Drought Monitoring Applications for Africa

Monitoring drought and crop production in Africa is not straightforward in part because of the large footprint of the NOAA-NDVI data (~4-8 km

H 04

H 04

120. 140. Julian date

57.5

32.5

480 560 640 720 800 880 960 Total above ground dry biomass (g/m2)

1040

32.5

220 250 280 310 340 C Grain yield (g/m2)

Figure 5.5 (A) Temporal evolution of NDVI (normalized difference vegetation index) of wheat during the growing season. (B) The relationship between NDVI and total above-ground dry matter accumulation. (C) The relationship between wheat yield and NDVI during the growing season (from Tucker et al., 1980, 1981).

Average Precipitation (mm/yr)

Figure 5.6 Relationship between July-October NDVI and rainfall over West Africa, 1980-97 (from Tucker and Nicholson, 1999).

spatial resolution), the complexity of crop mixtures with surrounding vegetation types, and the small farm sizes that dominate the African landscape. Many major organizations, however, operationally use NDVI data to monitor drought and famine in Africa, such as the Food and Agricultural Organization's Global Information and Early Warning System (GIEWS; http://www.fao.org/giews/), the U.S. Agency for International Development's Famine Early Warning System Network (FEWSNET; http://www. fews.net/; chapter 19), and the U.S. Department of Agriculture's Foreign Agricultural Service (USDA/FAS; http://www.fas.usda.gov/pecad/pecad. html). In all of the following examples for the Sahel, East Africa, and southern Africa, the monthly NDVI data for 20+ years were used (Tucker, 1996; Anyamba et al., 2001, 2002; Los et al., 2001).

The Sahel

Prolonged and severe droughts occurred in the Sahel region of Africa in the 1970s and 1980s (Tucker et al., 1983,1986; Hielkema et al., 1986; Justice et al., 1986). These droughts were monitored by comparing monthly NDVI values with the long-term means. Figure 5.7 shows reduction in NDVI values on account of reduced precipitation for Mali in the Sahelian region of Africa.

Figure 5.6 Relationship between July-October NDVI and rainfall over West Africa, 1980-97 (from Tucker and Nicholson, 1999).

Figure 5.7 (a) The NDVI monthly values (bars) and long-term (1982-2000) means (thick line) for Mali during 1983-85, and (b) the annual variation in millet production for Mali during 1981-2000.

During the period from 1983 to 1985, Mali, like other Sahelian countries, experienced one of the most severe droughts in recent history. This region has only one growing season from June to October, with a peak rainy season in July-August. The mean NDVI vegetation profile shows this growing season pattern, with a peak in NDVI during August-September every year. During the 1983-84 period, rainfall was below normal over most of the Sahel region, resulting in below-normal NDVI values. This was followed by a recovery in NDVI values to more normal levels in 1985. Drought evolved slowly in early 1983 before reaching a peak in 1984 when NDVI was 35% below normal.

Millet and sorghum are the staple crops in this semiarid region of Africa. Millet production for Mali, covering the period 1981-2000, is shown in figure 5.7b. The NDVI data showed that 1984 was the year of the most severe drought, resulting in the lowest production in 20 years, about 500,000 metric tons of millet. This was the lowest point over several years of progressively decreased production from 1980 to 1983, following years of persistent drought during the 1970s.

The spatial pattern of NDVI anomalies for 1984 (figure 5.8) shows a region-wide drought across the Sahel region, from Senegal to Sudan in the east, with NDVI anomalies ranging well below normal: -20% to as low as -80% in Chad. This drought led to large-scale famine, starvation, and loss of human life and livestock.

East Africa

Most of East Africa has a bimodal rainfall distribution and hence two agricultural growing seasons. The short growing season typically begins in late September and ends in November. The long growing season typically begins in February or March and continues through May or June. Figure 5.9 shows the evolution of monthly NDVI vis-à-vis the long-term monthly NDVI mean for Kenya (East Africa) for 1983-86 and 1996-99.

Like rainfall, NDVI values over East Africa exhibit a bimodal pattern, with maximum values in April and November. This pattern follows the evolution of both the long and the short rainy seasons that are governed by the Inter-Tropical Convergence Zone (ITCZ), following the twice-yearly solar passage over the equatorial zone. For 1983 and 1984, the monthly NDVI values for both short and long rainy seasons were lower than the

Sahel Growing Season NDVI Anomaly (1934)

Sahel Growing Season NDVI Anomaly (1934)

Figure 5.8 Spatial variation in NDVI anomaly across the Sahel region during the growing season of 1984.

Figure 5.8 Spatial variation in NDVI anomaly across the Sahel region during the growing season of 1984.

Figure 5.9 The NDVI monthly values (bars) and long-term (1982-2000) means (thick line) for Kenya during (a) 1983-85 and (b) 1996-98. (c) The annual variation in maize production for Kenya during 1981-2000.

long-term mean values, especially in 1984, and only slightly recovered to near their long-term average values in 1985. This deficit is a reflection of the large-scale rainfall deficit that affected the entire Sahelian region and extended into East Africa. The NDVI declined by 10-40% of normal NDVI during the period from the end of 1983 to the end of 1984 only slightly recovering to near normal in 1985. This indicates that the persistence of earlier drought conditions may have dampened the expected improvement in vegetation conditions with the return of near-normal rainfall conditions in 1985.

The NDVI patterns in 1996 to mid-1997 were associated with drought conditions that prevailed over most of East Africa during this period. NDVI declined by 20% for most of 1996 and early 1997 but was above normal for the period from late 1997 through 1998.

The drought of 1984 caused significant reduction in maize production (figure 5.9c). In 1984, only 1.4 million tons of maize was produced, compared to 3.0 million tons in 1994 and 2.4 million tons in 1998. Periods of reduced maize production (1984, 1992, 1996, 2000) corresponded to the occurrence of El Niño/Southern Oscillation (ENSO) cold events (which typically resulted in below-normal rainfall) and below-normal NDVI values (figures 5.9a,b).

The spatial anomaly patterns of drought conditions during March— May (critical growing period) 1984 are shown in figure 5.10. The drought pattern in 1984 was most pronounced in Kenya, especially the eastern part, and most of Ethiopia and Somalia. A large part of East Africa had NDVI values that were 10-60% below normal. Monthly NDVI in 1992 was well below normal (figure 5.11a). Figure 5.11b shows the interannual variability in maize production for Zimbabwe from 1981 to 2000.

Southern Africa

In southern Africa, the worst drought of the century occurred during 19911992 and affected nearly 100 million people. The most affected countries were Zimbabwe and Botswana.

The years 1983, 1987, 1992, 1995, and 1998 showed large deficits in production, with the lowest production in 1992 (figure 5.11b). All of these years represented periods of ENSO warm events, which were associated with below-normal rainfall and the prevalence of drought conditions over most of the region (Cane et al., 1994; Phillips et al., 1998). Drought years also led to inadequate forage and poor nutrition for animals, resulting in reduced carcass weight and increased cattle mortality (figure 5.12).

Unlike East Africa, southern Africa has only one growing season that normally begins in November and ends in April. Figure 5.11a shows the temporal evolution of monthly NDVI values and the respective long-term means for Zimbabwe for 1991-93. The time series data showed a unimodal pattern, with the NDVI reaching a maximum of about 0.4 units in February—March after the maximum rainfall in December—January. The dry season typically includes the period from late May to October,

Growing Season NDVI Anomaly (Mar-Ma/ 1934)

Growing Season NDVI Anomaly (Mar-Ma/ 1934)

Figure 5.10 Spatial variation in NDVI anomaly across East Africa during the critical growing season of 1984.

with associated low NDVI values (typically ~ 0.2). During 1991-93, NDVI values over the growing season were lower than their long-term means, indicating the prevalence of drought conditions. The NDVI during the 199293 period was below the long-term average. In 1992, the region received below-normal rainfall, causing region-wide drought conditions with NDVI anomalies reaching 80% below normal.

0 0

Post a comment