bodily fuel. Thus any information from carbon dioxide relating to digestion should be filtered out, which is not an easy step.

The invention also relates to a greenhouse gas emission tion device for estimating a greenhouse gas emission from a ruminant, according to a method of any of the preceding claims, the device comprising: - at least one greenhouse gas emission rate sensor, arranged to provide a greenhouse gas emission rate signal - a clock device arranged to provide a time signal each time when said ruminant eats and each time when the at least one greenhouse gas emission rate sensor measures a greenhouse gas emission rate signal - a control unit arranged to apply the method of an y preceding claim, and on the basis of the measured time signals and greenhouse gas emission rate signals.

This device s the method of the present invention, with a principal advantage of supplying useful ation on (total) emission of GHG in a simple way.

In embodiments, each greenhouse gas emission rate sensor comprises or is provided together with a nt ID device arranged to establish the ID of a nt upon visiting the sensor. This is useful for coupling the measured emission to a particular animal, especially in an environment where there are more animals.

Advantageously, the device ses a feeding unit with an animal ID station arranged to identify the ruminant upon g, the control unit being arranged to s the corresponding time signal together with the animal ID.

Herein, the ID station identifies the animal and couples a clock signal, i.e. eating moment signal, to the ID. Note that a similar signal may be ed otherwise, such as with a feed fence with an ID station, or a feed fence with a feed pusher having a control with a clock. Herein, when the feed pusher pushes feed towards the feed fence, according to e.g. a programmed route, it appears that the animals start eating anew. Thus, an eating moment is deemed present upon the feed pusher pushing feed.

In embodiments, there is provided a greenhouse gas emission sensor with an animal ID station at one or more of a g place, at a feeding place, such as a feeding box or feeding fence, and/or at a resting place. For the present invention, there needs to be at least one GHG . Preferably, such sensor is positioned in a way in which an animal is more or less separated from the others. In this way, emissions from other animals will not, or to an acceptable , interfere with the emission to be ed. Advantageously, a milking box and/or feeding box comprise such a GHG emission sensor. Having multiple sensors spread throughout the housing may be advantageous, as this may provide a larger set of measurements during the relevant time period.

[Followed by page 9a] Unless the context clearly requires otherwise, throughout the description and claims the terms “comprise”, “comprising” and the like are to be construed in an inclusive sense, as opposed to an exclusive or tive sense.

That is, in the sense of “including, but not limited to”.

[Followed by page 10] [Followed by page 10] Further aspects, embodiments and ages of the invention will be apparent from the detailed description below of particular embodiments and the drawings thereof, in which: Figure 1 is a very diagrammatic drawing, not quite in perspective, of a loose-housing system ing to the invention Figure 2 is graph for determining the basic function EM(t) based on measurements Figure 3 is a graph for ining an individual function EM(t) for a specific animal, again based on measurements; and Figure 4 is a graph showing a ucted function ER(t), according to the ion.

Figure 1 is a very diagrammatic drawing of a system 1 with a housing 2 with a stable 3 and optional pasture 4 with gates 5 there between. Cows are denoted by 6.

Reference numeral 7 indicates a milking robot with a gas sensor 8 and an ID device 9, all connected to a computer 10, here shown to comprise a transceiver 11. Afeeding box 12 is shown, with gates 13, a feed trough 14, as well as a gas sensor 8 and ID device 9.

A feed fence is indicated at 15, with feed 16 being pushed by a feed pusher 17, also having a transceiver 18. Cubicles are indicated by 19, in one of them a cow with an ID tag 20.

At the pasture, a cow is grazing, wearing a rumination sensor 21 and a height sensor 22.

In the housing, an important part is the milking robot 7. Here, the animals are milked and optionally fed concentrates. When visiting the milking robot, the animals are identified with the ID device 9. Additionally, the milking robot comprises a GHG sensor 8, in particular a methane , arranged to measure (greenhouse) gas emission from an animal in the g robot 7. For examples of such gas sensors, which are known per se, nce is made to e.g. document /0192213, in particular e.g. paragraph 0018. Also, as is described in this document, it is possible to monitor the ratio of carbon dioxide and methane in order to obtain an indication for the absolute emission of methane. Note that it is seen as an advantage when no bolus is used to determine GHG emission, such as those using slow release of SF6 as a tracer gas. No bolus means a more natural behavior.

A rly important part may be played by the feeding box 12. This may similarly comprise an animal ID , as well as a gas sensor 8 as described above. In the feeding box 12, roughage may be provided to an animal when visiting and identified by the ID device 9. Not only does this serve to offer the best ration available for the animal, but it also ensures that the eating moment is recorded with high cy. This may be achieved through e.g. a weight sensor (not shown) for the , that registers eating moments when the feed weight diminishes. Simultaneously, GHG emissions are measured by the sensor 8, and combined with a timing signal.

It is noted that the computer 10, as well as each ID device 9, will have a clock device for recording moments in time.

The ation from both the milking robot 7 and the feeding box 12 are sent to a central computer 10, where it is processed according to the invention, to be explained in connection with Figures 2-4.

In addition, there are provided cubicles 19 for cows 6 to lie down and ruminate. Shown here is only one cow 6 in a cubicle 19. The cow carries an ID tag , with which she is identified by the ID device 9. A local gas sensor 8 again serves to determine a GHG emission from that cow. It is remarked here that it is only al to provide so many gas sensors 8, as in principle one would do.

The feed fence 15 shown in the Figure 1 serves for the cows 6 to get feed 16, in particular ration, such as hay or the like. There may be added some concentrates. Also shown is a feed pusher 17, such as the Lely Juno(tm), that is arranged to push the feed 16 towards the feed fence 15, Le. towards the cows 6. It is remarked in practice that the pushing of the feed 16 is a trigger for the cows to come to the fence 15 and eat. Therefore, if the fee pusher 17 pushes, this is considered an eating moment, for all practical purposes. The feed pusher, programmed to follow certain routes and during certain times of day, will send a clock or timing signal to the central computer 10 via their respective transceivers 18 and 11, such as a Bluetooth(tm) connection or the like. The central computer will handle that signal from the feed pusher as an eating moment.

There is also shown a pasture 4, with gates 5. If these gates are tic, their opening, combined with a time , may be considered an eating moment, as the cows being presented with fresh grass may for all practical purposes be considered the start of eating said grass. Alternatively or additionally, the cows 6 may be provided with a rumination sensor 21. If this sensor 21 is arranged to discern eating sounds and rumination sounds, as disclosed in EP1301068, then sending a signal whenever there is an eating moment suffices.

However, even in cases when only rumination sounds are detected, an eating moment may be inferred, in particular by the central er to which the signals may be sent, e.g. by Bluetooth(tm). For the computer may infer the periods of time when there is minimal rumination activity, in between periods with high rumination activity. In practice, the moments of minimal rumination activity can be taken to be the eating moments. These are also stored by the er 10. Moreover, also shown is a height sensor 22, arranged to determine whether the cow 6 is standing.

In combination with the rumination sensor 21, this may well distinguish between eating and ruminating, as a cow will hardly ruminate when standing up, in particular if the height sensor 22 and/or the rumination sensor 21 is onally able to tell whether the cow has her head towards the ground.

Thereby, the total system 1 can ine the moments when an animal eats, either at the roughage trough 14 or at the feed fence 15, and even in the pasture 4. It furthermore establishes the GHG emissions of an animal (ID controlled) at various moments in the milking robot 7, and/or the feeding box 12 and/or in the es 19.

With the help of the above data, the system determines a scatter plot of GHG emissions versus time after an eating moment, in Figure 2. Note that this figure s either to the data for one animal, or (less preferably) for one type of animal. Next, on the basis of a selected basic model function, a mathematical curve is fitted through the data (the dashed line). As a basic model function, one could take e.g. EM(t) = P(t)-exp(-t) wherein P(t) is a polynomic function of a desired order, such as a first or second order. A concrete example would thus be EM(t) = 4-t"2-exp(-t). In any case, it should be a function beginning at zero, having a m, and then slowly falling off to zero again, to reflect the natural course of normal GHG emissions.

In an optional step, relating to Figure 3, this basic function EM(t), which holds for example for a certain race of cows, such as Jersey or Holstein, may be zed for an individual animal. Suppose that the general function EM(t) has been established for an average animal of that type. This is shown in the Figure 3 as the solid line, peaking at a value X (relative units). Now, for a particular animal, a number of emission values have been established, shown as the five circles. Then, a curve is fitted, that has the same shape, but is multiplied by a fixed nt F. This individualized curve is shown as the dashed line, which thus peaks at a value F-X. It is of course possible to use different fitting techniques to establish the individual curve, but this suits fine. Note that F is shown to have a value smaller than 1. It may of course also be larger than 1 for other s.

Then, in Figure 4, this function EM(t), optionally individualized, is plotted for various time intervals, each starting and ending with subsequent eating moments, such that a continuous mathematical function ER(t) is obtained.

Thereto, from a first eating moment T1, one arbitrarily starts from zero or a zero measurement, and the function EM(t) is plotted from t = T1. This is cut off at t = T2. At such time T2, the function for the first al will have a certain value V1.

Next, another function (again) EM(t) is plotted, now from the time T = T2 and beginning at the value V1. Thereto, the function EM(t - T2) has been shifted over a constant C2 to reflect the starting at a ro value. This comes down to the constant mentioned in claim 4. This next function is cut off at T3, where the whole procedure is repeated, and so on, until the point in time "Tdesired" has been reached. The resulting function ER(t) can now be integrated over the desired period TO through Tdesired. Obviously, this will often be a 24 hour period, but it can be any desired period, as long as it is likely that sufficient data points can be gathered in the time in between.

Note that for the construction method for ER(t) it follows that there is always the same maximum value X. This ly unnatural behavior is not present in adapted construction methods, still according to the invention. For e, it is possible to not cut off the function EM(t - T1) after T2, but let it continue, and simply add another function (EM(t - T2) on top of it from T2, and so on. Then the peaks will be higher, up to a maximum, when the first on EM(t - T1) has fallen to zero sufficiently. Furthermore, the invention also asses a c (re)determination of EM(t) when new data are added. This may also ensure that, according to the method bed for Figure 3, the peak height may be adapted to these new data, providing for more accuracy.

Claims (23)

Claims

1. A method for estimating a greenhouse gas emission from a ruminant in a loose housing environment, and in a predetermined time period from T0 to 5 Tdesired, wherein the method comprises: - determining a model emission rate function EM(t) for the ruminant - determining the moments of eating feed by the rum inant, at least during the predetermined time , as a series of points in time {T1, T2, T3, ..., Tn} - constructing the ted real emission rate fun ction ER(t) on the basis of 10 the model emission rate on and the moments of eating feed, and - integrating ER(t) from T0 to ed.

2. The method of claim 1, n the function ER(t) is constructed as ER(t) = EM(t - T1 - C1) for T1≤t<T2, EM(t - T2 - C2) for T2≤t<T3, ..., EM(t - Tn-1 - Cn-1) for Tn-1≤t<Tn, and EM(t - Tn - Cn) for Tn ≤t, wherein C1, C2, ..., Cn are 15 constants.

3. The method of claim 2, wherein all constants are zero.

4. The method of claim 2, wherein all constants C1 ... Cn are determined such that the on ER(t) is a continuous function.

5. The method of claim 1, n the function ER(t) is constructed as 20 ER t )( = ∑ EM (t - Ti ) for all eating moments T1, ..., Tn relevant for the desired period of time T0 - Tdesired.

6 The method of any one of the preceding claims, wherein the feed is ration, containing roughage such as grass, hay, silage, maize, and optionally containing concentrates. 25

7. The method of any one of the preceding claims, wherein determining the model emission rate function comprises: - measuring a greenhouse gas emission rate at a plu rality of measurement points in time TM0, ... ,TMn, providing respective values EM0, ... , EMn - establishing for each of said measurement points in time TM0, ... , TMn the 30 most recent feeding moment TF0, ... , TFn before th e respective measurement point in time, and - determining the function ER(t) by fitting a mathe matical curve that fits through the sets of values TF0), EM0), ... , ((TMn-TFn), EMn).

8. The method of claim 7, further comprising repeating the steps of - measuring at an additional point in time Tadd sai d greenhouse gas emission rate EMadd, and - rmine the function by adapting and refittin g the mathematical curve 5 through one of the last X sets of values of TM and EM, wherein X is a predetermined number, and all sets of values of TM and EM.

9. The method of claim 8, wherein, in the step of redetermining the function, one or more predetermined constraints are applied, comprising predetermined rules with respect to the relative and/or te change of one or 10 more coefficients used in the function.

10. The method according to any one of claims 7 to 9, wherein the mathematical curve is selected from a set of standard curves depending on the type or race of animal, and wherein the fitting comprises multiplying the curve by a constant (F). 15

11. The method of any one of the ing claims, wherein determining the moments of eating feed comprises ining the s when at least one of the following conditions is met: - new ration is provided to the ruminant - feed is displaced s a feed fence at which t he nt is allowed to 20 feed - the ruminant is provided access to a pasture.

12. The method of any one of the preceding claims, wherein determining the moments of eating comprises identifying the ruminant at a feeding place and determining that said ruminant is eating, by means of a eating . 25

13. The method of claim 12, wherein the eating sensor comprises a camera with image processing software for recognizing eating behaviour, a feed weight sensor arranged to determine a weight change in a feeding device, and/or a microphone with sound processing software for izing eating behaviour.

14. The method of any one of the preceding claims, wherein determining 30 the moments of eating comprises determining when the ruminant produces eating sounds by means of a sound sensor, in particular provided on or near the neck and/or mouth, that is designed to detect sounds produced by the nt when moving its mouth in particular for chewing.

15. The method of claim 14, wherein, when determining the moments of eating, rumination sounds are ed.

16. The method of claim 14, wherein, when determining the moments of eating, only rumination sounds are detected, and the moments of eating are 5 determined to be the moments of a local minimum in rumination sounds.

17. The method of any one of the preceding , comprising: - providing a herd having a ity of ruminants - providing one or more ruminant ID devices, at lea st at one or more positions for feeding the ruminants and at one or more positions for measuring a 10 greenhouse gas on rate and arranged to fy the ruminant and provide an ID signal - performing the method for each of the ruminants i n the herd, and - providing a total ouse gas emission in the predetermined time period as the sum of the greenhouse gas ons from each of the ruminants. 15

18. A greenhouse gas emission estimation device for estimating a greenhouse gas emission from a ruminant, according to a method of any one of the preceding claims, the device comprising: - at least one greenhouse gas emission rate sensor, arranged to provide a greenhouse gas emission rate signal 20 - a clock device arranged to provide a time signal each time when said ruminant eats and each time when the at least one greenhouse gas emission rate sensor measures a greenhouse gas emission rate signal - a control unit ed to apply the method of an y one of the preceding claims, and on the basis of the measured time signals and greenhouse gas 25 emission rate signals.

19. The device of claim 18, wherein each greenhouse gas on rate sensor comprises or is provided together with a ruminant ID device arranged to establish the ID of a ruminant upon visiting the sensor.

20. The device of claim 18 or 19, comprising a feeding unit with an 30 animal ID station arranged to identify the ruminant upon feeding, the control unit arranged to process the corresponding time signal together with the animal ID.

21. The device of claim 20, n there is provided a greenhouse gas emission sensor with an animal ID station at one or more of a milking place, at a feeding place, such as a feeding box or feeding fence, and/or at a resting place.

22. A method for estimating a greenhouse gas on from a ruminant substantially as herein described with reference to any one of the embodiments as illustrated in the accompanying drawings.

23. A greenhouse gas emission estimation device substantially as herein 5 described with nce to any one of the embodiments as illustrated in the accompanying drawings. WO 62404