The measurement principle, first described by EHLERT & SCHMIDT (1996), is the basis for the advanced measurement principle described in this work.
The following mechanical formulas (BEITZ & KÜTTNER 1990) are used for discussing the measurements based on the simplified theory seeing the cereal stem as a cantilever:
for the mass moment of inertia of the stem at the fixed point B, the root, according to STEINER is equation 1
J
_{B}
= J
_{G}
+
m
e
^{2}
[1]
where J _{B} is the mass moment at the fixed side at point B, then J _{G} is the mass moment in the gravity point G of the stem, m is the mass of the stem, and e is the height of the gravity point G of the stem to the soil. From equation 1 is calculated equation 2 for the mass moment of the stem at the height of the contact with the pendulum
[2]
where F is the force due to the mass moment of inertia, v _{D} is the driving velocity, r is the height of the contact point of the cylindrical body with the stem, and x is the bending distance of the bended stem, and J _{B} is the mass moment at the fixed side at point B. In long bending distances there will be an additionally force due to the acceleration of the mass of the upper parts of the stem, expressed by equation 3
F = m
a
[3]
where F is the force necessary to accelerate the stems, m is the mass of the accelerated parts of the stem, and a is the acceleration of the stems during measurement.
For the bending moment of resistance of the stem is used equation 4 to calculate the force F necessary to bend the stem
[4]
where E is the elasticity of the material, r is the height of the contact point of the cylindrical body with the stem, x is the bending distance of the bended stem, and I is the second moment of the
The second moment of the area I is a measure of the distribution of mass around the longitudinal axis of a structure, hence dependent upon transverse geometry, and is calculated for hollow cylinders using equation 5
[5]
where do is the outer diameter and di is the inner diameter of the stem.
The bending distance x is the distance at which the stem is in contact with the pendulum and it can be derived from equation 4
[6]
where F _{P} is the force of the pendulum, r is the height of the contact point of the cylindrical body, E is the elasticity of the material, and I is the second moment of the area.
Friction force is the product of the mass of a sliding body and the friction coefficient of its sliding surface. The friction force can be derived by equation 7
[7]
where F F is the friction force, where F V is the vertical force component of the pendulum, and µ G is the friction coefficient between the two surfaces of the stem and the pendulum.
A light weight four wheel construction, consisting of two bicycles, welded together with a square frame, served as a carrier. Figure 1 provides a good illustration of the research carrier. The height of the frame being adjustable to keep it above the crop. The carrier has a handle to push the carrier manually along the rails. The bicycle wheels ran on rails to ensure exact sideways position in the replicated experiments.
The carrier was equipped with the measurement sensors, an electronic cabinet, two 12 Volt batteries as energy supply, and a 486 laptop to record and save the data. All original data was saved, and the vast measurement data was not reduced by the measurement system, but later by an Excel Visual Basic macro
The measurement software was the program Nextview, using the measurement system µ‑Meter 4, produced by BMC-Systeme GmbH. The measurement system had four inputs for analogue signals, and was calibrated for each sensor individually. It recorded the measurements of all four sensors as numerical data, together with the measurement time, and saved them as ASCII files.
An Excel Visual Basic macro reduced the original measurements into averages per plot, marked in the file by the trigger, thus reducing the data material to a single value per plot.
Based on this plot average all further statistics were calculated:
The sample size of the replicates necessary for the parameter optimisation trials, was calculated according to KÖHLER et al. (1992) for comparing two averages of two different samples with the same standard deviation, based on equation 8
[8]
where n is the unknown sample size, s is the standard deviation in preliminary tests, and
As reduction method to reduce the vast number of measurements to values per plot, were the average angle of deviation, the median of the angle of deviation, and the average of the vector
[9]
where x
_{i} is the length of the horizontal vector component, y
_{i} is the length of the vertical vector component, and n is the number of measurements to be reduced.
Observed and predicted residuals were plotted in SPSS 7.0 to check for possible biases in the measurement values.
Standard deviation and coefficient of variation of the measurement replicates of the plot average angle of deviation were calculated in Excel 7.0 using a Visual Basic Macro.
The relationship between biomass and angle was determined using simple linear regression (Pearson regression), to calculate linear goodness of fit (equation 10) and square goodness of fit (equation 11), and calculate the standard error of estimate of the regression formula (square root of the residual mean square), and its significance in STATISTICA 5.0.
y = a + bx
[10]
y = a + bx + cx
2
[11]
The relationship between additional plant parameters and angle was determined using simple linear regression (Pearson regression), calculated in STATISTICA 5.0.
The relationship between pendulum settings and angle was determined using simple linear regression, and multiple stepwise forward regression calculated in STATISTICA 5.0.
The interpolation of the crop biomass map in the Baasdorf winter wheat field was done in ARCVIEW 3.1, with the interpolation method inverse squared distance weighted IDW,
The growth-stages were determined according to the BBCH code (WITZENBERGER et al. 1989, LANCASHIRE et al. 1991). The BBCH code was developed from ZADOKS code
Fresh mass was determined by cutting all plants in a specific plot at soil surface, and weighing immediately thereafter on a balance with an accuracy of at least 0.1 gram. The cutting of the plants was done mostly in the afternoon, when the plants were hand dry. Winter rye 1998 was cut with a motor mower, while all the other crops were cut with sickles. With regard to rice, the field was drained before cutting. The fresh mass of the plot was calculated in kg m^{-2}.
Dry mass was determined for winter rye in 1998 according to method 1 used by KUHLA & WEIßBACH 1994, 48 hours oven drying at 60°C and then oven drying for 3 hours at 105°C. But since this is only possible for small samples, it was not useful to clarify the question, whether the biomass sensor was determining dry mass better than fresh mass, or vice versa. In this way, it was only possible to calculate the total dry mass per test area, based on the fresh mass and then multiplied with the determined dry matter content, thus biasing total dry mass.
With reference to irrigated rice, winter wheat 1999, and winter rye 1999, dry mass was determined according to the IRRI quality assurance (1996) for drying large biomass samples in cloth bags. Drying time for rice was 72 hours at 80°C, and 105°C during 72 hours for wheat and rye respectively. Repeatedly weighed test samples in cloth bags showed a constant weight after 65 hours. By using cloth bags, it was possible to determine the total dry mass of all plots. After weighing the fresh mass, the material was oven dried. The dry mass of the plot was calculated in kg m^{-2}.
Plant height was measured in centimetres from the top of the untouched plant – including awns – to the soil using a gauge-stick. The average plant height was considered to be the height of the main crop and was averaged out of 50 randomly selected stems of a plot.
Crop density in this context means, depending on the growth-stage, either the number of shoots (tiller density), or stems, ears, or panicles (stem density) per unit area. The account of crop density used in this work, was the number of stems counted on a full square meter per plot.
In paddy rice is the number of hills a measure of the yield structure and is related to crop biomass, and therefore, the total number of hills was counted per square meter.
Fungi infestation was determined for downy mildew (Erysiphe graminis) according to EPPO Standard PP 1 / 26 (3). Eye spot (Pseudocercosporella herpotrichoides) was determined according to EPPO Standard PP 1 / 28 (2).
The biomass sensor has been tested with winter wheat (Triticum aestivum ssp. Vulgare), winter rye (Secale cereale L.), and irrigated rice (Oriza sativa L.). Test sites for winter rye and winter wheat were located in Germany, at or near the Institute for Agricultural Engineering ATB near Berlin (figure 2), while the test sites for irrigated rice were located at the International Rice Research Institute IRRI, Los Banos, near Manila, Philippines.
Cultivation, management methods, and soil types of the German test sites for the winter rye and winter wheat optimisation trials are given in table 2. Soil types can have a wide range, depending on the field heterogeneity, and were not specified for the tests, since the sensor is not able to recognise the soils. The cultivation of the other test sites of winter wheat and winter rye are given in the appendix. The winter wheat optimisation trial was at the same time the test site of the trial for spatially reduced fungicide application in winter wheat.
WINTER RYE 1997 / 1998
Königsfeld
Variety: AMILO
Soil types:
sandy loam, loamy sand
Preceding crop:
Winter rye
Date
Name
Application rate
Seeding
30^{th} October 1997 AMILO 120 kg ha^{-1}
Fertilizing
25^{th}March 1998 21 / 21 / 21 N P K 90 kg ha^{-1} nitrogen
Growth-regulator
8^{th}May 1998 MODDUS 0.6 l ha^{-1}
WINTER WHEAT 1998 / 1999
Grube field
Variety: BATIS
Soil types:
loamy sand, clayey loam
Preceding crop:
Summer barley
Date
Name
Application rate
Seeding
12^{th} October 1998 BATIS 80 kg ha^{-1}
Fertilizing
31^{st}March 1999 Calcium-ammonium-nitrate 137 kg ha^{-1} nitrogen
Herbicide
2^{nd}May 1999 STARANE 0.7 l ha^{-1}
Fungicide*
20^{th}May 1999 JUWEL TOP 1.0 l ha^{-1}
Growth-regulator
20^{th}May 1999 MODDUS 0.4 l ha^{-1}
* site-specific application in some parts of the field as site-specific field trial
IRRIGATED RICE Variety: IR 64
Research Farm:
IRRI, Philippines, 1998 / 1999, Dry season
Preceding crop:
Irrigated rice, Double cropping system
Date
Name
Application rate
Transplanting
27^{th} December 1998 IR 64, 14 days old DAPOG seedlings 33 hills m^{-2},
Fertilizing
26^{th} December 1998 Urea 65 kg ha^{-1} nitrogen
Rat protection
15^{th}January 1999 Plastic fence
Weeding
28^{th}February 1999 By hand
2-3 plants per hill
At the German sites near Berlin there is predominantly a sub-continental temperate climate, characterised by long cold winters and warm summers, often accompanied by droughts in spring or early summer. The water deficit is the most important factor hindering plant production. Daily precipitation and temperatures for the German test sites are given in figures 3 and 4 for the growing season of the years 1998 and 1999 respectively.
Both seasons were warmer than usually and exceptionally dry, with 22 mm of rain between the 8^{th} April and the 24^{th} May 1998, and 21 mm of precipitation between the 25^{th} April and the 31^{st} of May 1999, causing in both years considerable drought symptomes in the crops, and in the 1999 season even crop extinction on sandy soils.
At the Philippine sites at IRRI there is predominantly an oceanic sub-humid tropical climate, expressed with a dry season and a wet monsoon season. In the dry season there are occasionally showers, but mostly dry and sunny conditions. The water deficit is no factor in plant production due to irrigation. The monsoon season is characterised by high precipitation, as heavy showers or long-lasting rainfalls. Daily precipitation and temperatures for the IRRI test sites are given in figures 5 for the growing season 1998 / 1999.
The biomass sensor optimisation trials were set up in fields, where there was the highest visible difference in biomass (figure 6). This difference in plant mass was visible in winter wheat and winter rye through large differences in crop height, while in paddy rice it was visible in crop density, that means how far one can see through the crop. The optimisation trials in winter rye – variety AMILO – were done in 1998 at the ATB research field near Berlin, Germany, in winter wheat in 1999 – variety BATIS – at a farmers field near Berlin, Germany, and in irrigated rice – variety IR 64 – in the dry season 1998 / 1999 at the IRRI research farm in the Philippines.
At selected places with differences in biomass, the rails were laid out in order to run the carrier. Every 5 meters, plot marker sticks were set in the soil to push over the trigger, and thereby mark
Measurement Replicates in the Optimisation Trials
The measurements of the 12 plots in these optimisation trials were the basis for determining the accuracy of the replicates. The accuracy of the 5 replicates was determined by comparing the plot measurements obtained with the same pendulum parameter setting, and calculating the standard deviation and the coefficient of variation.
Dependency of the Measurement on the Biomass in the Optimisation Trials
The measurements of the 12 plots and the plant mass of the 12 plots of these optimisation trials were the basis for determining the accuracy of the biomass measurements. The relationship between the biomass measurements in 5 replicates and the destructively sampled and weighed biomass was determined by correlating the values of the 12 plot averages with the weights of the biomass in the 12 plots. The accuracy of the regression was determined by the goodness of fit and the standard error of the regression.
Dependency of the Biomass Measurement on the Sensor Parameter Setting
The dependency of the measurements on the pendulum’s parameters was determined by simple and multiple regression for the change of the plot measurement when the pendulum parameters were changed. Therefore, the regressions were calculated for one plot of the optimisation trials of the 5 replicates of each parameter setting.
Trials for the Relation of the Measurements to Additional Plant Parameters
In addition to the optimisation trials for biomass sensing, on the same test strip, further measurements were undertaken. For determining the relationship between other plant parameters, such as plant height or crop density, and the sensor measurements, the angles of
Trials for the Influence of the Carrier Speed on the Measured Angle
As with the determination of other plant parameters, the basis of the measurements were the test strips of the optimisation trials after the optimisation trial was finished. The tests for determining the influence of the carrier speed on the biomass measurements were done with one parameter setting and a measurement frequency of 75 Hz. In the tests only the carrier speed varied in increments of 0.5 m s^{-1}. Tests started with 0.5 m s^{-1} as the lowest speed, passing on to 1.0 m s^{-1}, 1.5 m s^{-1}, 2.0 m s^{-1}, 2.5 m s^{-1}, 3.0 m s^{-1}, and finishing with 3.5 m s^{-1} as the highest possible speed, each with 5 replicates. The measured plot averages of the 5 m^{2} plot were correlated with the carrier speed, and were measured repeatedly for all crops and different growth-stages. Only in winter wheat, at BBCH 49, was the influence of the carrier speed on the measurements taken, in all 12 plots of the test strip, and calculated for the individual plots, with the exception of the first three, where a speed of 3.5 m s^{-1} was not possible to test due to the short acceleration way.
Trial for the Effect of Wind on the Measurement
To determine the influence of wind on the biomass measurements, trials were undertaken with a singular parameter setting, a measurement frequency of 75 Hz, a carrier speed of 1 m s^{-1}, for one 5 m^{2} plot, in rice at the growth-stage BBCH 69, or 80 DAT respectively. The influence of wind was not tested in other crops. In addition to the usual equipment on the carrier, two fans were attached to the carrier (figure 7). The wind of the fans was directed right onto the pendulum’s cylindrical body, the penetration angle of the wind into the crop canopy was 30° in this combination. The wind direction was opposite to the measurement direction. The fans were infinitely variable in speed, thus giving the possibility of measuring the crop with wind speeds of 0 m s^{-1}, 1 m s^{-1}, 2 m s^{-1}, 3 m s^{-1}, and 4 m s^{-1}. The lower parts of the fans were covered with cardboard to prevent a threshing of the panicles. The wind speed was measured with an anemometer directly in front of the pendulum’s cylindrical body, at the centre and at both ends of the cylindrical body. The measured plot averages of the 5 m^{2} plot was then correlated with the wind speed.
Trial for the Effect of Water Height on the Measurement in Irrigated Rice
The influence of the water height of the irrigation water in irrigated rice was tested in rice at BBCH 39, or 42 DAT respectively, for one 5 m^{2} plot, using a measurement frequency of 75 Hz, and a carrier speed of 1 m s^{-1}, and one single parameter setting. Water depth was determined using a measuring-stick, that gave the water height measured from the root base in centimetres, while the paddy field was flooded. The measurements were done during flooding. The measured plot averages of the 5 m^{2} plot was then correlated with the height of the irrigation water.
Trials for the Effect of Weeds on the Measurement
To determine the influence of weeds on the biomass measurements, trials were undertaken with a singular parameter setting in one 5 m^{2} plot, a measurement frequency of 75 Hz, and a carrier speed of 2.5 m s^{-1}. The influence of weeds was tested in winter wheat. The effect of Creeping Thistles (Cirsium arvense (L.) Scop.) was tested in winter wheat BBCH 39 in a plot with one single thistle 0.1 m higher than the crop. The effect of Loose Silky-Bent (Apera spica-venti (L.) Pal. Beauv.) was tested in winter wheat at BBCH 69 in a plot with a large number of them. The measurements were done in 5 replicates with the weeds in the crop and later without them.
Trials for Biasing Factors of Influence on the Measurements
The objective of the trials for the influence of biasing factors on the measurements of the biomass sensor pendulum-meter was to observe whether there is a factor that is able to bias the
In the trials for the influence of the day and daytime the test strip analogue to the optimisation test strips was measured repeatedly during two consecutive days with a singular parameter setting, 2.5 m s^{-1} carrier speed, and 75 Hz measurement frequency. The measurements were done in winter rye at BBCH 39. The averages of the plots were arranged in their respective time of measurement for the two days to see the effect. To indicate the effect of the growth-stage and rapid growth rate of the crop, an identical parameter setting was used in two growth-stages in winter rye. The carrier speed was 2.5 m s^{-1}, and the measurement frequency was 75 Hz for all measurements.
The trials for the effect of cultivar, variety, season, and year were made in all crops based on the results of the parameter optimisation trials. In winter wheat and rice, the effect of the variety was tested by measuring a different variety with an identical parameter setting as one of the settings used in the optimisation trials, a carrier speed of 2.5 m s^{-1}, and a measurement frequency of 75 Hz. Thus, it was possible in rice to compare two different varieties, one seeded and the other transplanted, and one crop tested in the dry season while the other was tested in the monsoon season, both tested at BBCH 65, or 80 DAT respectively. With regard to winter wheat it was possible in this way to compare different varieties, one ZENTOS and the other BATIS, in two different years at BBCH 69. In winter rye, the effect of the year was also compared. In the first year the measurements were obtained in the optimisation trials, and in the second year measurements were replicated with one specific parameter setting to check for the repeatability of the measurements in succeeding years. The measurements in winter rye in the second year were not taken over a 60 m long test strip, or 12 plots, but over a 50 m long strip, or 10 plots.
The trial for the effect of the ratio of biomass versus crop height was done in winter wheat at BBCH 69, to check for the influence of the crop height versus the crop density. Therefore, a test strip with differences in crop height was measured with a specific parameter setting, 75 Hz measurement frequency and 2.5 m s^{-1} carrier speed. Then one plot was measured repeatedly, while the crop of the plot was randomly thinned out in increments of 25 stems between the measurements. Each measurement was replicated 5 times and averaged for the plot. The thinning results were then compared with the measurement of the entire test strip.
The trial for the influence of the stem inclination was done in winter wheat at BBCH 69 with identical parameter settings, 75 Hz measurement frequency and 2.5 m s^{-1} carrier speed. The
Tests for the Limits of Non-Destructiveness of the Biomass Sensor
Before the parameter optimisation trials were started, the parameters were tested for their limits at which they induce symptoms of destruction on the plants. Therefore, the sensor was moved with possibly critical parameter settings in the crop. Those parameters showing symptoms of destruction were usually not used in the parameter optimisation trials, but some parameters showed symptoms of destruction not in the first replicates, but in the 5^{th} replicate.
To investigate the potential of site-specific reduced plant protection, according to the biomass data measured by the biomass sensor pendulum-meter, two field trials were set up. One in winter rye, with a site-specific reduced application of the growth-regulator MODDUS^{®}, and one in winter wheat, with a site-specific reduced application of the strobilurin-fungicide JUWEL^{®}.
Site-Specific Application of Growth-Regulator
The trial consisted of two neighbouring tram lines, of which one was the strip with the site-specific reduced application, and the other one the strip with the uniform application of the growth-regulator MODDUS^{®} (trinexapac-ethyl). In the uniform variant, the entire strip was sprayed with 0.6 l ha^{-1} MODDUS in 300 l ha^{-1} water. In the strip with the site-specific reduced amount of the growth-regulator MODDUS, the field sprayer was switched on and off. That means, in areas with low biomass, there was no application, and in areas with average or high biomass, there was a full application. In addition, there was a third tramline left without spraying of the growth-regulator as a control strip. Cultivation methods are given in the appendix. Plot length was 10 m, and both strips were 440 m long. Permanent plot markers in the crop indicated the plots till harvest. The carrier was pushed along the tramlines with a velocity of 1.0 m s^{-1}. The measurements were done with the pendulum settings of 0.5 m h_{P}, 0.1 m h_{A0}, 1 kg m _{P}, and a frequency of 75 Hz. After measuring the winter rye crop, variety AMILO, at BBCH 49, the average angles were calculated for one meter. Flags were set up in the crop, where the application rate changed according to the measurements, since an on-line control of the sprayer was technically not yet possible. Grain yield was determined by harvesting each strip separately in a harvester without a yield measuring device, and weighing the yield on a truck balance.
The trial consisted of two strips along neighbouring tramlines, one for the site-specific reduced application, and the other one for the uniform application. The fungicide JUWEL^{®} was applied, consisting of the two active agents kresoxim-methyl and epoxiconazol. For the other cultivation methods see winter wheat in table 1. In the uniform variant, the entire strip was sprayed with 1.0 l ha^{-1} JUWEL in 300 l ha^{-1} water. In the strip with the site-specific reduced application, the field sprayer alternated between full application, an application of 0.66 l ha^{-1} JUWEL, and the lowest rate using 0.33 l ha^{-1} JUWEL, according to the biomass data. Plot sample length was 10 m, and both strips are 320 m long. Permanent plot markers in the crop indicated the plot positions. The carrier was pushed along the tramlines in the field with a velocity of 1.0 m s^{-1}. The measurements were done with the pendulum settings of 0.6 m h_{P}, 0.3 m h_{A0}, 1 kg m _{P}, and a measurement frequency of 75 Hz. After measuring the winter wheat crop, variety BATIS, at BBCH 41, the average angles were calculated for one meter. The angle, at which the application rate changed, was determined according to a visual assessment of the specific risk of low and high biomass yields for fungi infestation. Flags were set up in the crop, where the application rates changed. Flags indicated the area with the same application rate, since an on-line control of the sprayer was not yet possible. The field sprayer varied application according to the flags. Grain yield was determined by harvesting each strip separately with a harvester without a yield measuring device, and weighing the yield on a truck balance. Fungi infestation of downy mildew was determined in the uniformily sprayed tramline.
In the year 2000, a prototype pendulum-meter was tested for a tractor-based, automatic on-line sensing of cereal crop biomass, which was attached to the three point linkage in the back of the tractor. The tractor ran in the tramlines while sensing the biomass. The measured crop was a winter wheat crop of the field Baasdorf at BBCH 39, variety CONTRA. Positioning was determined with a TRIMBLE^{®} 132 GPS (INSAT 2000), and recorded with an advanced version of the NEXTVIEW^{®} measurement program. The measurements were done with the pendulum settings of 0.6 m h_{P}, 0.1 m h_{A0}, 1 kg m _{P}, a carrier speed of 2 m s^{-1}, and the measurement frequency was technically limited to 1 Hz. Positioning data were recorded in World Geodetic System of 1984 – WGS 84 (EUROPEAN ORGANISATION FOR THE CONTROL OF AIR NAVIGATION 2000), and re-calculated in Gauß-Krüger as projector and Bessel as ellipsoid. A video of this measurement with the prototype pendulum-meter is attached.