Coping Strategies with Agrometeorological Risks and Uncertainties for Drought Examples in Brasil

O. Brunini, Y. M. T. da Anuncia^ao, L. T.G. Fortes, P. L. Abramides, G. C. Blain, A. P. C. Brunini, J. P. de Carvalho

17.1

Introduction tte 1997-1998 El-Niño caused an extreme drought in the northeastern region with considerable losses for agriculture, livestock, water resources and society. Regionally, the impact of these anomalies can be striking. In the southeastern region, for example, in the State of Sâo Paulo in the El Niño period, the effects caused by this phenomenon were quite different with above average rainfall in months like May and June, ttis situation can be observed, as indicated by the rainfall anomalies represented by the monthly Standardized Precipitation Index (SPI) for the month of May in 1998 (Figure 17.1). tte occurrence of these anomalies lead the State Government to create a task force involving the various sectors of society, such as, research institutes, universities and the civil defense, to propose mitigation measures.

Map Quantil Frequency Drought

Fig. 17.1. Monthly precipitation anomalies as indicatedby the monthly SPI (SPI-1) for the month of May 1998 in the State of Sâo Paulo.

Fig. 17.1. Monthly precipitation anomalies as indicatedby the monthly SPI (SPI-1) for the month of May 1998 in the State of Sâo Paulo.

tte National Meteorological Institute (INMET) determines the occurrence of droughts by means of the SPI, and also in deciles and the monthly deviation in precipitation compared to the climatological standard from 1961 to 1990. Studies have shown that 18 to 20 years of drought occurs every 100 years, tte frequency of the drought occurrence in the Brazilian northeast is associated with the frequency of the El-Niño and of the Atlantic Ocean dipole; and the frequency of the drought occurrence in the southern region is associated with the frequency of the La-Niña, tte areas affected by the drought vary in intensity, extension and time duration.

When a drought situation is confirmed through precipitation anomaly indices, technical material is prepared containing the precipitation monitoring for the affected region with a climate prognosis for the following quarter and this material is forwarded to the federal authorities in order to support the Brazilian government emergency actions. In the northeastern region of Brazil, there are several institutions and technical and technological infrastructure to detect drought. A limiting factor to ease detection of drought and corresponding mitigating actions is the lack oftrainingand capacitytodefinethe applicablemethodologies.

In addition to the National Meteorological Institute, some States of the Federation developed specific studies for droughts to support not only agriculture, but also the civil defense activities and water resources planning and studies. An example is the State of Sâo Paulo, through its Integrated Agrometeorological Information Center (CIIAGRO), and the Drought and Hydrometeorological Adversities Mitigation and Monitoring Center (INFOSECA). In this aspect, the assessments of the drought conditions and prognosis are prepared and distributed to farmers, rural cooperatives and other sectors of society.

tte Ministry of Agriculture at Federal level and the Agricultural Secretariat of the Sâo Paulo State Government, at State level, apply the reports and bulletins of drought monitoring in the Agricultural Activity Assurance Program (PROAGRO) and as a subsidy to the Agricultural/Livestock Expansion and Agricultural Insurance (FEAP) for the federal, and for Sâo Paulo State government, respectively.

tte immediate results of these actions are a reduction in the request for coverage for climatic events and the reduction of risks in Meteorological Adversities upon agriculture, in addition to the monitoring of the insurance operations and the agrometeorological management of PRO AGRO and FEAP (sources: www.agri-cultura.gov.br; www.agricultura.sp.gov.br.)

Regionally, there are programs that involve research institutions and the community in order to minimize risks for agriculture during drought situations. In the northeastern region, the state governments have mechanisms of their own to aid the population, such as distribution of water and foodstuff. In the drought areas, communities are supported by the federal and state governments and NGOs that orient the population. In order to improve health and reduce infant mortality, efficient methods of collecting and storing water by means of rural cisterns, underwater reservoirs and desalinization units are being applied.

In the northeastern and southern regions, regional forums for the quarterly climate prognosis for the rainy seasons are held, ttere is no specific forecast for drought, but the climate prognosis indicates beforehand if there is a probability of precipitation remaining below or above the normal, tte State of Ceará, through its Secretariat of Rural Development and the Ceará Meteorological Foundation indi-

cates the beginning of the sowing time by means of the climate prognosis, and the drought probability studies using real-time monitoring of precipitation and soil moisture content.

In the southeastern region, the government of the State of Sao Paulo implemented the Drought and Hydrometeorological Adversities Mitigation and Monitoring Center (INFOSECA), which is subordinated to the Instituto Agronomico (Agronomy Institute), tte work performed by INFOSECA along with the activities carried out at CIIAGRO is pioneering in Brazil the agrometeorological monitoring of drought and its effect on agricultural activities (source: http://ciiagro.iac.sp.gov.br - www.infoseca.sp.gov.br).With its major territorial portion restricted to the equatorial humid or tropical areas, the effects of meteorological adversities on the Brazilian territories, and most notably drought, are very distinct. An assessment of the Humidity Index as proposed by ttornthwaite and Mather (1955) is presented on Figure 17.2, involving some states in the southern, northeastern and midwestern states.

In general, the macroclimatic characteristics indicate humid climate conditions for the states in the southeastern and midwestern regions. Nevertheless, even for humid regions, the climatic oscillations cause, in specific years, a drought condition that is highly unfavorable to crops, ttis statement is supported by the monthly variation of the SPI for the areas of Campinas and Ribeirao Preto in the State of Sao Paulo for the month of January (Figure 17.3). Even though the month of January normally presents high rainfall indices, on specific years a meteorological drought occurs, ttis phenomenon has an elevated consistency with values that are highly unfavorable and prejudicial to crops, tte same aspect can be observed by the monthly variation of the Palmer Drought Severity Index for the locations ofVotu-poranga and Assis during the month of October (Figure 17.4).

Figure 17.4 further indicates an incisive factor which is the higher incidence of dry periods in the month of October in the last 15 years, shifting the beginning of planting of the summer crops to early November, ttis relationship with the PDSI, as well as with the SPI oscillations support the importance of monitoring and prognosis of drought in Brazil from the meteorological, hydrological and agronomic standpoints, with greater focus on the socio-economic effects of this meteorological adversity.

tte methodologies and parameters used at federal level by the National Meteorological Institute (INMET) and at a state level by the Integrated Agrometeorological Information Center - CIIAGRO, and by the Drought and Hydrometeorological Adversities Mitigation and Monitoring Center (INFOSECA), of the Agricultural Secretariat of the State of Sao Paulo are described below.

Fig. 17.2. Macroclimatic characteristics of some states in the southern, southeastern, midwestern and northeastern states based on the climatic classification proposed by ^ornthwaite-Mather (1955).

Campinas

□ Ribeirao Preto

I960 1363 1966 1963 1972 1975 1978 1381 1384 1387 1330 1933 1996 1993 2002 2005

Campinas

□ Ribeirao Preto s

I960 1363 1966 1963 1972 1975 1978 1381 1384 1387 1330 1933 1996 1993 2002 2005

Year

Fig. 17.3. Seasonal variation of the Standardized Precipitation Index (SPI) on a monthly scale (SPI-1) for the month of lanuary in the regions of Campinas and Ribeirao Preto in the State of Sao Paulo.

1970 1973 1976 1979 1982 1985 1988 1991

994 1997 2880 2083 2006

Fig. 17.4. Seasonal variation of the Palmer Drought Severity Index (PDSI) for the month of October in the regions of Assis and Votuporanga in the State of Sao Paulo.

1970 1973 1976 1979 1982 1985 1988 1991

994 1997 2880 2083 2006

Fig. 17.4. Seasonal variation of the Palmer Drought Severity Index (PDSI) for the month of October in the regions of Assis and Votuporanga in the State of Sao Paulo.

17.2

Methodologies to Assess Precipitation Anomaly and Drought 17.2.1

Meteorological Indices

SPI Standardized Precipitation Index tte Standardized Precipitation Index (SPI), proposed by McKee et al. (1993), corresponds to the number of standard deviations that the observed accumulated precipitation deviates from the climatological average, for a determined period of time, tte State of Sao Paulo (Brunini et al. 2000, INFOSECA 2005), Pernambuco (Santos and Anjos 2001) as well as INMET have been monitoring droughts through the SPI, presenting results that enable the use of the information to anticipate and mitigate adverse effects.

It is common to see in literature an association between a range of values for the SPI and the qualitative assessment of precipitation observed during the corresponding period, tte most frequent association is suggested by IRI (2005), as per Table 17.1.

Calculation of the index begins with the adjustment of the gamma probability density function to the monthly rainfall series. After this phase, the accumulated probability of the occurrence for each monthly total observed is estimated, tte normal inverse function (Gaussian) is applied to this probability and the result is the SPI.

In this method, precipitation can be totalized in several scales (1 to 72 months). When the time scale used is small (1, 2 or 3 months, for example), the SPI moves frequently above or below zero, observing the meteorological drought regime. As the assessment scale increases (12 or 24 months, for example) the SPI responds slower to changes in precipitation observing the hydrological drought regime.

Table 17.1. Arbitrary correspondence between the SPI values and the climate categories (adapted by Mckee et al., 1993)

SPI Values

Categories

SPI >+2

Extremely Wet

+1.50 a +1.99

Very Wet

+1.00 a +1.49

Moderately Wet

-0.99 a +0.99

Near Normal

-1.00 a -1.49

Moderately Dry

-1.50a-1.99

Severely Dry

<-2.00

Extremely Dry

tte gamma probability density function (GPDF) assumes distinct forms, according to the variation of a. Values for this parameter inferior to 1 indicate a strong asymmetric distribution (exponential form) with g(x) tending to infinite when x tends to 0. In the case of a = 1 the function intercepts the vertical axis in P for x=0. tte increase in the magnitude of this parameter reduces the asymmetric degree (deviation from the mode) of the distribution (the probability density is displaced to the right). Values for a greater than 1 result in a GPDF with the maximum point (mode) in p*( a-1). An increase in the ^ parameter stretches the GPDF to the right, lowering its height and reducing the probability of the occurrence of the mode value. Similarly, as the density is compressed to the left (reduction of the P magnitude) and the height of the function becomes greater, the probability of the event increases.

ttus, the spatial variation of a and ^ in a state or country, indicate which are the regions with greatest degree of asymmetry in the temporal distribution of precipitation (rainfall irregularity). Considering the phenomenon of drought, anomalies in relation to environmental conditions of each area, these regions are at a greater risk ofbeing subjected to meteorological droughts.

Palmer DroughtSeveritylndexAdapted to the State ofSao Paulo - PdsiAdap tte most important step of the PDSI is the calculation of precipitation, "Cli-matologically Appropriate Existing Conditions" (P) which can be understood as the amount of monthly precipitation necessary for a given area to remain under normal climatic conditions, ttis parameter is calculated as described by Palmer (1965). For the calculation of the monthlywater anomaly (d), the precipitation observed in the month (Pi) is compared to P in the same period.

As Palmer (1965) developed a standardized index compared to different locations at any period of time, it needs to be standardized (weighted) on a regional basis (Karl 1986). ttus, Palmer (1965) developed the climatic characterization factor designated by the letter K.

Where,

T - the ratio between the demand and supply of water in a region, and D - the monthly average of the absolute values for d.

Table17.2. Arbitrary correspondence between the PDSIadap and drought categories

PDSI adap

Categories

> 3.00

Extremely Wet

2.00 a2.99

Severe Wet

1.00 a 1.99

Moderately Wet

0.51 a 0.99

Slightly Wet

0.50 a -0.50

Near Normal

-0.51a-0.99

Slightly Dry

-1.00 a -1.99

Moderately Dry

-2.00 a-2.99

Severely Dry

< - 3.00

Extremely Dry

According to Blain (2005) adaptation of the PDSI to the State of Sao Paulo, had its major focus on the K factor of climatic characterization, tte other elements of the original methodology, such as precipitation, "Climatologically Appropriate Existing Conditions" and the d index were calculated as described in the original paper by Palmer (1965). Drought categories, according to the PDSIadap are presented in Table 17.2. tte final expression for K in State of Sao Paulo is:

and the final equation adapted to the State of Sao Paulo is:

17.2.1.3 Decile Method tte method consists of, initially, the organization in ascending order and subsequent classification of the historic precipitation data accumulated during the period of interest (normally 1,3,6,12 or more months) in 10 intervals of equal frequency (10 percent probability of occurrence in each class), ttese intervals are denominated deciles and are normally numbered 1 to 10. N being the number of historic observations registered, the first decile will contain the nl smallest values for precipitation, where nl corresponds to the integer part of (N/10), the second decile will contain the following values (n2 & nl), where n2 = (N/20), and so on.

Subsequently, a category will correspond to each decile, in other words, a descriptive concept of the rainfall intensity, in which deciles may be grouped, this means more than one decile may be associated with the same category. If we asso-

Table 17.3. Alternative classifications used with the decile method

Proposed

Classification

Classification Currently Adopted by the Australien Office ofMeterology

Classification Adopted by INMET

Category

Category

Category

Index

1

Lowest on Record

Extremely Below Normal

-3

Much Below Normal

Very Much Below Average

2

Below Average

Below Normal

-2

3

Below Normal

Slightly Below Normal

-1

4

0

5

Near Normal

Average

Normal

1

6

2

7

Above Normal

3

8

Above Average

Slightly Above Normal

1

9

Much Above Normal

Above Normal

2

10

Very Much Above Average

Extremely Above Normal

3

Highest on Record

ciate a color coding to each category, for example, we can plot precipitation behavior maps verifying, for each point, a class corresponding to the rainfall value observed during the period of interest, painting the point on the map with the color associated with this category.

Originally, the proponents of this method suggested using a rainfall classification per deciles as defined in the first part ofTable 17.3. More recently, the Australian Bureau of Meteorology adopted the classification defined in the second part of Table 17.3. On the other hand, INMET adopted the convention defined in the second part ofTable 17.3. On the other hand, INMET adopted the convention defined on the third part ofTable 17.3 and further associating, at each concept, a numerical indexbetween -3 and +3.

QuantileMethod

In summary, the Quantile method, consists in the classification of the accumulated precipitation values during the period of interest (timescale), X, in five categories as defined below:

Table 17.4. Rainfall anomaly classification based on Quantile methodology

Preciptation level

Associated Probability

Categories

(Observed Precipitation)

Quantile 1

15%

Very Dry

Quantile 2

20%

Dry

Quantile 3

30%

Normal

Quantile 4

20%

Wet

Quantile 5

15%

Very Wet

• First Quantile, 0<X<Qb where Qi is such that the Probability (X <Qi) = 0.15

• Second Quantile, Qi <X< Q2, where Q2 is such that the Probability (X < Q2) = 0.35

• ttird Quantile, Q2 <X< Q3, where Q3 is such that the Probability (X < Q3) = 0.65

• Fourth Quantile, Q3 <X< Q4, where Q4 is such that the Probability (X < Q4) = 0.85

Similarly to the SPI, to determine the Qi, i=l,...5, values, a probability model is adjusted (normally a Gamma distribution) to the historic data observed. Xbeing the precipitation for the period and F(x) the Accumulated Density Function adjusted to the historic values for X, and F1 to the inverse F function, thus:

Qi = F1 (0.15), Q2 = F1 (0.35), Q3 = F1 (0.65) eQ4 = F' (0.85) (6)

Each of the five quantiles defined above is associated with a qualitative classification as indicated on Table 17.4. As with the previous methods, the period of interest is normally 1, 3, 6, 12 or more months.

Comparison between methods

With the exception of the Palmer Index, the intrinsic principle of the various methods discussed above is the same and their results will differ only in the distinct conventions adopted for the classification of precipitation in categories and by the treatment, parametric or not, applied to the historic data, ttis comparison is discussed by Fortes et al (2006), which presents the chart reproduced on Figure 17.5.

Agrometeorological Indices tte understanding of the effect of the meteorological variables and their effect on crops is vital to determine the indices that adequately reflect the climate-plant in-

Fig. 17.5. Numerical scale indicating the estimated probability through the historical values of precipitation, in order to verify if a specific recorded rainfall value is smaller, equal to or larger than the historical case.

teraction and crop yield and that can be used in a constant, dynamic and easily handled manner.

It is worth mentioning that the drought phenomenon can be assessed or monitored, with emphasis on the meteorological, hydrological, agronomic and social-economic aspects, however, from an agronomic standpoint, this monitoring and prognosis must be evaluated with tools that involve agronomy and agrometeoro-logical knowledge and that integrate them in the process for weather and climate forecast.

In this aspect, the Institute Agrondmico do Estado de Sao Paulo (Agronomy Institute) has been developing the pioneering work with the implementation of CI-IAGRO in 1988, and subsequently with the Drought and Hydrometeorological Adversities Mitigation and Monitoring Center (INFOSECA) in 2005. tte agro-meteorological indices used on a routine and continuous basis by CIIAGRO and INFOSECA are given below.

Actual Evapotranspiration Standardizedlndex(IPER)

Developed by Blain and Brunini (2006), and based on the SPI methodology, this index begins with the adjustment of the beta probability density function to the series of water balance in a ten days step. After this phase, the cumulative probability of a given estimated value for ETR is calculated, tte normal inverse function (Gaussian), with a zero average and unit variance is applied to the accumulated probability, tte result is the value of the new index, named "Actual Evapotranspiration Standardized Index" (IPER).

Considering that the beta distribution is defined in the interval [0 and 1] and that the de average temperatures in the State of Sâo Paulo do not allow decendial ETR values above 100mm, the following variable transformation was chosen:

Where,

ETR" actual évapotranspiration variable transformed so that 0 < ETR" < 1

G(ETR") is then transformed into a normal variable (final value for the IPER) through the equations developed byAbramowitz and Stegun (1965)

IPER = -11--C° + Clt + Cltl-1 paraO < H (x) < 0.5

And the values of the constants are defined as:

c0 = 2.515517; c2 = 0.802853; c2 = 0.010328; d1= 1.432788; d2 = 0.189269; d3 = 0.001308

IPER values close to or greater than 0 indicate that the accumulated ETR in a 10-day period is close to or greater than the climatologically expected value of this parameter in this period. Negative values for the index indicate that the actual évapotranspiration in a given 10-day period is below the expected level for this given period. Variation of this index is directly related to the number of standard deviations that a given value of the ETR is below the climatologically expected value for

Table17.5. Arbitrarycorre-spondence between the IPER values and the drought categories to address the crop water requirements

IPER Values

Categories

IPER > -0.5

Near Normal

-0.5 to -1.0

Moderately Dry

-1.1 to 1.99

Severely Dry

IPER< -2.0

Extremely Dry

a given period of time and location. Table 17.5 offers an arbitrary correspondence between the IPER value and the water conditions for the soil to address the needs of crops.

Crop Moisture Index(CMI)

Palmer (1968) developed the Crop Moisture Index (CMI) in order to perform weekly monitoring of crop conditions on a climatological scale, based on the average temperature and the total precipitation for the current week. According to this author, in simple terms, agricultural drought is an "évapotranspiration deficit". However, if the potential évapotranspiration is used as the maximum estimated moisture required by plants, sub-humid and semi-arid areas will have a évapotranspiration deficit during summertime. It is suggested that the actual évapotranspiration anomaly be used, in other words, an estimating the total, the actual évapotranspiration dropped in relation to the expected actual évapotranspiration for that week, tte CMI quickly responds to changes in climatic conditions for a region, being as such, appropriate for monitoring in small time scales (weeks or 10-day periods), tte index is not adequate for a larger time scales, such as months, quarters and others.

Crop Development as a Function ofSoilMoisture

Developed by Brunini (2005), this index seeks to relate the current soil moisture conditions and the development of the crop, aiming at quantifying and qualifying the water conditions in the soil which are favorable or unfavorable to plant development. In this case, the crop water development factor (CWDF) is the function between the ratio between the amount of water available in the soil (DAAS) and the maximum available water (DISPMAX).

CWDF = DAAS/DISPMAX

Where:

DAAS = DISPMAX indicates that CWDF = 1

Based on the agronomic, pedologic and agrometeorological aspects, the following relationship is established as seen on Table 17.6.

Considering that the soil moisture factor (CWDF) can be observed as a punctual value, as well as an average assessment, or an averaged value for soil moisture and characteristics of the crop development above or below the median value, a index was introduced to the crop development (CWDI), which considers the average

Table 17.6. Arbitrary relationship between the average soil ratio (CWDF) and the plant agrometeorological development conditions.

Average Soil Water Ratio

Plant Development Conditions

0.8 <= CWDF <= 1

Very Good

0.6 <= CWDF < 0.8

Favorable

0.4 <= CWDF < 0.6

Reasonable

0.3 <= CWDF < 0.4

Not Favorable

0.2 <= CWDF < 0.3

Harmfully

0.1 <= CWDF < 0.2

Severe

0.0 <= CWDF < 0.1

Critical

Table 17.7. Arbitrary relationship between the average soil moisture index and the conditions related to the plant water satisfaction index

Soil Moisture Index

Conditions Related to the Plant Water Satisfaction Index

1.0<= CWDI <= 1.5

Good

0.5 <= CWDI < 1.0

Favorable

0.0 <= CWDI < 0.5

Reasonable

-0.25 <= CWDI < 0.0

Not Favorable

-0.5 <= CWDI < -0.25

Harmfully

-0.75 <= CWDI < -0.5

Severe

CWDI < -0.75

Critical

characteristics of soil moisture and the crop, ttis relationship is indicated by the formula below and in Table 17.7.

ttese parameters and indices enable the monitoring of a culture, considering the periods of crop development, type of soil and culture, as well as date of sowing and the phenological phases.

Soil WaterSupply Conditions and WaterStress on a Crop

Many drought indices consider either rainfall only or, in some cases, the interaction with the water available in the soil as passive. As such, Brunini (2005) introduced the Crop Water Stress Index (CWS), which is based on the relationship between actual évapotranspiration, potential évapotranspiration and the water available in the soil. In addition that to the water availability follows the evolution of the root system.

In this case, values are estimated for general crops, in which the crop coefficient Kc is not employed. However, assessments are made involving specific groups of plants defined by Zl, Z2, Z3, Z4, which corresponds to the depth of the root system, as shown below:

Zi (25 cm) = potato, onion, garlic, rice, garden produce, beans Z2 (50 cm) = beans, peanuts, corn, sorghum Z3 (75 cm) = soybean, citrus, coffee, sugarcane, cotton Z4 (100 cm) = coffee, citrus, sugarcane ttis diversity in depth aims at differentiating crops, as well as to the different water retention capability of the soil, which can reflect in a larger or smaller exploration volume of the roots.

tte water stress concept, based on the ETR.ETP relationship was developed as a result of works from Brunini (1981,1987); Camargo and Hubbard (1994), Camargo and Hubbard (1999), in which the reduction in crop yield or plant development is based on the sum or product of the (ETR/ETP) in the period.

In this case, we analyzed only the response of a plant and the average (ETR/ ETP) values during this period indicating the relationship between these two parameters. We then have a combination of Z for each value of the DAAS, which is a double entry table. In other words:, for each value of water available in the soil and for each potential évapotranspiration at the same period there is a unique value of Z; having in mind that:

i) CWDF = (DAAS/ DISPMAX) and37. Mund- und Rachentherapeutika Judith Günther ii)Z = / [(CWDF) (ETR/ETP)]

Table 17.8 indicates the relationship between the Crop Water Stress Index (CWS) and the plant water supply, while Table 17.9 represents the relationship between the average value of stress for a given culture in a given time interval (ACWS), which is determined by the relationship:

n being the number of intervals used

17.3

ResultsandAnalysis 17.3.1

Meteorological Aspects of Drought Monitoring and Prediction

Considering the Brazilian territory, the INMET performed the follow-up of the drought in the 2005/2006 period, monitoring the monthly rainfall values and

Table 17.8. Arbitrary relationship between the Crop Water Stress Index (CWSI) and the plant agrometeorological development conditions

Crop Water Stress Index

Plant Development Conditions

0 = CWS < 0.1

Good

0.1 <= CWS <=0.2

Favorable

0.2 <= CWS < 0.4

Ordinary

0.4 <= CWS < 0.6

Reasonable

0.6 <= CWS < 0.8

Not Favorable

0.8 <= CWS <= 1.0

Critical

Table 17.9. Arbitrary relationship between the average Crop Water Stress Index (CWS) and the average development conditions of the plant during the period

Average Crop Water Stress Index

Average Development Conditions of the Plant

0.8 <= ACWS <= 1

Good

0.6 <= ACWS <= 0.8

Favorable

0.4 <= ACWS < 0.6

Ordinary

0.2 <= ACWS < 0.4

Reasonable

0.1 <= ACWS < 0.2

Not Favorable

ACWS = 0.1

Critical

Fig.17.6a. Monthly monitoring of rainfall anomaly in the Brazilian territory as indicated by the SPI for June/05 to August/05.

Fig.17.6a. Monthly monitoring of rainfall anomaly in the Brazilian territory as indicated by the SPI for June/05 to August/05.

0 0

Post a comment