Regionalization of characterization factor in brazil: fresh water eutrophication category

Brazil is privileged to have the most important natural resource in its territory, the water. But currently, the increased concentration of phosphorus (P) affects the water quality. It has two main routes to get into aquatic environmental: through the dump of untreated sewage and fertilizers runoff. The P excess may promote eutrophication, a process characterized by microalgae uncontrolled growth, affecting several parameters of freshwater. Due to the great differences at the Brazilian regions, the current Life Cycle Impact Assessment methodologies are not capable to evaluate properly the eutrophication impact in Brazil. The most viable method to obtain a more suitable model is regionalizing it by estimating the characterization factor (CF). Therefore this is the first study that presents a regionalized model to estimate Brazilian CF for freshwater eutrophication. The regionalization was based on the models proposed by Helmes et al. (2012) and Azevedo et al. (2013), which are considered the most complete and suitable models. Due to the lack of data it was possible to calculate the CF for Alto Iguaçu micro watershed and four more subwatersheds: Paraíba do Sul, Parnaíba, Litorânea do Ceará and Litorânea Pernambuco Alagoas. The processes assessment of advection, retention and water use provides valuable information of each region and also results in more realistic Fate Facto (FF), since Brazilian sanitation is completely uneven, and the sewage treatment must be modelled to not overestimated or sub estimated the FF. At Alto Iguaçu is the advection rate is the most relevant and its CF is 7.43 103. m3 .KgP-1 day. For that reason, the same amount of emitted phosphorus promotes a bigger eutrophication potential at Paraíba do Sul and Parnaíba than other basins. Phosphorus income rates estimate is possible to know the origin of its most significant input. Based on this information, financial resources can be better used. This was the first attempt to develop a Brazilian CF and some improvements need to be done. Firstly, new studies ought to concentrate to promote good quality of data, because the unavailability of data was one of the greatest difficulties of this study. Then the regionalized model should be improved modeling treatment of industrial sewage and, finally, CF needs to be calculated for all subwatershed.


INTRODUCTION
The increase of phosphorus (P) emission compounds in water bodies is responsible for poor water quality. Although P is a nutrient, in excess it may cause eutrophication (Esteves 1998).
The domestic and industrial effluents and agricultural activities are the main sources of anthropogenic phosphorus. A big amount of sewage is released to water bodies without any treatment and part of the fertilizers percolate through the soil reaching aquifers, and another portion is carried out by irrigation and rain water to rivers and lakes (Esteves 1998).
To evaluate freshwater eutrophication's potential impact supports to prevent the environmental impact management, giving information for decision makers (private or public).
The eutrophication potential of products can already be obtained using Life Cycle Impact Assessment (LCIA) methods, however it is important to regionalize the characterization factors (CF), as the same amount of emission does not affect the regions equally due to the fact that, among others, it depends on the phosphorus transport and some water body characteristics (Tundisi, Tundisi, Siagis Galli 2006). Therefore, the goal of the current study is to calculate the eutrophication freshwater CF for Brazilian subwatersheds with different sanitations and climate conditions.

METHOD
Equation (1) shows the generic model to obtain CF.
Where , , : substance characterization factor, at the compartment m, that is transferred to the receiving environment; : Fate factor; and : Effect factor The FF characterizes phosphorus persistence in the environment and the EF shows the connection between the ecological damage and the mass change of phosphorus in the freshwater (Rosenbaum, Margni, Jolliet 2007

FF regionalization
Helmes et al. (2012) proposed the calculation of three processes to estimate the FF, advection, retention and water use. In the appendix I the variables are explained in more detail.
The Brazilian data was available at subwatershed level. Moreover, the sewage treatment was included, which was not modeled in Helmes et al. (2012), due to the massive difference at Brazilian sanitation. This model adaptation is based on the model proposed by Gallego et al. (2010).
Income rate of P by advection from upstream subwatershed ( , ) is calculated using equation 2. To estimate it some assumptions were made: a) Only stream water contributes to the advection process and, in a watershed, secondary rivers flow to the main stem, so the water flowing of the river base level represents the advection from upstream subwatersheds and;

b)
, was calculated adding the volume of rivers, lakes and reservoirs of upstream subwatershed.
Originally, the model estimated P transference from upstream grid ( ) to downstream grid ( ) through the water flow using equation 2.
(2) , = , Where: : The flow rate of the main stem base level of upstream subwatershed and; , : was calculated adding the volume of the mean rivers, lakes and reservoirs of subwatershed which is connected with the analyzed subwatershed Outcome rate of P by retention ( , ) was estimated using equation 3. The assumptions were: a) There was no information of affluent rivers, therefore the river volume at analyzed subwatershed was considered equal to the main river volume; b) Removal rate of phosphorus depended of the mean river water flow and; c) Phosphorus uptake velocity was 3.80 10 -5 at all subwatersheds (Alexander, Smith, Schwarz 2004).  Where: , : rate of P by water used for domestic purposes by subwatershed.
, : Fraction of water returned to the subwatershed after being used by the domestic sector at subwatershed level. , : Share of the total water use that is used for domestic purposes at subwatershed. , : Share of the total water use that is used for irrigation at subwatershed.  = (1 − ). Where: : Percentage of treated sewage at analyzed subwatershed : Percentage of phosphorus removed at effluent treatment : Fraction of P at the not treated sewage The regionalized equations of income and outcome rates were used to estimate the phosphorus persistence at freshwater ( ) and transported phosphorus ( , ). The persistence was obtained by the inverse of the sum of outcomes rates, as in equation 8. The transported phosphorus was calculated according to equation 9.    The Average Effect Factor model (AEF) was recently proposed as an alternative to MEF, because it reflects the average distance between the current state and the preferred state of the environment. AEF is calculated using equation 14, which is detailed at the Table 3.

FF results
The figure 1 presents FF results, figure 2 income rates and the figure 3 the outcome rates. was relevant, in spite of the fact that this basin is in an urban area, which imply a lower amount of water used for irrigation when compared to the amount used to domestic supply.
The advection was also the most important variable. , was significant, because water flow from Iguaçu River to Alto Iguaçu was low (around 86 m 3 /s) and this increased the P retention time increasing its assimilation and precipitation. Therefore, investing at sewage treatment at Ribeira do Iguapé subwatershed seems to be the most efficient way to improve the eutrophication condition at Alto Iguaçu.
The results of FF for Parnaíba do Sul using the original model was also between 30-300 days, and using the regionalized data and the model adaptation, it was 80.05 days.
The original FF results of Parnaíba, Litorânea do Ceará and Litorânea Pernambuco and Alagoas ranged from 3 to 10 days, whereas using the regionalized model the FF were respectively 31.78, 5.0, 13.37 days, none of them within the previous range. Paraiba River is the main river and it is connected to Bacia Grande and Bacia Doce watersheds.
Although Paraiba do Sul has 37% of its sewage treated and Parnaiba only 21%, in the former, the situation is worse than the latter, because it is located at a populous region, so the , , and , , are extremely high. Besides, Paraíba's River water flow is high, approximately 1120 m 3 /s, consequently, , is under most and , is the highest of all watersheds studied. The , from Bacia Grande and Bacia Doce (1.13 10 -3 and 4.51 10 -3 days -1 respectively) is quite relevant, therefore, to improve the eutrophication at Paraiba do Sul it is not enough to expand the sewage treatment, it is also necessary to invest at the wastewater treatment of Bacia Grande and Bacia Doce. Industrial activity is very intense in this region, but the regionalized model does not model it, therefore the FF should be even worse than the estimated.
Parnaiba, Litorânea Pernambuco Alagoas and Litorânea do Ceará are located at northeast of Brazil and in a megathermal rainy zone, which has an average temperature of 16 C to 32° C.
They are not connected with any subwatersheds. Litorânea do Ceará is divided in five regions: Acaraú, Coreaú, Curú, Litoral and Metropolitana.
For those three watersheds, , is zero because the spring rivers are located inside the basin, therefore there is no income rate from upstream watershed. At Parnaiba, Litorânea Pernambuco Alagoas, the dominant input rate was , , and at Litorânea do Ceará it was , , . At the coast, very low results of , , can be explained due to the fact that a large amount of sewage is dumped at the sea and the current regionalized model is for freshwater eutrophication.
Nevertheless, this information is relevant for marine eutrophication.
These three basins are located at a region with optimal temperature conditions for microalgae development so , is the dominant output process. As Litorânea Pernambuco Alagoas and Litorânea do Ceará suffer with water shortage, this resulted in low rates of , .

EF results
At Alto Iguaçu micro watershed TP concentration data of affluent rivers is not available, so the average of the four values collected at Iguaçu River was used. The coefficients α and β were used for water stream, because the rivers' water volume is much bigger than in lakes and reservoirs.
The Table 4 presents the data for Alto Iguaçu micro watershed, and Table 5   The figure 4 shows the results of EF obtained using the three models. LEF was used to estimated the effect if the concentration of P is unknown. As LEF does not depend of TP concentration, and since there is not Brazilian data to estimate the water stream coefficient, the same α w from literature was used for all subwatersheds. As a result, LEF causes the same value for subwatershed, making it impossible to evaluate the effect using it. For this reason, the AEF allows a better evaluation of the Brazilian subwatersheds than LEF and MEF. According to AEF model, the same amount of dropped P affects more species richness at Alto Iguaçu, Parnaíba and Paraíba do Sul, hence preventive measures should be a priority at these regions.

Characterization Factor results
The results of the Alto Iguaçu CF using three different EF models are shown at Table 6. = .
The calculation of the other basins is detailed at the Appendix I and CF calculated using AEF model of all subwatersheds are shown in figure 5. Because of this high eutrophication potential there is concentration of hypereutrophy water bodies at these watersheds.

CONCLUSIONS
The proposed goal was achieved through the LCIA method of freshwater eutrophication that was regionalized promoting the first CF to estimate the eutrophication impact at Brazilian subwatersheds.
This is the first study that presents a regionalized model to estimate Brazilian CF for freshwater eutrophication. The regionalization was based on the models proposed by Helmes et al. (2012) and Azevedo et al. (2013), which are considered the most complete and suitable models.
Due to the lack of data it was possible to calculate the CF for Alto Iguaçu micro watershed and four subwatersheds: Paraíba do Sul, Parnaíba, Litorânea do Ceará and Litorânea Pernambuco Alagoas. Among these watersheds, Alto Iguaçu has the highest CF.
The processes assessment of advection, retention and water use provides valuable information of each region and also results in more realistic FF, since Brazilian sanitation is completely uneven, and the sewage treatment must be modelled to not overestimated or sub estimated the FF.
It is not feasible to apply the original models in Brazil, because there is no data of geographical differentiation in 0.5°x 0.5°grids. Therefore, regionalizing the CF is the most reasonable way to promote higher quality information, which can be used to make strategic decisions in order to avoid eutrophication impact.
Finding the necessary data was the crucial part of the work because the lack of available data in Brazil only considers the main rivers with low quality and low representativeness. To improve the CF quality, is very important having data of affluent rivers, such as lakes and reservoirs.
Understanding the subwatershed, it becomes possible to direct financial resources to prioritize preventives actions, such as sanitation and environmental education programs at regions as  (HÍDRICO, 2016). It is located at a tropical altitude zone, and the average temperature is from 17°C to 22°C (MMA/ANA). Paranaiba River is the main stem and it is connected to Bacia Grande and Bacia Doce watersheds.
• FF calculation Applying the Subwatershed model FF takes 35.89 days. The process calculations are detailed below.
o Calculation of , The income rate by advection was estimated for Preto (Bacia Doce watershed) and Pomba River (Bacia Grande watershed) and the total advection rate is 5.63 10 -3 days -1 , which is the sum of the rate of both rivers.
The main stem water flow rate was collected just before it disembogues in the Paraiba do Sul watershed. The water volume at Paraiba do Sul was estimated by the same process as Alto Iguaçu's water volume. • EF calculation TP concentration is measured between Santa Branca and the Paraíba River base level. The coefficients α and β are used to analyze stream water, because the rivers watershed volume is more representative than the lake and reservoir volumes.
• Parnaiba subwatershed Parnaiba is located at the northeast of Brazil and in a megathermal rainy zone, which has an average temperature from 16º C to 32° C. Approximately 4.152.865 inhabitants live in this region. Parnaiba subwatershed is not connected to any subwatersheds, because all the rivers spring at the subwatershed. Therefore, the income rate by advection is zero.

FF calculation
• Litorânea do Ceará Litorânea do Ceará is also located at the northeast of Brazil and is not connected with any subwatersheds. It is divided in five regions: Acaraú, Coreaú, Curú, Litoral and Metropolitana.
• FF calculation Applying the Subwatershed model FF takes 5.00 days. The process calculations are detailed below.
o Calculation of , Litorânea do Ceará (LC) , is 3,28 10 -3 . Its water volume is calculated adding the volume average of each region, which is estimated multiplying the average percentage storage capacity by the maximum volume. Consequently, the water volume of LC subwatershed is 6,30 10 -2 km 3 . LC water flow rate is the sum of the average values of the five main stem flow rates (96,68m 3 /s). Therefore, the phosphorus removal rate is 0.012 day -1 (ALEXANDER et al., 2004). More details can be found at the table 4. The main stem volume is estimated using the shapefile of water availability. The lake and reservoir surface areas are calculated using GIS, as described before at Alto Iguaçu's topic. = .

• Litorânea Pernanbuco Alagoas
Litorânea Pernambuco Alagoas is located at the northeast of Brazil and it is not connected with any subwatersheds.
• FF calculation o Calculation of , Litorânea Pernambuco Alagoas (LPA) , is 1.44 10 -2 day -1 . Its water volume is estimated multiplying the average percentage storage capacity by the maximum volume. LC water flow rate is the sum of the average values of the five main stem flow rates (110.4 68m 3 /s). Therefore, the removal rate of phosphorus is 0.012 day -1 (ALEXANDER et al., 2004). More details can be found at the table 4. The main stem volume is estimated using the shapefile of water availability. The lake and reservoir surface areas are calculated by GIS, as described before at Alto Iguaçu's topic.