AGRI4401 Assessment 1
Management Zones Exercise 2019
You are required to work on your own.
Worth 55%, Due 5 pm Friday 23 August 2019
Spatial prediction and mapping can be used by agronomists and farmers to understand soil, plant growth and yield variability in paddocks. Such maps can be made when there are relatively few data points (<500 e.g. from a collection of soil samples in a paddock) or with more data intensive measurements (e.g. remote sensing, yield monitor on a harvester). The data from a number of maps of different types (e.g. gamma radiometric/EM map with yield maps) can then be combined into relatively uniform zones using cluster analysis. Areas with similar characteristics (e.g. soil characteristics and/or yields/production) can then be used for taking soil samples and paddock observations to better understand the cause of the variability. The same zones, or new/different zones, may be used to apply different management (called management classes) (e.g. application of an input like lime or phosphorus, spraying of weeds in certain areas, different nitrogen rates aiming for different yield goals etc.).
In this exercise you are given a set of field data for Paddock BF66 on Bungulla farm in the Tammin area. The paddock is about 244 ha and the soil varies between alkaline heavy clay, lateritic gravel, granite outcrops and loamy clay. As an agronomist you are given some soil and crop yield data for paddock BF66 and you need to produce a report with recommendations for the farmer to make some immediate (current season – Sections 1 and 2 of this report) and longer term decisions (section 3 of this report) to improve the productivity of this paddock.
Use references where appropriate. See the marking guideline below for allocation of marks/effort. Note that there is no perfect way to do this analysis and it will involve some trial and error – sometimes, the simple solutions work best.
The report should consist of about 3500 to 4000 words excluding the title page and attached maps (use size 12 font). There are three sections – 1) current soil pH decisions, 2) current yield goals and 3) future site investigations). Start each section on a new page. Attached all the relevant maps at the back and refer to all attached maps in the text (Fig 1, Fig 2 etc.). Note that you must be concise in your writing.
The three sections: each section should be in the format of a short paper.
Introduction (must be brief) – to outline the aim of the section and provide some background to the different data sets you are using in that section (where from etc.).
Methods – explain the steps used to derive the various maps (single or combined – which data sets were combined to produce e.g. Potential Management Class (PMC) maps and why, which software etc.) Include enough information so that someone else could repeat why you were doing, without unnecessary detail (e.g. you don’t have to explain all the buttons/you clicked or how a program like VESPER works etc. – just explain what settings you used – e.g. local variogram because >500 data points etc., so that someone could repeat it).
Results and Discussion – present and discuss the results (i.e. what recommendations you made and what evidence/information was used make these recommendations (which maps you have used and why; other references to the literature etc.) Remember to label the map captions correctly and refer to the maps in the text – all attached Figures (maps) should be referred to in the text.
Recommendations – Conclude with a short summary of the recommendations/key points.
The aim of the assignment
As an agronomist you are given some soil and crop yield data for paddock (Bungulla Farm) BF66 and you need to make some immediate (current season Sections 1 and 2) and long term decisions (section 3) to assist the farmer.
Yield goals – Based on the yield data for 2008, 2009 and 2011 (Files – BF66 Canola yield08 short.csv BF66 Wheat yield09 short.csv and BF66 Wheat yield 11 short.csv), you should map yields in each of the years to assess spatial yield variability in BF66. Determine if there is temporal variability. Then advise the farmer if he/she should identify different management classes (based on the yield variability – i.e. the farmer might apply different amounts of input onto different areas of the field) or if he/she should treat the paddock as one (i.e. uniform application of inputs over the whole field) for the coming season.
Imagine that the farmer is going to seed Mace wheat in the paddock BF66. You should determine the yield potential (yield goal/s) for the different areas of that field (or the whole field if it is to be treated uniformly). You should explain the basis for calculating the different yield goals (removal of nutrients in grain, or aiming for maximum or average yield potential based on past yield data). You should also produce a management class map (only based on the yield data – which can combine the yield maps over the three years (or two of the wheat years?) – depending on temporal variability). The map should show the different zones (areas) associated with the management classes (unless you recommend treating the field uniformly), along with recommendations for application of nitrogen and phosphorus in the coming season.
Soil pH decisions – based on existing soil pH data (File – BF66 Soil pH.csv), you are required to make recommendations for management of soil pH on BF66 (given the soil pH data). First map spatial variability for soil pH, then decide/recommend if the farmer should apply lime and, if so, how much (with reasons – you will need to do some of your own research here – one source might be the Department of Agriculture and Food WA website https://www.agric.wa.gov.au/soil-constraints/developing-liming-program?page=0%2C0 ) or the Booklet produced by Gazey and Davies (2009) (Soil acidity : a guide for WA farmers and consultants), which is on LMS. As part of this, you need to specify if the lime should be applied at a single rate to the whole field or applied at different rates, according to soil pH variability, and how many management classes (for lime application rates) should be used. Attach all relevant maps. The soil pH(CaCl2) is given at three depths – base your lime recommendations on the top 30 cm of soil i.e. over the three depths 0-10 cm, 10-20 cm and 20-30 cm depths, being the soil depths with most of the plant roots.
Combining the information into ‘potential management classes’, for more detailed investigation (You will need to use all the files BF66 radiometrics and EM cleaned.csv, BF66 Soil pH.csv as well as BF66 Canola yield08 short.csv, BF66 Wheat yield09 short.csv and BF66 Wheat yield11 short.csv).
You have advised the farmer that you need to ground-truth the information collected so far, so you can better assess the reasons for the yield variability and to decide if there are other factors having a major effect on yield (water-logging, gravel, weeds, pests etc.). You should make recommendations as to how many soil samples are required to characterise paddock BF66, based on potential management classes (derived using some or all the available information on soil type and yield potential – depending on the amount of variability and correlations to yield etc).
To do this you should firstly decide which are the important data/parameters to use for your potential management classes. For example, if some of the soil parameters (like EM38 (shallow), gamma radiometrics – K and thorium, pH at three depths) are correlated with yield, then your potential management zones would include those parameters as well as the appropriate yield data (the yield data coming from Section 1 above – e.g. use canola yield 08, wheat yield 2009 and wheat yield 2011, OR if the yield patterns are consistent across years then you might use a single, combined, yield map from averaging the normalised yield data)
Next produce a map of each of these selected parameters for your report (the yield maps are already done from section 1). Then you should combine this data (cluster analysis) to produce your final potential management class map. Include on this map the locations of your proposed soil samples/paddock observations. You should also explain what other factors (other than soil samples/analyses) will need to be investigated at these selected sites.
Key references: (available on LMS)
Fridgen et al. (2004). Management Zone Analyst (MZA): Software for subfield management zone delineation. Agronomy Journal, 96 (1):100-108
Bereuter (2011). Management zone delineation techniques on irrigated corn in Nebraska. MSc Thesis University of Nebraska
Some general guidelines
Open the data in Excel and familiarise yourself with the data. Note which columns the various variables are in. You will need to do spatial prediction (kriging)on the different variables in the various .csv files using the same paddock boundary and grid for all predictions (A free software program called VESPER (Variogram Estimation with Spatial Prediction with Error – developed by the Australian Centre for Precision Agriculture, Precision Agriculture Laboratory, University of Sydney) is used for spatial prediction.
You will then need to combine some/all of this predicted data into one csv file, based on what you find/determine with the data (i.e. which is correlated with yield) and then analyse the data in the combined file to come up with paddock/potential management classes. The analysis will be done with various statistics and cluster analysis in the software Management Zone Analyst (MZA – developed by the University of Missouri-Colombia). For example, if the coefficient of variation for yield is consistently >15% then it is highly suitable for making different zones for further analysis i.e. there is sufficient variation in the field to warrant making management classes. The output from MZA will need to be saved into another data file. Finally, mapping software called QGIS (Qantum GIS a free open source program) is used to produce the soil, yield and management class maps.
Example of the software ‘workflow’
Familiarise yourself with the data Excel. Clean/edit any columns you think necessary (you might change the names to be simpler etc.).
Open the .csv file in VESPER and select the correct data column that you want to do kriging on. Give the kriged output file an appropriate name. Then Draw/create the boundary and grid for spatial prediction on a 10 m grid. (First draw a boundary then generate the grid). I suggest you use a 10 m grid, otherwise it may take a long time to run each analysis, although 5 m is ok. REMEMBER to rename the Kriged file in the FILES Tab each time otherwise it will be overwritten (and to use the same grid file if you take a break when doing the analysis and close VESPER down and reopen it – to select it check the box in the KRIGING tab doe “Define Grid File” and select the grid file you have been using for this analysis. Follow the VESPER quick guide doc and Vesper_1.6_User_Manual pdf provided.
When the kriging is complete, copy all the data (from the separate kriged XXX.txt files – EM, radiometric yield etc) into one Excel file (save this as an Excel.xlsx and also “save as” a csv file from Excel.
Import the csv file into Management Zone Analyst (MZA) and look at the statistics for the different variables including the variance/covariance matrix and correlation matrix. After consideration of the information, test various combinations of the data/variables to produce the management zones (try between 2 and 6 zones). Follow the MZA quick guide doc, Management Zone Analyst user guide pdf and Overview of MZA discussion Powerpoint presentation Ortiz pdf
The output from MZA may have the file extension .dat – in which case rename it to csv. You can do this by opening in Excel (remember to look for “All files” in Excel) and then saving as MSDOS CSV format.
Import the file into QGIS and produce the required maps. Follow the QGIS quick guide doc
Complete the report.
Title (1 mark)
Short and descriptive
Section 1 (20 marks)
Describes the aim of the section and provide some background to the different data sets used in in the section.
Materials and Methods
Explains the steps used to derive the various maps. Includes enough information so that someone could repeat the process (e.g. the models and associated parameters used), without unnecessary detail (especially about how to use the program).
Results and discussion
Clearly and concisely describes the results. Includes some discussion (critical evaluation) of implications of the results (i.e. leading to the recommendations) and any perceived issues/weaknesses in the analysis or recommendations.
All recommendations are clearly articulated
Section 2 (20 marks)
As described above
Materials and Methods
Results and discussion
Section 3 (40 marks)
As described above
Materials and Methods
Results and discussion
Overall report (19 marks)
Overall map presentation, captions, legend etc.
Produces maps that are high quality – easily legible and have all the required features.
Good writing style, correct grammar, no typos or incorrect spelling
Structure and flow of ideas
Has a logical flow of ideas that is easy to follow
Consistent and suitable reference style
Assignment status: Already Solved By Our Experts
(USA, AUS, UK & CA PhD. Writers)