Читать книгу Congo Basin Hydrology, Climate, and Biogeochemistry - Группа авторов - Страница 43

The interannual variability of rainfall in equatorial Africa is relatively low, compared to other regions. Figure 3.17 presents the coefficient of variation (standard deviation divided by the mean), based on CHIRPS2 for the years 1981 to 2019. The coefficient of variation is lowest over the Congo Basin and surrounding regions, generally between 10 and 20%. While interannual variability tends to be low in humid regions (Nicholson, 2011), the low‐spatial correlation between stations over the Congo further reduces the interannual variability on a regional scale. Notably, the interannual variability of convective activity over the Congo is also relatively low (Hart et al., 2019).

As demonstrated in Nicholson et al. (2019), it is difficult to ascertain the rainfall trends over the central basin because of low station density and relatively poor performance of satellite products in the region. To produce the most reliable estimates of rainfall variability in the region, three data sets are examined: the NIC131 raw gauge data, NIC131‐gridded reconstructed data set, and CHIRPS2. Rainfall is represented by a standardized anomaly, the standard deviation divided by the long‐term mean. For the raw gauge data set, this permits the calculation of regional averages based on climatologically diverse stations and a temporally varying gauge network. Initially, year‐to‐year fluctuations for multi‐region sectors (Figure 3.18) are examined, then long‐term trends are quantified for individual regions and for the Congo Basin as a whole. The regional grouping into the six sectors is determined by linear correlation between adjacent regions (e.g., Nicholson et al., 2018b).

Figure 3.19 shows the year‐to‐year variations in rainfall in these sectors, based on both gauge data over the length of record and on CHIRPS2 between 1981 and 2019. Because the post‐1980 mean evident in the gauge data is often quite different from that of the earlier period, an adjustment is made to the CHIRPS2 anomaly, based on the ratio of the post‐1980 mean to the long‐term mean. This makes the two data sets more comparable. The three Congo sectors (Figure 3.18) are labeled Central Basin, North Basin, and South Basin. The remaining three sectors (Cameroon, Congo/Gabon, and East Africa) have much more extensive gauge coverage and satellite estimates are more reliable in these areas (Nicholson et al., 2019). Agreement with these regions increases the credibility of the estimates for the Congo Basin regions.

Figure 3.17 The coefficient of variation (standard deviation divided by the mean) of annual rainfall, based on CHIRPS2 and averaged over the time period 1981 to 2019.

Figure 3.18 Location of six multi‐region sectors for which interannual variability is assessed.

The three Congo Basin series are shown in Figure 3.19a. Note that the record for the central basin only extends through 2005. In that sector similar variations are apparent in annual rainfall and for the two seasons. A brief, relatively dry period commences around 1979 but an abrupt change to extremely wet years occurs around 1993. Because the wet period occurred so abruptly and was so extreme, the station network was examined to determine whether or not discontinuities in the station network were associated with these major shifts. That was not the case. Moreover, the CHIRPS2 data are consistent with this. Thus, the very wet period is likely real and not an artefact of the paucity of stations in the region.

The trends in the northern and southern portions of the basin are quite different from those in the central basin. In the south there are no obvious long‐term trends in either annual or ON rainfall. However, it is arguably drier since around 1970, with a preponderance of years with below normal rainfall since then. During MAM an abrupt shift to drier conditions is evident around 1970. In the north, negative anomalies likewise become more prevalent since around 1970 in the annual series and the season series. CHIRPS2 suggests some recovery in recent years in the MAM series.

Figure 3.19 (a) Interannual variability of rainfall for three of the sectors shown in Figure 3.18: North Congo, Central Basin, South Congo. Rainfall is expressed as a standardized departure (y‐axis), with a value of one equivalent to one standard deviation. Top: annual rainfall. Middle: March–April–May rainfall. Bottom: October–November rainfall. The diagrams on the left are based on gauge data and cover the period for which gauge data are available in the region. The diagrams on the right are based on CHIRPS2 and cover the period 1981 to 2019.Vertical lines in the gauge diagrams indicate the start of the CHIRPS2 analysis. (b) Interannual variability of rainfall for three of the sectors shown in Figure 3.18: Cameroon, Congo/Gabon, and East Africa. Rainfall is expressed as a standardized departure (y‐axis), with a value of one equivalent to one standard deviation. Top: annual rainfall. Middle: March–April–May rainfall. Bottom: October–November rainfall. The diagrams on the left are based on gauge data and cover the period for which gauge data are available in the region. The diagrams on the right are based on CHIRPS2 and cover the period 1981 to 2019. Vertical lines in the gauge diagrams indicate the start of the CHIRPS2 analysis.

Figure 3.19b shows the rainfall series for Cameroon, the Congo/Gabon area, and East Africa. The long‐term variability of annual and MAM rainfall in the Congo/Gabon region shows some similarity to that in the northern Congo. However, the shift to drier conditions occurred a few years earlier, in the early 1960s. Driest conditions occurred in a sequence of years around 1980. Rainfall then increased to a period of wet years centered around 2007. A later decline is apparent in the gauge data and in CHIRPS2. The trends in MAM are similar. During ON a shift to wetter conditions occurred in the early 1940s but since that time no obvious trends are evident. A similar increase in annual rainfall occurred in the early 1940s.

Rainfall in Cameroon shows the most pronounced long‐term trends. A shift to drier conditions around 1970 is apparent in annual rainfall and in both seasons. Some recovery is seen in recent years. This pattern is markedly similar to that over the Sahel (Nicholson et al., 2018c), where the seasonal cycle is similar to that over Cameroon.

Pronounced changes are also evident over East Africa, especially during the ON season. An abrupt shift to wetter conditions occurred around 1960 and has continued to present. Annual rainfall is determined primarily by conditions during this season, although it is the secondary rainy season over East Africa (Camberlin & Philippon, 2002, Hastenrath et al., 2011; Nicholson, 2017). Thus the shift to wetter conditions around 1960 is also evident in annual rainfall but to a lesser extent than in ON. MAM rainfall does not show any abrupt changes, but drier conditions have prevailed since around 1970.

The common denominator in rainfall variability in the six regions evaluated is a shift to drier conditions around 1970. It was apparent in March–April–May in all areas but the central Congo Basin, in annual rainfall in the northern and southern basin, and in October–November rainfall over the northern basin and in Cameroon. This coincides roughly the beginning of a major period of aridity in the Sahel that commenced in 1968 (Nicholson et al., 2018b). Notably, at about that same time a major change occurred in the relationship between Sahel rainfall and ENSO (e.g., Janicot et al. 1996, Losada et al. 2012, Camberlin et al. 2001).

In the central basin a sequence of dry years occurred commencing around 1970, but conditions shifted abruptly to high rainfall in the late 1970s to early 1990s. That wet episode is not apparent in any of the other regions. Notably, the orographic influence in that region is quite different than in surrounding areas (see Figure 3.3), providing a physical distinction that could play a role in interannual variability.

Zhou et al. (2014) noted a downward trend in rainfall over the Congo during the AMJ season over the period 1985 to 2012. The trend that paralleled a widespread decrease in productivity in the Congo rain forest. The magnitude of the rainfall trend is shown in Figure 3.20. The net trend is downward throughout except for a small region in the central basin. It is significant at the 10% level in 6 of the 12 regions shown. Fortunately, in central and northern areas of the Congo Basin, there is evidence of a return to better conditions of rainfall in recent years.

Figure 3.20 AMJ rainfall trends 1985 to 2012 for the individual regions in and around the Congo Basin (from Nicholson et al., 2018a; © American Meteorological Society. Used with permission). An asterisk indicates regions with trends significant at or above the 10% level. Units are standard departures per year averaged over the analysis period.

Hua et al. (2016) presented evidence of a longer term trend towards drier conditions. They specifically evaluated AMJ rainfall over the period 1950 to 2014, using GPCC data. A downward trend was evident throughout the Congo Basin. This trend was even stronger in eastern equatorial Africa. The NIC131‐gridded data set suggests an even longer term trend and a very sharp discontinuity in the mid‐1960s. The trend was evaluated for a somewhat longer season (March to June) over the period 1921 to 2014, the period of record for the NIC131‐gridded data set (Figure 3.21). A decline is seen in all but 3 of the 35 2.5‐degree grid boxes covering the Congo Basin. It significant at the 10% level in half of those regions.

A complete explanation for the drier conditions is beyond the scope of this chapter. Detailed analyses can be found in Camberlin et al. (2001), Todd and Washington (2004), Balaset al. (2007), Dezfuli and Nicholson (2013), Nicholson and Dezfuli (2013), and Hua et al. (2016, 2018). Overall, the causes are complex, with changes in SSTs and large‐scale circulation patterns being implicated. However, the links vary greatly by region and season (Balas et al., 2007). The factor that encompasses the region as a whole appears to be the Walker Circulation, which is strongly impacted by SSTs and in particular by the SST contrast between the various ocean basins (e.g., Hua et al., 2016, 2018; Williams & Funk, 2011).

Figure 3.21 Spatial patterns of linear trend per decade (1985 to 2012) and regional mean anomalies in rainfall during March through June, as evidenced in the NIC131‐gridded data set (from Liming Zhou, personal communication). The regional time series represents the area enclosed in bold in the figure on the left.

Figure 3.22 Five‐year averages of (left) relative number of MCSs per year and (right) total volumetric rainfall (km² mm/h × 10⁴) from MCSs (from Jackson et al., 2009, based on TRMM data; © American Meteorological Society. Used with permission). All data are averaged for a 1° × 1 °grid box.