Drought Monitoring

Drought is monitored using different indices. To characterize drought spells in mainland Portugal, some simple indices such as the percentage of normal have been used. To monitor drought situations in Portugal, the Palmer drought severity index (PDSI) (Palmer, 1965) is now being used. The next step will be the implementation of the standardized precipitation index (McKee et al., 1995) and its comparison with PDSI.

Precipitation Deviation and Deciles

The percentage of the normal precipitation is one of the simplest methods of quantifying rainfall for a given location, and it is very effective when used for a single region or a single season. However this criterion can be misleading, because it is not standardized for varying environments.

The decile was developed by Gibbs and Maher (1967). The distribution of occurrences over a long-term precipitation record is divided into tenths of the distribution. Each of these categories is called a decile. The first decile is the rainfall amount not exceeded by the lowest 10% of the precipitation, and so on, until the rainfall amount identified by the tenth decile is the

1970 1975 1980 1985 1990 1995

Year

Figure 14.2 Wheat yield variation in mainland Portugal (from GPPAA, 1999).

1970 1975 1980 1985 1990 1995

Year

Figure 14.2 Wheat yield variation in mainland Portugal (from GPPAA, 1999).

largest precipitation amount within the long-term record. By definition, the fifth decile is the median, and it is the precipitation amount not exceeded by 50% of the total precipitation over the period of record (Hayes, 2002).

Palmer Drought Severity Index

The Palmer drought severity index (PDSI) was developed by Palmer (1965). This index is based on the supply-and-demand concept of the water-balancing equation. The objective of the PDSI was to quantify moisture conditions that were standardized, so that both spatial and temporal comparisons in drought conditions could be made (Palmer, 1965). The PDSI is calculated on the basis of precipitation and temperature data, as well as on the locally available water content (AWC) of the soil. The PDSI varies roughly between -6.0 and +6.0. A PDSI value between -2.0 and -3.0 refers to a moderate drought; between -3.0 and -4.0 refers to a severe drought; and below -4.0 refers to an extreme drought.

The PDSI is applied within the United States but has little acceptance elsewhere (Kogan, 1995). One explanation for this is provided by Smith et al. (1993), who suggested that it did not perform well in regions with extreme variability in rainfall or runoff. To overcome this situation, the PDSI was adapted and calibrated to the specific climatic conditions of mainland Portugal. The calculations pertaining to runoff, the procedure for water balancing, and the identification of the beginning and the end of a drought or a wet spells were modified. The climatic coefficient (K) was prepared using drought periods from the time series and from different regions of Portugal (Pires, 2003).

To begin with, the PDSI was studied to provide spatial and temporal representations of historical droughts. The PDSI was calculated on a monthly basis for three southern stations with a long time series (Lisbon, Evora, and Beja) for the period 1901-2000. On evaluating the trends of moderate, severe, and extreme droughts, the results indicated that, generally, some categories of droughts occurred more frequently just near the end of the time series, especially in the last 20 years. In Beja, 50% of the extreme droughts occurred after 1980. In fact, during the 1901-2000 period, this station recorded the greatest frequency of extreme droughts (4.0%), with Lisbon and Evora showing similar frequencies (2.0%; Pires, 2003).

The PDSI time series for the three meteorological stations is shown in figure 14.3. They reveal a high-frequency oscillation of the PDSI between negative and positive values, superimposed by periods of consecutive months with negative or positive values, which are almost coincident for the three stations presented. With respect to the change in variability of the PDSI, the negative values seem to dominate the last 20 years of the 20th century, especially in the south inland stations of Evora and Beja. The 1980s begin with a sudden and large decrease in the PDSI, maintaining a trend for negative values through several years. According to figure 14.3, the values of the PDSI in the cooling period 1946-75 are less negative than in the warming period 1976-2000, suggesting an increased frequency of droughts in the south of Portugal (Pires, 2003).

A Geographical Information System (GIS) is used to map the PDSI and monitor the historical evolution of the index in the southern regions of Portugal that are the most affected by droughts (Pires, 2003). The percentage of time with mild, moderate, and severe drought was mapped for the southern region from 1961 to 2000 (Pires, 2003). During this period of 40 years, the percentage of time with mild drought (PDSI values below -1.0) in a great part of the region occurs between 30 and 40% of time. The percentage of time in moderate drought is lower; however, a large area with 15-20% of time in moderate drought is still observed. The percentage of time with severe drought is low, although in a small region it reaches nearly 10%.

A statistical analysis of long climatological series of the PDSI was made for the southern region of mainland Portugal. The PDSI average was calculated for the last four decades since 1961. An increase in severity is observed in most of the months in the first three decades, while in the last decade the drought intensity, although it has increased in some regions, does not increase significantly as compared with the previous one. In the period from February to April the increase is more significant, changing from normal conditions (PDSI with small positive values) to conditions of mild and moderate drought, especially in the months of February and March (figure 14.4). As these are decade-average values, this change is very significant. No relevant change is noticed during the summer period (June to August) over the four decades because this period is normally very dry (Pires, 2003).

As a means of validation of the PDSI as a tool to monitor agricultural droughts in Portugal, three situations, which occurred in the 1990s (199192, 1994-95, and 1997), were analyzed. The example of the meteorological station of Beja is presented in table 14.1 (Pires, 2003). In these three periods of drought, the PDSI values show the occurrence of severe and

Figure 14.3 Time series of Palmer drought severity index for Lisbon, Evora, and Beja stations in southern Portugal (from Pires, 2003).

extreme droughts, as expected due to the consequences on agricultural production.

The next step to improve agricultural drought monitoring is to implement the crop moisture index (CMI), an index also developed by Palmer (1968) and based on PDSI procedures. Whereas the PDSI monitors long-term meteorological wet and dry spells, the CMI was designed to evaluate short-term moisture, as it is based on the mean temperature and total precipitation for each week and also for the CMI value from the previous week. The CMI responds rapidly to changing conditions and, like PDSI, it is weighted by location and time, allowing the preparation of maps covering moisture conditions at different locations (Hayes, 2002).

Figure 14.4 Average March Palmer drought severity index for (a) 1961-70, (b) 1971-80, (c) 1981-90, and (d) 1991-2000 (from Pires, 2003).

Drought Monitoring Using Satellite Data

The Portuguese Meteorological Institute (IM) routinely calculates normalized difference vegetation indices (NDVI) based on the National Oceanic and Atmospheric Administration's Advanced Very High Resolution Radiometer (AVHRR) satellite data (Tucker and Sellers, 1986; Gomes et al., 1989; Duchemin and Maisongrande, 2002; chapters 5 and 6) to quantify vegetation stress and monitor droughts. Since 1987 drought effects have been monitored on the main region of cereal winter crops in Alentejo. Usually, only one parameter (NDVI) is used to monitor land, which fails to fully characterize the surface. Thermal infrared channels can provide

Table 14.1 Palmer drought severity index values for Beja, Portugal, during three drought spells (Pires, 2003)

Year

Month

PDSI

1991

Sept.

-2.17

1991

Oct.

-1.98

1991

Nov.

-2.64

1991

Dec.

-3.56

1992

Jan.

-4.11

1992

Feb.

-4.28

1992

March

-4.80

1992

Apr.

-4.13

1992

May

-3.62

1994

Sept.

-2.20

1994

Oct.

-2.50

1994

Nov.

-3.22

1994

Dec.

-4.31

1995

Jan.

-4.51

1995

Feb.

-4.02

1995

March

-4.37

1995

Apr.

-4.29

1995

May

-4.65

1995

June

-4.91

1995

July

-4.76

1995

Aug.

-4.15

1995

Sept.

-3.40

1995

Oct.

-4.59

1997

Feb.

-1.67

1997

March

-3.97

1997

Apr.

-3.75

additional information on surface conditions. These channels are used to estimate mass and energy fluxes between the surface and the atmosphere (Cooper et al., 1989). The difference between surface temperature (Ts) and atmospheric temperature (Ta) is a decreasing function of the plant's real evapotranspiration (Jackson et al., 1977).

Taking into account this relationship, another index, the forest fire index (FFRI), is computed at the IM on an operational basis:

where k is a constant, Rg is the global radiation, and NDVImax is 10-day maximum NDVI value. The above index is calculated on a daily basis; Ts is calculated using AVHRR data, and Ta and Rg are collected from the IM's meteorological stations. The surface temperature, Ts, is calculated using a split window method (Melia et al., 1991) as expressed by equation 14.2:

Ts = T4 + [1.31 + 0.27(T4 - T5)](T4 - Ts) + 1.16 [14.2] where T4 and T5 are the brightness temperatures of respectively channel

4 and channel 5 of the AVHRR. The temporal variation in the FFRI is analyzed for different test sites and compared with the temporal variation at each respective site for a representative year to detect drought conditions.

Satellite Application Facilities (SAFs) are specialized centers within the European Organization for Exploitation of Meteorological Satellites Applications Ground Segment and are hosted by the European National Meteorological Services in member states. The Portuguese Institute of Meteorology (www.meteo.pt) is responsible for the development of the SAF for land surface analysis in Portugal (www.meteo.pt/landsaf). The main purpose of this SAF is to enhance the benefits of the EUMETSAT satellite systems, Spinning Enhanced Visible and InfraRed Imager/Meteosat Second Generation and EUMETSAT Polar System related to land, land-atmosphere interactions, and biophysical applications by developing the techniques, products, and algorithms for more effective use of the satellite data. The land SAF involves the near real-time generation, archiving, and distribution of a coherent set of products that characterize the land surface by surface temperature, albedo, evapotranspiration, snow/ice cover, soil moisture, and vegetation parameters that are especially relevant to drought management. Some of these products include the leaf area index (LAI), the fractional vegetation cover (FVC), and the fraction of absorbed photosynthetic active radiation (fAPAR) or the fraction of green vegetation (FGV). The quality of these products depends to a large extent on the sensor characteristics (spectral, radiometric, and geometric), cloud detection, atmospheric correction, and angular distribution of the observations.

The algorithms used to estimate biophysical parameters in the land SAF depend on empirical relationships for vegetation indices (Asrar et al., 1985), inversion models (Roujean et al., 1992; Knyazikhin et al., 1998; Bicheron and Leroy, 1999), and physical and empirical models (Qin and Goel, 1995; Weiss and Baret, 1999; Lacaze and Roujean, 2001). Two complementary inversion approaches for the retrieval of FVC and LAI include kernel-driven reflectance models (Roujean et al., 1992) that obtain nadir-zenith reflectance as an input, before applying a more robust technique—namely, variable multiple endmember spectral mixture analysis (VMESMA; Garcia-Haro et al., 2002a). With VMESMA it will be possible to estimate the subpixel abundance of vegetation, soils, and other spectrally distinct materials that fundamentally contribute to the spectral signal of the mixed pixels. Although the primary output is FVC, some empirical relationships can be then used to derive LAI (Lacaze and Rou-jean, 2001). The application of a directional strategy relates the Bidirectional Reflectance Distribution Function of the surface with the directional signatures of vegetation and soil and meaningful biophysical parameters (Garcia-Haro et al., 2002b). Reflectance of an individual pixel is assumed to consist of an area-weighted linear combination of the soil and vegetation radiances. Canopy geometrical effects are considered in the first-order scattering. The model makes a simple treatment of the multiple scattering effects using an approximate analytical solution to the radiative transfer equation.

The volume scattering formulation is similar to the G-function and Hot Spot model. However, the probability between the crown and gap is formulated in terms of the FVC and a geometric variable (n) associated with the shape of plants. The canopy model can be coupled with a leaf-level radiative transfer model, which provides a stronger connection between leaf optical properties and biochemical and structural parameters. The model can be inverted to estimate both FVC and LAI.

Early Warning Systems

In Portugal an early warning system is being developed involving data acquisition, analysis leading to drought monitoring, and prediction using a software application (Pires, 2003). Such a system will help planners devise the best strategy for drought management. In the near future this product will be available on the Web for direct access by users.

A project is also being developed to identify, characterize, and predict local and regional droughts (INTERREG IIIB, 2002) using general circulation models, soil water balancing, and drought indices (e.g., PDSI, SPI) and probabilistic and stochastic modeling of regional droughts.

Crop Growth Models

In Portugal, crop growth simulation models have been used to assess the impacts of climate variability on agriculture. The models used are CERES models included in the Decision Support System for Agrotechnology Transfer series (Pinto and Brandao, 2002; Jones et al., 2003) and, more recently, the World Food Studies model (Supit and van der Goot, 2002). These models will be used for yield estimation, together with seasonal and annual forecasting, to complete the early warning system.

0 0

Post a comment