2. The effect of variables within the crop on yield measurement

Transcription

1 2. The effect of variables within the crop on yield measurement 2.1 Overview Traditionally, yield sensors on combine harvesters have been used to collect information on the total yield or grain weight from individual fields and farms. The total crop weight or yield has been used by the farmer and agronomist to fine-tune crop husbandry techniques and to promote the efficient use of inputs, Birrell et al. (1994). Today, however, yield sensors are not only used to measure total weights, but are also being utilised to measure crop flow. The additional data on crop flow, and corresponding positions within the field, form the basis of any yield mapping system. Commercially available yield sensors are based on either one of two different principles for measuring the mass flow of grain: 1. Volumetric - a volume of grain is measured and then converted into mass by using the equation: Density = Mass x Volume. 2. Mass flow detection - a measurement to determine grain mass. Various techniques for mass flow measurement have been reviewed by Bull (1988). Bull states that there is clear distinction between volume and mass flow rate. The former is a measure of the rate of mass of material conveyed while the bulk density remains constant. Mass flow sensors can be sub-divided in two distinct types: (i) True mass flow meters which have a sensing element that reacts to the mass flow of solids through the instrument and (ii) Inferential mass flow meters which determine both the instantaneous solids concentration and flow velocity within the sensing element and multiply these two measurands to give mass flow rate. Where the flow velocity is constant, a solid concentration sensor can be calibrated in terms of mass flow rate. This is approximately so in a combine harvester for small cereal grain, Stafford et al. (1991). Mark Moore (1997) 6

2 2.2 Yield sensors and their accuracy - (Mass detection Versus Volumetric) The yield measurement principle and system used to monitor and record yield levels within a field provides the first source of potential error within any yield mapping system. Therefore, attention must be given to the way other crop variables may affect the accuracy of yield data derived from either of the above principles of yield measurement. There are five main yield sensors which are commercially available today, as summarised in Table 2.1, all of which utilise one of the above principles of yield measurement. Table Main commercially available yield sensors and the organisations marketing these systems Company Measurement Technology Manufacturers Principle Utilised Claimed Accuracy (+/-%) Massey Ferguson Mass Detection Absorption of 1+/- (Flow Control) Gamma Radiation RDS Technology Volumetric Opto- 2+/- (Ceres) Electronics Micro-Trak Mass Detection Force on Impact 1-2+/- Plate Ag-Leader Mass Detection Force on Impact 2+/- Plate Claas Yield-o-meter Volumetric Positive 1+/- (Claydon) Displacement Auernhammer et al. (1993) compared the performance of a mass detection system with a volumetric based system. They concluded that the overall performance of both yield sensors were very similar. However, it was noted that when using the volumetric system, the variation of specific weight within the crop on measurement accuracy was highly significant. Auernhammer stated that with volume measuring systems, accuracy is only guaranteed if regular measurements of the crop density are taken and manually programmed into the yield sensor s computer. Schroder et al. (1995), arrived at a similar conclusion. They reported that grain bulk density will change continuously with varying field conditions, and therefore volumetric systems require considerable effort in maintaining an accurate calibration. Mark Moore (1997) 7

3 Current methods for analysing the accuracy of yield sensors, by comparing true weigh-bridge weights with yield sensor estimates, do not take into account the effect of other variable factors, such as crop density or moisture, on the flow of grain. The current procedure for comparing the weights of large volumes of grain (combine grain tank or grain trailer loads) merely averages out any effect a variable, such as crop density, may have on the measured flow of grain. Where tests have revolved around large volumes of grain, the accuracy obtained from mass detection and volumetric yield sensors are very similar, Auernhammer et al. (1993). As the instantaneous yield, rather than the total yield, is the prime measurement required to produce an accurate yield map, the effect of other variables on the measured flow of grain are also to be considered. 2.3 Methodology To quantify the effect of varying crop density and moisture content on the measured crop flow through both mass detection and volumetric sensors, a number of trial fields were established over a period of three seasons. Grain samples were regularly taken from the combine bubble auger as clean grain flowed into the combine grain tank. Each of the sample points were logged using the combine s on board Global Positioning System. From tests it was concluded that it took 16 seconds from the crop being cut to grain reaching the grain tank. Therefore, a 16 second time delay was allowed between, firstly, logging the GPS position and, secondly, taking the grain sample. This enabled the grain sample to be correlated to its true position. The samples were taken on an approximate grid of 30 metres, which correlates to approximately 12 samples per hectare (Appendix 1). The samples were sent to the laboratory for analysis of crop density and moisture content. The results of the laboratory analysis were entered into a Geographical Information System (GIS) for interpolation. The irregularly spaced samples were converted into a regular 10 x 10 grid using simplified Kriging. Contour maps and histograms showing the variations and distributions in grain density and moisture contents within the trial fields were generated for presentation purposes (Appendix 2 and 3). Mark Moore (1997) 8

4 To ascertain any difference between mass detection and volumetric yield measurement in a field situation, the contour maps of grain density and moisture content were integrated with yield and gross margin data. The gross margin data was obtained from farm records (Appendix 15), and the yield data from a yield mapping combine using a mass detection yield sensor. Contour maps were produced showing the percentage error and the actual yield and gross margin errors between both mass detection and volumetric yield measuring systems (Appendix 4 and 5). A full statistical analysis was carried out to establish the effect of varying crop density and moisture content on the accuracy of the two yield measurement principles, and the resulting yield and gross margin maps produced from data. To isolate the errors between volumetric and mass flow measurement it was assumed during this analysis that the data derived from the mass flow sensor was accurate, however, in reality this would not be the case as the mass flow sensor is susceptible to errors which are quantified in chapter The extent of within field variation in crop density Baldwin (1984) summarised the effect of agronomic factors on the specific weight of grain, and concluded that there are two major aspects: 1. Occurrences which adversely affect the growth of the plant during the grain fill period will tend to reduce the specific weight as well as yield. These would include a large number of factors, such as drought. Water available to a crop is mainly influenced by physical soil properties such as soil type, compaction and depth of soil. He suggested that a combination of reduced rooting and take-all will have a high degree of influence on the specific weight of grain. Nutrient and trace element deficiencies, leading to leaf scorch, disease, lodging, etc. can all affect the grain fill and hence the resulting specific weight. 2. Occurrences which result in wheat producing a very large number of grain sites may in turn result in the plant being unable to fill the grain. There is little value in producing a crop with a 12t/ha potential in a field with only a 6t/ha potential. Baldwin compared this to the Black Fen near the Wash, where wheat s grown on fen peat soils are quite lush in their early growth Mark Moore (1997) 9

5 stages, but result in low specific weights because their roots are impeded by an acid drummy peat layer causing the plant to almost collapse in dry summers. Fenwick (1990) analysed different cereal varieties for variations in quality factors such as specific weight and Hagberg falling number. He showed that quality factors between different varieties, and also within the same variety, vary considerably. Table 2.2 quantifies the variations in crop density that existed within the six trial fields of winter wheat. The average variation in specific weight was found to be 9.76%. However, large variations, of up to 16% (Football Field ), that existed within the fields suggests that using a volumetric system could lead to significant errors when measuring the instantaneous yield, and therefore could have an adverse affect on the accuracy of a resulting yield map. Field Name Year Field Size (ha) Table Within field variation of grain density for six fields Number of Samples Min. value (kg/hl) Max. value (kg/hl) Mean value (kg/hl) Std. Dev. (kg/hl) Variation (%) Football Football * Oak Field * Oak Field * Underwood Underwood * Average * - part of field allocated to headland set-aside Varieties : Football Field 1994 Spark Football Field 1996 Oak Field 1995 Oak Field 1996 Underwood 1994 Underwood 1995 Consort Rialto Rialto Riband Rialto Mark Moore (1997) 10

6 Variation in Crop Density Date harvested - 14 August 1994 Crop - Winter Wheat Variety - Spark Meter Specfic Weight of Crop (kg/hl) Football Field Meter Figure Variation in grain density - Football Field (1994) Figures 2.1 and 2.2 illustrate the distribution of the variation in crop density within Football Field (1994). In theory a yield sensor based on the principle of mass detection compensates for any variation in grain density. Therefore, yield data derived from such a sensor has the advantage of not being influenced by any variation in crop density that may exist within fields. No automatic compensation takes place for a volumetric based sensor, which requires crop density to be constant if its accuracy is to be maintained. Therefore, given the extent of the within field variation in crop specific weight, it can be concluded that this has major implications for the accuracy of data obtained from yield sensors based on the principle of volumetric yield measurement. However, little work has been done to establish the degree of Mark Moore (1997) 11

7 error introduced into yield data derived from a yield sensor based on the principle of volumetric measurement. Distribution of Crop Density - Football Field (1994) Area of Field (%) Specific Weight (kg/hl) Figure Distribution of crop density - Football Field (1994) Many of the conclusions drawn from tests carried out on volumetric yield sensors concur that careful and regular calibration of crop density is a prime requirement if the sensor is to be accurate, Schroder et al. (1995) and Auernhammer et al. (1993). To quantify the effect of varying crop density and the frequency of calibration required to maintain accuracy on data obtained from a volumetric yield sensor, comparisons were made between the principle of mass detection, which in theory compensates for any variations in specific weight, and the principle of volumetric measurement, which assumes the specific weight of crop is constant until re-calibrated. For each of the six fields that were grain sampled, two values for total yield were obtained. The first value was measured using a mass detection yield sensor. The second value was derived from the first by integrating the measured yield with crop specific weight data. In the second instance, the specific weight of the crop was assumed to be constant, therefore simulating data derived from a volumetric yield sensor which had been calibrated once for the Mark Moore (1997) 12

8 field. The single calibration factor used to calculate errors was related to the average crop density measured for the field from the grain samples. The total field weights from the simulated volumetric sensor were compared with the results from mass detection sensors for the six trial fields, in Table 2.3. It was found that the difference between the total weights obtained from a volumetric system were similar to those of the mass detection system. The average difference between the two different methods of measuring total yield was -0.72%. Therefore, providing that a single calibration factor related to the average crop density is used, a volumetric system was shown to compare with the results obtained from a system based on the principle of mass detection. In a volumetric system, the heavy grain will tend to compensate for the light grain across the field, and result in an equally accurate figure for total weight. This supports the conclusions drawn by Auernhammer et al. (1993) and Reitz et al. (1996). Table Mass detection versus volumetric yield measurement Field Year Calibration Factor (kg/hl) Comparison of total yields from individual fields Total Yield - Volumetric Sensor (t) Total Yield - Mass Detection Sensor (t) Difference (%) Football Football Oak Field Oak Field Underwood Underwood Average Value The effect of volumetric yield measurement on yield and gross margin maps Recognising that the instantaneous yield is the primary measurement required to produce a yield map, the total weight of grain becomes of little importance. The data relating to varying crop density was integrated with the corresponding yield data to demonstrate potential errors associated with measuring the instantaneous yield with a volumetric sensor. In this instance, it was assumed that the system had been calibrated once with a value that represented the average crop density for the field. Table 2.4 illustrates the effect on a yield map if the crop density is assumed to be constant. Mark Moore (1997) 13

9 Of the six fields that were sampled, the average maximum variation in crop density was 9.76%. This variation, when ignored and assumed to be constant, produced a maximum potential error of 0.72t/ha on the yield maps for the six fields. Therefore, based on a single calibration value a volumetric yield sensor would under-estimate the extent of the yield variations. For example, having assumed the crop density to be constant, the yield for Football Field in 1994 would have been under estimated by -0.36t/ha to 0.58t/ha (standard deviation of 0.20t/ha). Table Volumetric yield measurement - Potential yield map errors associated with Field Year Calibration Factor (kg/hl) constant crop density Yield map error range - based on one constant calibration factor Min. Yield Max. Yield Standard Error (t/ha) Error (t/ha) Deviation Maximum Potential Error (t/ha) (t/ha) Football Football Oak Field Oak Field Underwood Underwood Average Value The data on crop yield and specific weight from the six trial fields were processed and maps illustrating the difference between mass detection and volumetric yield measurement produced. The difference between the two systems correlates to an error introduced into a any system based on the principle of volumetric yield measurement. Figure 2.3 illustrates one of the maps for Football Field (1994). When the specific weight of the crop was assumed to be constant, the distribution of the error (%) within Football Field (1994) and the potential implications for instantaneous yield measurement on the yield map is shown in Figure 2.3. Error expressed as a percentage The effect the error has on the yield map (t/ha) Mark Moore (1997) 14

10 600 Potental errors using a volumetric yield sensor to measure the flow of grain 600 Potential errors using a volumetric yield sensor to measure the flow of grain 500 Crop density assumed to be a constant 72 kg/hl 500 Crop density asssumed to be a constant 72 kg/hl Meter 300 Meter Football Field - (1994) 0 Football Field Meter Meter Error (%) Yield Error (t/ha) Figure Volumetric yield measurement - Distribution of potential yield map errors For each of the six fields, further analysis was carried out on the data by integrating it with Gross Margin data. Table 2.5 quantifies the effect on calculated Gross Margins assuming crop density to be constant. It was found that an average potential error of 83/ha was produced in the resulting Gross Margins when the average variation in crop density of 9.8% was ignored. The maximum Gross Margin error range was - 42/ha to 67/ha ( 109/ha) in Football Field in This illustrates that the derived Gross Margin data is very sensitive to the initial yield data, and therefore, errors are amplified. Figure 2.4 illustrates the distribution of potential Gross Margin map errors associated with the measurement of crop flow with a volumetric yield sensor. Table Volumetric yield measurement - Potential Gross Margin map errors associated with constant crop density Gross Margin map error range - based on one calibration factor Field Year Calibration Min. GM. Max. GM. Standard Maximum Mark Moore (1997) 15

11 Factor (kg/hl) Error ( /ha) Error ( /ha) Deviation ( /ha) Potential Error ( /ha) Football Football Oak Field Oak Field Underwood Underwood Average Value Football Field Potential errors using a volumetric yield sensor to measure the flow of grain Crop density assumed to a contant 72 kg/hl Meter Gross Margin Error ( /ha) Meter Figure Volumetric yield measurement -Distribution of potential Gross Margin map errors The errors in the Gross Margin map were considered to be significant, (up to 109/ha Gross Margin) and could, therefore, be a major influence for management decisions to be taken for the field. For example, if the cost of sub-soiling part of the field was 30/ha, the error on the Gross Margin map could suggest to the farmer that is not economic to sub-soil, when in fact it would be economic to carry out subsoiling in this example. In the above analysis, it was assumed that only one calibration value was used in conjunction with the volumetric yield sensor when harvesting the whole field. Auernhammer et al. (1993), state that a considerable amount of effort is required to continually update the calibration Mark Moore (1997) 16

12 value for specific weight to ensure volumetric yield meter accuracy. In practice this could be carried out for each grain tank load, however, in the farm situation it is often only carried out once, at the start of each field, Scott (1996) as time pressures during the harvesting operation are great. To demonstrate the impact of yield mapping with the volumetric principle of yield measurement, even when care was taken to regularly up-date the calibration value, further analysis was carried on the data. For each of the six fields, two individual harvest runs with the combine were chosen at random, (Appendix 6). Table 2.6 illustrates the degree of the variations in specific weights that were measured in each of the 12 harvest runs (two runs for each field). Field (year) Table Variation of crop density for 12 single harvest runs Run No. Number of samples Min. Value (kg/hl) Max. Value (kg/hl) Mean Value (kg/hl) Std. Dev. (kg/hl) Variation (%) Football Field Run (1994) Run Football Field Run (1996) Run Oak Field Run (1995) Run Oak Field Run (1996) Run Underwood Run (1994) Run Underwood Run (1995) Run Average Values The average maximum variation in specific weight in a single harvest run was found to be 7%, with a maximum of 13% (Football Field run 1). Mark Moore (1997) 17

13 The measured yield data from the mass detection system was extracted for each single run, and was integrated with the corresponding specific weight data from the grain samples. The degree of error was calculated for a volumetric yield sensor, which as stated previously, would have assumed the crop density to be constant. For each harvest run comparisons were made between mass detection and volumetric systems. The results of the analyses are shown in Table 2.7. Table The influence of variations in crop density on instantaneous yield during a single harvest run if measured using a volumetric system Field (year) Run No. Calibration Value Yield Error Range Min. Yield (t/ha) Max. Yield (t/ha) Standard Deviation (t/ha) Maximum Potential Error (t/ha) Football Field Run (1994) Run Football Field Run (1996) Run Oak Field Run (1995) Run Oak Field Run (1996) Run Underwood Run (1994) Run Underwood Run (1995) Run Average Values Assuming that it is impractical to stop the combine harvester during a harvest run and recalibrate the yield meter, Table 2.7 illustrates that significant errors can still occur. Note that the principle of volumetric yield measurement does not compensate for any variation in the specific weight of the crop within the harvest run. Therefore, even if particular care is taken in continually updating the calibration value for every grain tank load, the errors demonstrated in Table 2.7 represent the minimum errors that can be practically obtained within the trial fields, when measuring instantaneous yield with a volumetric sensor. The 12 individual harvest runs produced an average maximum potential yield error of 0.53t/ha, which occurs when the average within harvest run variation in crop density of 7% is ignored. The maximum error calculated during a single harvest run, if instantaneous yield was measured with a volumetric system, was 0.85t/ha (-0.32 to 0.53) in Football Field in Mark Moore (1997) 18

14 Figure 2.5 compares yield measurement taken with the mass detection sensor against that for a volumetric system during two different harvest runs in Football Field (1994). The first harvest run is represented by the two blue lines and the second by the two red lines. In run 1, the difference between the two blue lines indicates the error associated with assuming crop density to be constant, as is accepted with volumetric yield measurements. Similarly for run 2, the error is represented by the difference between the two red lines. 7.5 Comparison of Mass Detection Vs Volumetric yield sensors in varying crop densities Yield (t/ha) Distance Travelled (m) Mass detection - Run 1 Volumetric Sensor - Run 1 Mass detection - Run 2 Volumetric Sensor - Run 2 Figure The influence of variations in crop density on yield in a single harvest run Football Field (1994) Figure 2.5 illustrates that during run 1 (blue lines) a volumetric system would have under estimated the yield between -0.28t/ha and 0.57t/ha (standard deviation -0.23t/ha), as compared to the mass detection system which would have taken the variation of specific weight into account. A similar error was calculated with run 2 (red lines), where the volumetric system under estimated yield between -0.32t/ha and 0.53t/ha (standard deviation t/ha). Mark Moore (1997) 19

15 For each of the 12 single harvest runs, the yield data was also integrated with Gross Margin data to analyse the effect on Gross Margins between mass detection and volumetric systems. The results shown in Table 2.8 quantify the effect of measuring the instantaneous yield with a volumetric yield sensor on calculated Gross Margins in a single harvest run. Of the 12 harvest runs analysed, it was found that an average maximum potential error of 62/ha existed in the resulting Gross Margin calculations when the average variation in crop density in a harvest run was assumed to be constant. Table The influence of variations in crop density on Gross Margin during a single harvest run when calculated from a volumetric system Field (year) Run No. Calibration Value GM Error Range Min. Yield ( /ha) Max. Yield ( /ha) Standard Deviation ( /ha) Maximum Potential Error ( /ha) Football Field Run (1994) Run Football Field Run (1996) Run Oak Field Run (1995) Run Oak Field Run (1996) Run Underwood Run (1994) Run Underwood Run (1995) Run Average Values The maximum Gross Margin error range was - 33/ha to 67/ha ( 99/ha) in Football Field run 1 (1994). Again, these represent the minimum errors that can be practically achieved for a volumetric yield sensor within the six trial fields. It was concluded that the variations in crop density have a significant effect on Gross Margin calculations if the variance is assumed to be constant during a harvest run, as illustrated in Figure 2.6. Oak Field (1996) is illustrated in Figure 2.6 comparing two different harvest runs. For each harvest run the calculated Gross Margin is represented by two lines, one from a mass detection system and the other from a volumetric system. Each harvest run also depicts the correct Gross Margin which is indicated by the mass detection line. The difference Mark Moore (1997) 20

16 between the two lines indicates the calculated Gross Margin error associated with measuring crop flow with a volumetric system. 900 Potential Gross margin errors when measuring crop flow with a volumetric yield sensor 850 Gross Margin ( /ha) Distance Travelled (m) Mass detection - Run 1 Volumetric Sensor - Run 1 Mass detection - Run 2 Volumetric Sensor - Run 2 Figure The influence of variations in crop density on Gross Margin calculations in a single harvest run - Oak Field (1996) Figure 2.6 illustrates that during run 1, (blue lines), the Gross Margin value calculated if using a volumetric system would have been under-estimated by between - 9 /ha and 88/ha, (standard deviation - 25 /ha), compared to that calculated if using a mass detection system which automatically allows for the variation of specific weight. A similar error was calculated with run 2, (red lines), where the volumetric system under estimated the yield between - 1/ha and 56/ha (standard deviation 15/ha). It can be demonstrated, therefore, that significant errors are present in Gross Margin calculations if the yield is measured with a volumetric yield sensor Summary Mark Moore (1997) 21

17 Concluding that a volumetric yield sensor assumes the specific weight of the crop to be constant, then the within field variability of crop density, as demonstrated in the six trial fields, has profound implications for any yield sensor based on the of principle of volumetric yield measurement. However, the error range is dependent on the frequency of calibration of such a device. Within the six trial fields, a maximum error of 10% was demonstrated for a volumetric system that had been calibrated once for the whole field. This could be reduced to a minimum error of 7% if the system is re-calibrated at the start of each individual harvest run. 2.5 Grain Moisture Content Inevitably, grain moisture content will also vary within fields. If not taken into account when interpreting a yield map, the information that the map provides to the farmer may be open to misinterpretation. The problem is associated with the fact that varying grain moisture contents do not allow a direct comparison of grain yield from different parts of the same field. Obviously, if the grain moisture content varies within fields, then the grain yield will also vary as the moisture contained within the grain is a proportion of the grain weight. Therefore, depending on when and where the crop was harvested within the field, the yield may vary accordingly. This is due to changes in grain moisture content and not as a result of an actual increase in the amount of grain. The problem is exaggerated when the crop is harvested at high moisture contents and then dried to a constant level. In these circumstances it is important that the variability of the grain moisture content is taken into account in order to allow the yield map to be corrected. To quantify the effect of varying grain moisture contents on yield and Gross Margin maps, four trial fields were established. Grain samples were taken from various points within each field, (Appendix 1) and sent away for analysis to determine the moisture content. The resulting data from the analysis was used to produce contours maps illustrating the within field variability of the grain moisture content, (Appendix 3). The derived grain moisture content maps were then integrated with yield and Gross Margin data to produce further contour maps (Anon. 1996). These maps illustrate the errors contained within yield and Mark Moore (1997) 22

18 Gross Margin maps, where varying grain moisture was not taken into account (Appendix 5). When computing the error a standard grain moisture content of 14% was used, therefore, all yield and Gross Margin data has been adjusted to a 14% grain moisture base line The extent of within field variation in Grain Moisture Content Table 2.9 presents a summary of the within field variability of grain moisture from the four trial fields. The average grain moisture content within the four trial fields was 14%. In 1996, however, the weather was not so amenable towards harvesting, thus resulting in an average grain moisture content of 15% compared to the drier year of 1995 when the average moisture content was 12 %. The range of grain moisture content variability within the trial fields produced an average standard deviation of 0.56%. The standard deviation was relatively stable across both fields in 1995 and Therefore, the range of grain moisture content within a field appears to be reasonably static while the average level is more variable. However, the data has been gathered from only two years and four fields. The field that exhibited the most variation in grain moisture content was Oak Field in 1995, with a variation of 4%. The remaining three trial fields produced a variation of 3%. Table Within field variation of crop moisture content for four fields Field Name Year Field Size (ha) Number of Samples Min. value (%) Max. value (%) Mean value (%) Std. Dev. (%) Variation (%) Football * Oak Field * Oak Field * Underwood * Average * - part of field allocated to headland set-a-side Varieties : Football Field 1996 Consort Oak Field 1995 Oak Field 1996 Underwood 1995 Rialto Rialto Rialto. Mark Moore (1997) 23

19 Grain Moisture Content (%) :48 11:16 11:45 12:14 12:43 13:12 13:40 14:09 Time of day (Hour:Minute) Figure Within field variation of grain moisture content at various times of the day The first is the fact that the grain picks-up moisture overnight, and by morning is damp. Therefore, when harvesting started at 11.00am the grain was still relatively damp as shown in Figure 2.7. However, as the day progressed, the crop became drier as the heat from the sun became more intense; thus there is a downward trend in the data. By mid afternoon the crop had become approximately 1.5% drier than when harvesting started. The second cause is due to physical factors within the field, such as shade from trees and hedges, or the fact that the crop is not ripe in that part of the field. This is also demonstrated in Figure 2.7 and is represented by the spread of data points at any particular time of the day The effect of varying grain moisture contents on yield and gross margin maps If the yield data collected by the combine during harvesting is not corrected to a constant 14% moisture content, the yield maps will be subjected to the errors as presented in Table Mark Moore (1997) 24

20 The damper year of 1996 produced a positive error, as the grain was slightly heavier than it would have been if it were at 14% moisture content. The average error contained within the two yield maps produced from 1996 was 0.14t/ha (sd t/ha). Conversely, the drier year of 1995 produced an average negative error of -0.14t/ha (sd t/ha) as the grain was slightly lighter. A maximum spot error of -0.29t/ha was evident within Oak Field (1995) which also produced the largest error range of 0.36t/ha. Table Potential yield map errors associated with varying grain moisture Field Name Year Yield map error (t/ha) Average Minimum Maximum Std. Dev. Football Oak Field Oak Field Underwood The distribution of yield map errors associated with not adjusting yield values for grain moisture within Oak Field (1995) is illustrated in Figure 2.8. The error contained within the yield map is expressed as both a percentage and as actual yield (t/ha) Yield map error due to grain moisture correction Grain corrected to 14% moisture content Yield map error due to grain moisture correction Grain corrected to 14% moisture content Oak Field (1995) 0 Oak Field (1995) Error due to moisture correction (%) Error due to moisture correction (t/ha) Mark Moore (1997) 25

21 Figure Distribution of potential yield map errors Table 2.11 presents the errors as Gross Margins ( /ha). As with the errors encountered as a result of yield mapping with a volumetric system, the Gross Margin data is very sensitive to grain moisture content errors. The damper year of 1996 produced a positive average Gross Margin error of 12/ha (sd - 5/ha), while the drier year of 1995 produced a negative average error of 12/ha (sd - 3/ha). Oak Field (1995) also produced a maximum spot error of -27/ha whilst Oak Field (1996) produced a maximum error range of 13/ha. The distribution of the errors within Oak Field (1995) are illustrated in Figure 2.9. The error within the Gross Margin map is expressed as /ha. Table Potential Gross Margin map errors associated with varying grain moisture Field Name Year Gross Margin map error ( /ha) Average Minimum Maximum Std. Dev. Football Oak Field Oak Field Underwood Mark Moore (1997) 26

22 Gross Margin map error due to grain moisture correction Grain corrected to 14% moisture content Oak Field (1995) Error due to moisture correction ( /ha) Figure Distribution of Gross margin map errors Summary The within field variation of grain moisture content was found to be less variable that grain density. With less variation, grain moisture content had less influence on yield and gross margin maps. This was mainly as a result of the trial fields being harvested without interruption from over night rain. However, when making a direct comparison of yield levels from year to year, grain moisture content needs to be considered. It was concluded that the moisture content of grain is more likely to vary from season to season rather than within a single day s harvesting. In these circumstances grain weight should be corrected to a standard moisture content, normally 14%mc, allowing the direct comparison of yield levels within fields from a number of years. Mark Moore (1997) 27