Techniques to Predict Agricultural Droughts - Monitoring and Predicting Agricultural Drought: A Global Study - page 40

CHAPTER FOUR 1

2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46

T echniques to Predict Agricultural
D roughts
Z EKAI ¸EN AND VIJENDRA K. BOKEN

[40],

Line
——
-0.1
——
Norm
PgEn

In general, the techniques to predict drought include statistical regression,
time series, stochastic (or probabilistic), and, lately, pattern recognition
techniques. All of these techniques require that a quantitative variable be
identified to define drought, with which to begin the process of prediction.
In the case of agricultural drought, such a variable can be the yield (pro-
duction per unit area) of the major crop in a region (Kumar, 1998; Boken,
2000). The crop yield in a year can be compared with its long-term aver-
age, and drought intensity can be classified as nil, mild, moderate, severe,
or disastrous, based on the difference between the current yield and the
average yield.

[40],

St atistical Regression
Regression techniques estimate crop yields using yield-affecting variables.
A comprehensive list of possible variables that affect yield is provided in
chapter 1. Usually, the weather variables routinely available for a histor-
ical period that significantly affect the yield are included in a regression
analysis. Regression techniques using weather data during a growing sea-
son produce short-term estimates (e.g., Sakamoto, 1978; Idso et al., 1979;
Slabbers and Dunin, 1981; Diaz et al., 1983; Cordery and Graham, 1989;
Walker, 1989; Toure et al., 1995; Kumar, 1998). Various researchers in
different parts of the world (see other chapters) have developed drought
indices that can also be included along with the weather variables to esti-
mate crop yield. For example, Boken and Shaykewich (2002) modifed the
Western Canada Wheat Yield Model (Walker, 1989) drought index using
daily temperature and precipitation data and advanced very high resolu-
tion radiometer (AVHRR) satellite data. The modified model improved
the predictive power of the wheat yield model significantly. Some satellite

40