CFAB - About Us

Bengali

AN EXAMPLE OF
THE GENERATION OF A USER METRIC

click here
for a power point presentation of the User Metric concept.

In order to perform a risk analysis, it is necessary to determine the impact (or cost) of the occurrence of a particular environmental state. This information is best known by the user community and is shown in Figure 1. In the simple example we present here we assume that the user community has provided us with a yield reduction for each classification of rainfall intensity: If there is no rain, the yield is not reduced. Light rain would reduce crop yield by 30%, moderate rain by 50% and heavy rain by 80%. It is also assumed that input from the user community suggests that crop yield will be smaller the earlier that a crop is harvested because the crop may not have matured sufficiently. Here we assume that at 10 days before the optimal time of harvest the yield would be 80%.

Figure 3: The variation of yield for the various harvesting strategies for the cases of heavy, moderate, light and no rainfall. The impacts of various rainfall on harvesting is obtained from the user community. These yields have to be modified by the probabilities of the occurrence of each of the rainfall classifications obtained from the numerical weather and climate models. This information is static and provided by the user community.

The user community also provides possible harvesting strategies. For this example, we assume that there are six:

harvest 100% of the crop 10 days before harvest
harvest 75% of crop 10 days before harvest
harvest 25% at normal harvest time
harvest 50 and 50%
harvest 25 and 75%
harvest 0 and 100%

The latter strategy refers to the no action case.

The metric we envision contains three figures. The first figure user input data (Figure 1: this would be the upper right hand diagram in Figure 2)). A second figure is the probability of rainfall from the forecasting schemes. The third diagram is the crop yield based on aggregated risk which is a combination of the rainfall probabilities and the user community input. This is the figure that will aid the user community in making reasoned absolute decisions.

We consider two scenarios:

Scenario 1: 10-day forecast indicates a low risk of heavy rainfall (Figure 2)

Figure 2: Probabilities of rainfall occurrence during the ensuing 10-day period (left panel) obtained from the ensemble forecasts. Only a small risk of heavy rainfall (5%) is forecast and a 15% chance of moderate rainfall. Using the probabilities of rainfall and the statistics of the reduction of yield depending on the heaviness of rainfall and harvesting strategy (Figure 1 from user community), the yield based on the aggregate risk can be assessed for each harvesting strategy (left panel). For this case of low probability of heavy rainfall, the optimal strategy for maximum yield is no action. That is, maximum yield should occur with harvesting at the normal time or at full crop maturation.

• The probability pie chart indicates a 60% chance of no rain, a 20% chance of light rain, a 15% chance of moderate rain and a 5% chance of heavy rain. These probabilities are assessed from the plume diagrams (e.g., Figure 2) for the short-term forecasts. The second diagram in the User Metric would be Figure 4. As this diagram does not change between scenarios, it is not repeated here. The third diagram (bar chart) takes the data from the second diagram and ascribes the risk of a particular event occurring from the pie chart. Each bar denotes the probability of yield based on aggregated risk for each planting strategy. This diagram provides the most information for a user having to make a decision. The figure shows that yield grows monotonically from left to right with values ranging from 80% (harvest all crop early) to 82.5% for taking no action. Thus, in this particular climate environment, the farming community should take no action and harvest at the normal time.

Scenario 2: 10-day forecast indicates a high risk of moderate flooding (Figure 3)

Figure 3. Numerical models suggest a substantial risk moderate rainfall (20%) and heavy rainfall (15%). From Figure 3 such rainfall, if it were to occur would reduce yield by 50 and 80%. The yield based on aggregated risk (right hand diagram) suggests that the optimal strategy would to harvest early (80%). The worst decision would be to harvest late (63% yield).

• Probability pie chart indicates flood probabilities are 20% (no rain), 45% (light rain), 20% (moderate rain) and 15% (heavy rainfall).
• Panel 2 is the same as Scenario 1 because they depend on the harvesting strategy and yield associated with degree of flooding.
• Potential yields when weighted with the probability of rainfall occurrence shows a variation of 80% yield for immediate harvesting to 63% yield for no action. The prudent decision would be to choose to harvest immediately.

Labelled with ICRA