Search this site
Embedded Files
SIMAGRI-Ethiopia
  • Home
  • SIMAGRI Tutorial
  • Crop Simulation Info
  • User Feedback [Survey Form]
SIMAGRI-Ethiopia
  • Home
  • SIMAGRI Tutorial
  • Crop Simulation Info
  • User Feedback [Survey Form]
  • More
    • Home
    • SIMAGRI Tutorial
    • Crop Simulation Info
    • User Feedback [Survey Form]

SIMAGRI Tutorial

https://simagri-ethiopia.iri.columbia.edu/historical

http://simagri-ethiopia.iri.columbia.edu/forecast

Type a scenario name for running a crop simulation model with specific inputs (e.g., weather, soil, cultivar, and management)

  • Remember to first clear the existing scenario (which is empty) or the tool will not run

  • ONLY 4 characters (combination of number, alphabet, and symbols) are allowed for a scenario name. If the user's input is wrong, the Entry box turns into red color warning the input does not meet the requirement.

Select a weather station name from the dropdown list.

  • Initially seven stations in main agricultural areas were listed (red stars on the map below). Later centroids of 31 zones (green dots on the map) were added.

  • The daily rainfall, Tmin and Tmax were extracted from NMA-ENACTS database (1981-2018) and daily solar radiation was extracted from NASA-POWER.

Select a crop to simulate.

  • Once a user selects a crop, available cultivar types are automatically populated in #4) Cultivar section as shown below.

  • The genetic coefficients of each cultivar were calibrated by local partners or taken from literature. More thorough evaluation, particularly the cultivars of wheat and sorghum is required. DSSAT experts who are familiar with the genetic coefficients can refer to *.CUL, *.ECO and *.SPE files here used for the DSSAT simulations under SIMAGRI.

  • More descriptions on the cultivars can be found in the "Crop Simulation Info" section

Select a period (from Start year to End year) to run crop simulation model

  • Available weather observations are from 1981 to 2018. Therefore, the earliest Start Year (#5) and the latest End Year (#6) are 1981 and 2018 respectively (set up by default).

  • If a user has an interest in a specific period (e.g., the recent 10 years from 2009 to 2018), the default number can be adjusted.

  • If the user's input is beyond the available period (e.g., 2020), the Entry box turns into red color warning the input does not meet the requirement.

Select a target year as a benchmark to compare with the full distribution of the historical weather data.

  • Crop simulation results (i.e., yield) of this specific year are expressed as red dots in the boxplot of simulation results section on the right panel.

  • By default, a recent severe drought year, 2015 is selected. In this case, a user can check where the simulated yield using 2015 weather data (red dot) is located along with the full distribution (i.e., a box plot) and how much relative yield loss the 2015 drought year brought.

Select a soil type for crop simulation.

  • Soil characteristics are critical inputs in simulating soil water/nutrient dynamics and their interaction with crop growth. In spite of its importance, it is very difficult to have detailed soil information required by DSSAT crop models in reality.

  • Currently, the SIMAGRI tool takes a global gridded soil database developed by IFPRI at 10km resolution.

  • Each soil type has a unique soil ID (10 characters) to be used for DSSAT software. In the parenthesis, location ID (e.g., MELK), soil type (e.g., L representing Loam), and rooting depth (e.g., shallow) are shown in the dropdown menu. The acronyms of soil types, L, C, and CL represent Loam, Clay, and Clay loam respectively.

  • In addition to the default IFPRI soil profiles, rooting depth was reduced to less than 1m depth, in order to reflect degraded soil conditions in Africa. The root-depth adjusted soil profiles have a soil ID with "_" in the middle of the soil ID (as shown on the left).

  • DSSAT experts can refer to the soil input file (ET.SOL) here for more details.

Select an INITIAL soil water content as a percentage of Available Water-holding Capacity (AWC)

  • Initial soil water content represents soil water condition a day before planting.

  • AWC quantifies the amount of water available for plants that the soil can hold. The AWC is defined as the amount of water held by the soil between field capacity (FC) and permanent wilting point (PWP). Note that soil physical characteristics (FC, PWP, and AWC) vary depending on the soil type selected above in #8.

  • Initial soil water content calculation : Initial H2O [cm^3/cm^3] = IC_AWC_ratio * (FC - PWP) + PWP, where IC_AWC_ratio can be 0.3, 0.5, 0.7 and 1 based on the user's selection.

Select an initial NO3 level in the soil a day before planting.

  • In reality, it is almost impossible to have accurate NO3 information before planting. Therefore SIMAGRI takes a simple approach by setting up only two levels of NO3: Low and High. In the case of the High (65 N kg/ha) amount, initial 15 and 2ppm of NO3 are assumed for the first and second top soil layers, respectively. In the case of the Low (23 N mg/ha), 5 and 0.5 ppm of NO3 are assumed for the first and second layers. For other deeper layers, the NO3 amount is assumed zero.

  • NO3 amount [N kg/ha] can be derived from NO3 [ppm], soil bulk density and dept of soil layer (i.e., NO3 [N kg/ha] = NO3 [ppm] * BD * depth *0.1). For instance, initial soil with 15ppm of NO3 has 54 N kg/ha for a 30 cm depth of soil layer => 15ppm* 1.2 g/cm3 * 30cm *0.1 = 54 N kg/ha, when assumed bulk density = 1.2 g/cm3.

  • In general, an ideal high/low NO3 condition has 60~100/20~30 N kg/ha in the upper 30~50cm depth of soil.

Select a planting date for crop simulation, from a calendar.

  • The calendar was initialized as June 15, 2021, by default. The user can switch to another month or date by ignoring the default year (i.e., 2021) shown on the calendar. Only monthly and date will be reflected in the crop simulation.

  • The SIMAGRI uses the selected planting date as a fixed planting date when it runs the crop models for the historical weather observations. That is, for all years (1981~2018), the planting date is constant as 6/20/1981, 6/20/1982, 6/20/1983 .... 6/20/2018.

Type a reasonable number of planting density for the target crop and cultivar for crop simulation.

  • Note the unit is number of plants per square meters [plants/m2]

Select whether fertilizer is applied or not by selecting an appropriate radio button.

  • Once the "Fertilizer" option is selected, a table is automatically populated so that the user can type a specific date [days AFTER planting] and amount of the fertilizer [N kg/ha] application.

  • In reality, various types of chemical fertilizers are used. However, the DSSAT crop simulation model takes into account only N dynamics for crop growth. Therefore, the user has to drive N amount from the total fertilizer amount. For instance, 150kg/ha of NPK (15-15-15) is applied, N amount would be 150*15/100 = 22.5 N kg/ha.

  • Note: Current SIMAGRI allows only up to 4 fertilizer applications.

  • There are three options for irrigation: 1) No irrigation, 2) On Reported Dates, and 3)Automatic when required.

  • Option, 2) On Reported Dates, can be used when irrigation is applied a few times (in SIMAGRI, maximum of 5 times of irrigation is allowed) during the growing period.

(i) First, select an irrigation method in the Dropdown menu

(ii) Then, type when (i.e., Days after planting) and how much irrigation water for each column

  • Note that SIMAGRI takes into account irrigation AFTER planting (i.e., "Days After Planting" column takes only positive numbers. Therefore, in the case of irrigation before planting, the user needs to increase initial soil water content in #9.

  • Option, 3) Automatic When Required, applies irrigation water into the soil whenever water holding capacity in the soil drops down below the "Threshold" (% of maximum AWC) of the "Management soil depth" the user defines.

This automatic irrigation option may not be very realistic, but can be used to compute potential yield by allowing the assumption of no water stress.

Select whether to include Enterprise Budget analysis (Yes) or not (No).

  • Once the "Yes" option is selected, a table is automatically populated so that the user can fill out more detailed inputs including cost and expected income/crop price.

  • Enterprise budget analysis allows the user to look at different production combinations and see their financial impact.

  • It can be simpler or more complicated than the current SIMAGRI's calculation as follows.


Gross Margin [ETB/ha] = Revenues [ETB/ha] - Variable Costs [ETB/ha] - Fixed Costs [ETB/ha]


- Revenues [ETB/ha] = Yield [kg/ha] * Crop Price [ETB/kg] , where Yield values are taken from DSSAT simulated yield for the historical years.

- Variable costs include (i) fertilizer, (ii) seed, (iii) irrigation, and (iv) other costs

(i) Cost for fertilizer [ETB/ha] = N Fertilizer amount [N kg/ha] * cost [ETB/N kg], where N amount is from the user's input above in #13

(ii) Cost for seed [ETB/ha] can be calculated by taking into account planting density and seed price. For example, Cost for seed [ETB/ha] = planting density [# plant/m2] * seed price [ETB/kg] * seed weight [g/seed] * 10000 [m2/ha] * 0.001 [kg/g] *(1/seed survival rate)

**sample calculation: Cost for seed [ETB/ha] = 5 [#/m2]*10000* 6 [ETB/kg] *0.08 [g/seed] * 0.001 [kg/g] * (1/0.91) = 26 [ETB/ha] , where the price of hybrid maize seed is assumed as 600 ETB/100 kg

(iii) Cost for irrigation [ETB/ha] = irrigation water depth applied [mm] * cost [ETB/mm/ha], where water amount is from the user's input above in #14

(iv) Other variable costs [ETB/ha] may include pesticide, insurance, labor etc.

- Fixed costs [ETB/ha] may include interests for land, machinery, etc.

  • Once the user finished filling out all the inputs from #1 to #15, the user should click the blue button "Create or Add a new scenario" in order to save the current scenario set up (#1 ~ #15). The users can double-check if their input is correct or not by scrolling the grey bar to the end of the last column of the table.

  • If the users want to add another scenario, they can go up to #1~ #14 and make necessary changes, and then click the blue button "Create or Add a new scenario" again to add the new scenario to the Scenario summary table. Again, the user can check if all inputs are correct. If anything is wrong, the user can delete the row by clicking "x" button next to the first column of the table.

  • Users can save the current scenario set up by clicking "Download Scenarios" button and then re-run the same scenarios after importing the saved scenarios by clicking "Import Scenarios" button

  • #16 does not affect crop simulation, but helps to categorize dry/neutral/wet years among all historical observed years. Based on the selected months (e.g., July, August, and September in the example on the left), the SIMAGRI calculate total rainfall amount during the selected period for each year and match them with the simulated yields.

  • The user can drag the circle to adjust the selected period.

Once the user finished the scenario set up (i.e., satisfied with the scenario summary table above), the user should click the green button "Simulate all scenarios (Run DSSAT)" in order to run DSSAT crop simulations for all the scenarios.

  • Once the user clicks the green button, the browser will show a loading spinner while the crop simulation is in progress. The loading time will vary depending on how many scenarios the user created. Especially, "Forecast analysis" will require more time to run hundreds of years of synthetic weather data.

  • Once the simulation is done, output graphs will be automatically populated on the right side of the browser.

UNDERSTANDING CROP SIMULATION RESULTS

Simulated yield distribution shown in a boxplot

  • A boxplot is a way of displaying the distribution of data based on a five-number summary (“minimum”, first quartile (Q1), median, third quartile (Q3), and “maximum”). This type of graph allows you to quickly visualize the mean values ​​and the expected variability for a proposed scenario.

  • If the user hovers around the boxplot, minimum, maximum, and each quartile values (Q1, median, and Q3) are shown automatically (figure below on the right side).

  • How to interpret a boxplot? The first boxplot (from the scenario "MZ01") on the lower left shows that in the MZ01 scenario there is a 50% of chance that the yield is between 2001 and 2340 Kg/ha ("box" limits). On the other hand, the chance of obtaining 2161 Kg/ha or more is 50%, that is, it can be expected that in half of the years the yield will be at least 2161 Kg/ha (i.e., median). The probability of obtaining 2001 Kg/ha or less is 25%, that is, once every 4 years, yields less than 2001 Kg/ha can be expected. The same happens with yields above 2340 Kg/ha

Simulated yield distribution shown in a probability of exceedance curve

  • The Probability of Exceedance curve is another way to display the distribution of the simulated yield.

  • This graph allows you to estimate the probability of exceeding a certain value for different scenarios.

  • For the MZ01 scenario, there is a 50% probability of obtaining yields above 2156 Kg/ha (indicated by a black line). On the other hand, IN 3 out of every 4 years yields above 2001 Kg/ha are obtained (i.e., the probability is 75%), and once every 4 years your yield would be above 2340 Kg/ha (i.e., 25% prob. of exceedance.).

Simulated yield for each year (time-series)

  • The simulated yield corresponding to individual years can be found in the time-series graph.

  • In this graph, the user can find which year produces a higher/lower yield due to favorable/unfavorable weather/climate conditions. For example, in 2002 there was an exceptionally lower maize yield in Melkasa due to a severe drought.

  • If the user selects the option "compare data on hover" in the upper right corner (shown in a red box), the values of each point and the corresponding year are shown on the graph.

Yield Exceedance Curve in 3 categories

  • Based on the input #15 (i.e., seasonal total rainfall), the historical years and corresponding simulated yields are classified as dry, normal, and wet categories.

  • In the example output (on the left) resulted from the Melkasa weather data, we can see that there are more chances of extremely lower yields (~ 1500 kg/ha) in the dry category, probably due to drought. However, there are also higher chances of having higher yields (greater than 3000 kg/ha) in the dry category. This indicates that JAS seasonal total rainfall is not linearly related to the maize yield in Melkasa. This relationship may be different for different seasons or locations.

Download simulated yield/probability of exceedance/seasonal total rainfall in a CSV file

  • Simulated yield, probability of exceedance, and seasonal total rainfall are displayed in a table in columns, Yield_####, Y_Pexe_####, and Rain_####, respectively for each scenario (####).

  • The user can download each variable (i.e., simulated yield, probability of exceedance, and seasonal total rainfall) separately by clicking an appropriate grey button, as shown below.

  • The original simulation results can be sorted by clicking arrows next to the column name (highlighted in green boxes below)

Run Enterprise Budget Analysis

  • If the user provided inputs (crop price and cost) in #14, the SIMAGRI automatically calculates the expected gross margin based on the simulated crop yield and the user-provided cost data. For the calculation, the user should click the red button "Display figures for ..."

  • The original simulated yield can be adjusted by multiplying an adjustment factor. Crop simulation models cannot reflect all the factors in the natural system, particularly damages due to pest/disease. Therefore, simulated yields tend to overestimate actual yield. In order to take into account the overestimation, users can apply a yield adjustment factor and shift the simulated yields into a more reasonable range. For example, yield adjustment factor =0.8, 80% of the original simulated yields are used to compute gross margin.

  • Three different types of graphs, similar to the simulated yield are generated as shown below. Note that now the Y-axes in the boxplot and time-series graph are Gross Margin [Birr/ha], not Yield.

Download Gross Margin table into a CSV file

  • The user can download the computed gross margin data into a CSV file by clicking the grey button, "Download"

  • The table includes additional columns from the model output:

-PDAT: Planting date (YrDoy)

-ADAT: Anthesis date (YrDoy)

-MDAT: Physiological maturity date (YrDoy)

-HWAM: Yield at harvest maturity (kg [dm]/ha)

-NICM: Inorganic N applied (kg [N]/ha)

-IRCM: Season applied irrigation (mm)

Credit: Eunjin Han, Ph.D. at IRI, email: eunjin@iri.columbia.eduWalter Baethgen, Ph.D. at IRIJames Hansen, Ph.D. at IRIKesha Kumshayev, at IRI Jemal Seid Ahmed, at EIARKindie Tesfaye, Ph.D. at CIMMYTDawit Solomon, Ph.D. at CCAFS
Report abuse
Page details
Page updated
Report abuse