Drought Monitoring Methods

Drought Indices

The Korea Meteorological Administration (KMA) has 509 automated rain-gauge stations, which record precipitation on hourly or daily basis. Total precipitation from September to the next May and the mean and standard deviation of monthly precipitation from June to August, as well as monthly precipitation deviation from the long-term average are simple parameters used for drought monitoring. In addition, one-month forecasts of the temperature and precipitation are made and broadcasted every 10 days, and the outlook of temperature and precipitation for the next three and even six months is also announced.

The KMA determines the Palmer drought severity index (PDSI; Palmer, 1965; chapters 9 and 12). The Korean Institute of Construction Technology (KICT) also determines drought indices such as standard precipitation index (SPI; chapter 9), PDSI, and surface water supply index (SWSI). These indices are reported online (http://apply1.kma.go.kr/home/service/drought/ dro_home.htm; http://www.kict.re.kr/wed/researches/wres/DroughtWeb/).

A drought index, EDI, developed by Byun and Wilhite (1999), is the standardized value of the available water resources index (AWRI; Byun and Lee, 2002), which is the accumulated precipitation with a time dependent

1 101 201 301

Day Number

Figure 30.2 Daily precipitation and available water resources index (AWRI) averaged for 60 stations across Korea.

1 101 201 301

Day Number

Figure 30.2 Daily precipitation and available water resources index (AWRI) averaged for 60 stations across Korea.

weighting function. The difference between AWRI and precipitation that was averaged for 60 stations from 1974 through 1998 is shown in figure 30.2. It can be seen in figure 30.2 that the AWRI curve shows the concentration of water resources better than the precipitation curve. These indices and software for computing these indices are currently being verified and are available on a Web site (http://atmos.pknu.ac.kr/~mdr).

Figure 30.3 shows the spatial distributions of four drought indices for August 31, 2001, when a severe drought occurred. The EDI is different from the PDSI, SPI, and other indices (McKee et al., 1993; Hayes et al., 1999) because it can precisely capture the daily variation in drought intensity. The most important characteristic of the EDI is that the EDI shows both drought duration and severity. An EDI of less than -0.7, between -0.8 and-1.5, and between-1.6 and-2.5, means mild, moderate, and severe e0i pry e0i pry

http://atmos.pknu.ac.kr/ mdr) and in Byun and Wilhite (1999)."/>
Figure 30.3 Four indices that show the drought severity on August 31, 2001. Detailed information on these indices are available online (http://atmos.pknu.ac.kr/ mdr) and in Byun and Wilhite (1999).

drought, respectively. A severe drought with EDI of-2.1 and AWRI of 100 mm is seen in the central area. Figure 30.4 shows the variation in drought days (EDI less than -1.5) for a 25-year period (1974-98). It is evident from figure 30.4 that severe droughts occurred in 1978, 1982, 1988, and 1994.

Empirical Orthogonal Function

Figure 30.5 shows the empirical orthogonal function (EOF) analysis of the EDI. The first eigen mode explained only 55.9% of the total variance. Most parts of the peninsula show a positive value, with a center over southern inland area. Its time series shows four negative peaks near pentad (5-day period) number 13, 26, 44, and 61 that coincide with the four drought periods in early March, middle May, early August, and October, as shown by Byun and Han (1994) and Byun and Lee (2002). The second eigen mode includes 13.5% of the total variance shown by east-west contours. The time series shows drought concentration at mid-September. The third eigen mode shows 7.9% of the total variance with the meridional contours. In the eastern part of the peninsula, with negative values in contours, the drought occurs from March till July, as shown by positive values in time series.

Growing Degree Days

The Rural Development Administration (RDA) provides agricultural weather information to public on a Web site (http://weather.rda.go.kr/). The daily, monthly, and yearly data in time and space are available for temperature, humidity, radiation, wind speed, precipitation, and so on.

Figure 30.4 Yearly frequency, i.e. the number of drought days, of drought index (EDI < -1.5) based on data for 60 weather stations across Korea.

2nd Eigen Vector(13.5%) 1st Eigen Vector(55.9%) 3rd Eigen Vector(7.9E)

2nd Eigen Vector(13.5%) 1st Eigen Vector(55.9%) 3rd Eigen Vector(7.9E)

1 6 11 16 21 26 31 36 41 46 51 56 61 66 71 Pentad Number

Figure 30.5 Spatial structure of the first (55.9%) and second (13.5%) eigen vectors of 25-year mean drought index and the time series of the eigen vectors.

1 6 11 16 21 26 31 36 41 46 51 56 61 66 71 Pentad Number

Figure 30.5 Spatial structure of the first (55.9%) and second (13.5%) eigen vectors of 25-year mean drought index and the time series of the eigen vectors.

A drought index based on growing degree day (GDD) and soil moisture can be computed by either the Penman-Monteith equation or the Priestly-Taylor equation (Priestly and Taylor, 1972) for different soil types.

Use of Satellite Data

Combining the drought index with the vegetation index derived from satellite data can be a practical approach for agricultural drought monitoring. Normalized difference vegetation index (NDVI) composites with half-degree spatial resolution over the growing season in Korea (Szilagyi, 2002) were used to estimate areal evapotranspiration (AET) across Korea using the biosphere model (SiB2). Soil type, canopy structure, and phenology were taken into account to estimate the AET.

NDVI maps at a 1.1 km x 1.1 km grid interval were derived from the Advanced Very High Resolution Radiometer (AVHRR) onboard the U.S. National Oceanic and Atmospheric Administration's (NOAA) polar-

orbiting satellites to monitor vegetation conditions in Korea in 1999 (figure 30.6). These maps were used to classify the land cover of rice at a 1-km grid spacing for land surface parameterization of the biosphere model (Koo et al., 2001). Ha et al. (2001) analyzed the temporal variability in the NDVI, leaf area index (LAI), and surface temperature (Ts) estimated from AVHRR data collected from Korean Peninsula during 1981-1994. These products can be applied to estimate AET (Szilagyi, 2002):

where AETest is the estimated AET, ags is the standard deviation of monthly AET during the growing season (mm/day), and Egs is mean growing season AET. The estimated AET can be used for monitoring drought conditions.

Standardized Vegetation Index

A standardized vegetation index (SVI) based on calculation of a z score of NDVI distribution can also be produced for drought monitoring, as reported by Peters et al. (2002):

where Zijk is Z value for pixel i during week j for year k, Zijk ~ N(0,1), NDVIjk is weekly NDVI value for pixel i during week j for year k, wNDVIj is the mean NDVI for pixel i during week j over n years, and o ij is the standard deviation of pixel i during week j over n years. This per-pixel proba

Janurary February March April May June
Figure 30.6 Monthly maximum-value-composite normalized difference vegetation indices during 1999 for Korea. The values in the legend show NDVI values. The higher the NDVI values, the better the vegetation vigor.

bility, expressed as SVI, is an estimate of the probability of occurrence of the present vegetation condition (0 < SVI < 1).

SVI values were grouped by Peters et al. (2002) into five classes: very poor (0-0.05), poor (0.05-0.25), average (0.25-0.75), good (0.75-0.95), and very good (0.95-1), and were related to drought conditions. The SVI is a good indicator of vegetation responses to short-term weather conditions. High spatial resolution (1 km) and potential for near-real-time evaluation of actual vegetation conditions are the advantages of the SVI. But understanding the strengths and weakness of the SVI is important for determining when and how to use the index because climate conditions other than drought can also cause reduced vigor of vegetation.

In practice, the response of vegetation to precipitation can be examined using NDVI-precipitation relationships. Surface soil wetness derived from satellite images can be related to soil moisture for monitoring local drought conditions.

Was this article helpful?

+1 0
Renewable Energy Eco Friendly

Renewable Energy Eco Friendly

Renewable energy is energy that is generated from sunlight, rain, tides, geothermal heat and wind. These sources are naturally and constantly replenished, which is why they are deemed as renewable.

Get My Free Ebook

Post a comment