Monitoring Agricultural Drought

Decrease in rainfall below normal is a drought indicator commonly used by the Bureau of Meteorology and Geophysics (BMG), whereas the total area affected by drought is an indicator used by the Directorate of Plant Protection (Alimoeso et al., 2000,2002). Remotely sensed data such as normalized difference vegetation index (NDVI; chapter 5) is used by Lembaga Penerbangan dan Antariksa Nasional (Kushardono et al., 1999; Heryanto et al., 2002). The following sections describe how these indicators are monitored and drought forecasted in Indonesia.

Drought Monitoring and Forecasting System

The Indonesian BMG developed an operational seasonal climate prediction scheme for Indonesia in 1993. It is a statistical-analogue scheme based on a detailed historical analysis of rainfall data for 102 meteorological regions across the whole country. The forecast is issued in early March for the dry season (April-September) and in early September for the wet season (October-March).

The seasonal forecast products are forecast of seasonal monsoon onset dates at 10-day interval; seasonal cumulative rainfall; and monthly rainfall. There is an ongoing forecast verification process undertaken by the BMG, the results of which are shown in table 26.3 for the 1993-97 period. The verification of forecasts shows a good skill in predicting the onset period

Table 26.2 Rainfall (mm) at selected stations in rice-producing areas, l982-83a

Jan

Feb

Mar

Apr

May

Jun

Jul

Aug

Sep

Oct

Nov

Dec

West Java Jatiwangi 30-Year average

461

405

264

264

159

83

59

30

43

109

259

417

1982

538

612

245

310

90

20

2

1

0

55

40

Central Java Tegal

30-Year average

360 369

307

243

129

117

84

62

39

46

52

120

253

1982

403

346

274

236

12

55

2

6

0

21

14

62

1983 East Java Madiun

30-Year average

472 273

271

264

232

156

78

43

25

27

79

191

243

1982

118

183

330

259

0

0

3

0

0

0

NA

Denpasar 30-Year average

284 334

276

221

88

75

70

55

43

42

106

168

298

1982

334

267

145

53

2

1

0

3

0

1

79

35

1983

146

South Sulawesi Ujung Pandang 30-Year average

714

515

423

154

95

65

32

14

11

45

183

581

1982

648

482

434

109

77

6

0

0

0

0

30

323

1983

348

aUnderlining denotes dry season.

Source: USDA (1984).

aUnderlining denotes dry season.

of the wet season; reasonable skill in predicting the onset period of the dry season; and relatively poor skill in predicting the dry season rainfall, particularly during ENSO years.

Despite the availability of a reliable forecast for parameters such as the onset period for the wet season, decision-makers are unable to utilize the forecast because of its coarse resolution. Onset dates are indicated for a period of one month over a large area, and rainfall distribution is presented in terms of above-normal, normal, or below-normal categories for a six-month period over a large area. Based on the climate forecast received from the BMG, national organizations such as the departments of agriculture and water resources disseminate the climate information to provincial organizations in a routine manner. At present, these forecasts are used only as a general alert.

Another problem is that there is no consensus between related agencies on the threshold onset dates for the wet and dry seasons and the distribution of rainfall during these seasons. For example, a delay of up to 20 days

Table 26.3 Verification of seasonal prediction in Indonesia (percentage of districts in the respective categories)a

Onset of season Rainfall classification

Table 26.3 Verification of seasonal prediction in Indonesia (percentage of districts in the respective categories)a

Onset of season Rainfall classification

Year

Precise

Ahead

Later

Same

Close

Different

Wet Season

1993-94

84

5

11

52

45

3

1994-95

86

1

13

50

44

6

1995-96

83

8

9

49

42

9

1996-97

64

7

29

43

50

7

Average

79

5

16

49

45

6

Dry Season

1993-94

74

23

3

27

59

14

1994-95

56

10

34

63

28

7

1995-96

73

20

7

51

39

9

1996-97

76

19

5

21

58

21

Average

70

18

12

41

46

13

Source: ADPC (2000).

aPrediction of onset for the 102 seasonal prediction areas is estimated in 10-day (dekad) periods. Rainfall is considered normal if it is 85-115% of normal, above normal if it exceeds 115% of normal, and below normal if it is below 85% of normal. Precise means that onset occurred in the 10-day period predicted; ahead means that it occurred in an earlier dekad; and later means that it occurred in a later dekad. For rainfall classification, same means that rainfall total was as predicted, and different means that it was two categories from what was predicted.

Source: ADPC (2000).

aPrediction of onset for the 102 seasonal prediction areas is estimated in 10-day (dekad) periods. Rainfall is considered normal if it is 85-115% of normal, above normal if it exceeds 115% of normal, and below normal if it is below 85% of normal. Precise means that onset occurred in the 10-day period predicted; ahead means that it occurred in an earlier dekad; and later means that it occurred in a later dekad. For rainfall classification, same means that rainfall total was as predicted, and different means that it was two categories from what was predicted.

in the onset of the wet season may not upset established crop calendars at some locations because agricultural practices evolved over time are adjusted to these normal variations. Any variation beyond 20 days is likely to upset the established crop calendar, and the information on these thresholds could be of use to the end-users at the concerned locations. As another example, a 50% reduction of rainfall could still be enough to support the established crops in high rainfall zones such as Kalimantan, where average rainfall is around 3000 mm, because 1500 mm is enough to support a rice crop.

Crop Production Forecasting and Monitoring System

Forecasting rice production has long been a priority for the Indonesian government, particularly on the island of Java, which accounts for approximately 60% of the total rice production in the country. Data on the harvested area are collected on a monthly basis but, for the purpose of publication, are aggregated to a level that corresponds to time periods of recording of yields (using crop cuttings) in three sub-rounds: January-April, May-August, September-December. Official forecasts for rice production are made in mid-February, mid-June, and mid-August. All forecasts are disaggregated to provincial levels separately for padi sawah (wetland rice) and padi ladang (upland rice). Historically, the methods used to forecast harvested area have tended to rely on a combination of harvested area in the previous year and data available on planting in the current season.

The advance estimates consist of preharvest forecasts and quick estimates. Quick estimates refer to initial postharvest forecasts generally obtained through sample surveys and crop-cutting experiments. The sample surveys are conducted for rice, corn, soybeans, peanuts, cassava, and sweet potatoes. Data are collected by two field agencies, the Agricultural Extension Service (Mantri Tani) and the Subdistrict Statistical Office (Mantri Statistik). Data at the subdistrict level are collected through the discussions with village heads, the status of crops in block irrigation systems, and interviews with farmers and the sales agencies associated with the quantity of planted seed.

In the case of the crop-cutting experiments, each district is treated as a stratum, with the number of areas selected using random samples. One segment is selected randomly from each selected area. Households in each selected segment are requested to indicate whether they planted food crops and their expected harvest dates. A land parcel is selected from the household list, and one plot (2.5 x 2.5 m) is selected randomly. The crop inside the plot is harvested and weighed. The total number of sample plots for the crop-cutting experiments throughout Indonesia is 110,000. The crop cutting from 50% of the plots is done by the Sub-district Statistical Office and from the remaining 50% by the Agriculture Extension Service. The yield data thus collected are reported hierarchically (i.e., from subdistrict level to district, provincial, and national organizations).

The advance estimates from subdistrict to national levels are often tentative and subjective. No mechanism exists to validate the information flow. These estimates are revised from time to time, which poses difficulties for decision-making processes. For instance, for the 1997-98 El Niño, the initial estimates were around 45 million tons of rice, which were later revised to around 48 million tons. Therefore, decisions regarding rice imports could not be made accurately. The present system does not take into account the influence of rainfall and price variability on farmers' decisions to plant various crops. Hence, the present system has much scope for improvement, both in terms of lead time and accuracy.

Using Satellite Data LAPAN and the Research Institute for Soils and Agroclimate have evaluated the use of satellite data and GIS technology to monitor rice crop growth to estimate rice yield (Kushardono et al., 1999; Heryanto et al., 2002). The use of this technology is promising. The yield of rice can be estimated from the normalized difference vegetation index (NDVI) during the heading phase:

The above equation has been tested at four villages of Kronjo Subdistrict, Tanggerang District. The difference between the observed and estimated yield varied from 10% to 14% with an average of about 12% (Heryanto et al., 2002).

Using the Southern Oscillation Index Another approach to estimate the crop production in the coming season is the Southern Oscillation Index (SOI) data (Meinke and Hammer, 1997; Rahadiyan, 2002). A preliminary study indicated that the variation in national rice production could be explained partly by the SOI for April. The use of the April SOI was motivated by the evidence that after April the cumulative drought area increased rapidly, especially during El Niño years. Negative anomalies in rice production occurred mostly for El Niño years, while positive anomalies occurred for La Nina years.

Using Crop's Physical Appearance Drought severity is determined based on the proportion of the dry leaves in rice fields. If only the tip of the leaf gets dry, it is called slightly affected (Ringan); if one-quarter of the leaf is dry, it is called moderately affected (Sedang); if one-quarter to two-thirds of the leaf is dry, it is called heavily affected (Berat); and if all of the leaf is dry, it is called completely damaged (Puso). These drought symptoms are monitored every two weeks by the pest and disease observers at subdistrict levels. The data are sent to the district and then to the province and finally to the Directorate of Plant Protection. The Directorate of Plant Protection analyzes the data and provides a drought distribution map.

Drought Vulnerability Map The Directorate of Plant Protection has developed a map showing drought vulnerability in a district. Vulnerability is determined using frequency, intensity, and size of area affected by drought (Alimoeso et al., 2002). The analysis suggested that most of the vulnerable districts are located in West Java province, South Sumatra/Lampung, and South Sulawesi (figure 26.2). These three provinces are the main rice-growing areas of Indonesia. Among the three, West Java is the largest rice-producing province in the country. Any drought in West Java could have a significant influence on national rice production and food security. In West Java, districts considered to be very prone to drought are Indramayu, Bekasi, Sukabumi, Tasikmalaya, Bandung, and Cirebon. Loss of rice production in these districts increased significantly during El Niño years (figure 26.3) and could go up to 500,000 tons (Alimoeso et al., 2002).

From historical data, it was shown that, in general, the area affected by drought increased significantly during El Niño years (figure 26.4). However, from national production statistics, the impact of El Niño, apart from 1982, is not distinct, except for rice (figure 26.5). This condition appears due to a number of reasons (Suryana and Nurmalina, 2000; Meinke and Boer, 2002): (1) the statistics are based on calendar years rather than on El Niño years, (2) not all regions of the nation are affected by drought simultaneously, (3) shortage of water may force a farmer to switch from rice to secondary crops, (4) restricted water supply may reduce the area

Figure 26.2 Vulnerability of rice-growing area to drought by district (from Alimoeso et al., 2002).

planted under irrigation, but yield of crops may increase due to higher solar radiation, and (5) production may be affected in the year following an El Niño event because farmers have less money to spend on fertilizers or insecticides.

0 0

Post a comment