Total Nitrogen and Total Phosphorus Pollution Reshaped the Relationship Between Water Supply and Demand in the Huaihe River Watershed, China

As total nitrogen (TN) and total phosphorus (TP) pollution is the main source of water pollution in the Huaihe River watershed in China, it is important to understand how TN and TP pollution affect the relationship between water supply and demand. Quantifying their impacts and describing the spatiotemporal distribution of this relationships are necessary for furtherly deepening the theory of TN and TP pollution on water bodies, and this information is also particularly essential for managing water resources regionally. In this study, based on the potential water supply, the water demand and the Integrated Valuation of Ecosystem Services and Tradeoffs (In-VEST) water purification models, we estimated the TN and TP pollution from agricultural fertilizer, livestock and poultry breeding, and rural residents in the Huaihe River watershed and simulated TN and TP impacts on the relationship between water supply and demand. We found that if the impact of TN and TP pollution on water supply was not taken into account, on average, there was excess water supply in 79.20% of the watershed and excess demand in 20.80% of the rest during 1980–2018. Under the TN concentration limit, Grade-II (The water quality meets the secondary level of water body qualified in GB3838-2002, classified as Grade-II) water was the main water-supply type in 1980–2018, followed by Grade-I and Grade-III water. The total water shortage showed an inverted V-shaped trend: first increasing and then decreasing at the same period. The proportion of the water shortage of Grade-I water in the total water shortage was the largest, followed by Grade-II and Grade-III water. Areas with excess demand were located on the north bank of Wang-Beng, Yishuhe, and Huxi regions, although the water in these sub-watersheds met the water quality standards of Grade-I water. Under the TP concentration limit, Grade-II and Grade-I water were the main water-supply types. The overall water shortage trend first increased and then decreased, exhibiting an inverted V-shape from 1980 to 2018. The water shortages of Grade-I and Grade-II water showed similar inverted V-shape trend over time. Areas that met the water quality standard of Grade-I included the north banks of Wang-Beng and Huxi regions, where there was a surplus of demand. This paper suggests a way to analyze the interaction between water pollutants and the water supply-demand ratio as the example of TN and TP pollution at a watershed scale, which can broaden water pollution theory for relative water resources departments when water supply and demand will be evaluated.


Introduction
The benefits of water-related services are obtained from healthy watersheds, which provide the water supply necessary for sustainable economic growth and well-being. Research on hydrological services has focused on water supply and water purification, flow regulation, erosion and sediment control, and habitat preservation (Postel and Thompson, 2005;Brauman et al., 2007;Lü et al., 2015;Wang et al., 2019;Zhang and Wei, 2021), of which water supply and water-quality purification play important roles among the various ecosystem services (Postel and Richter, 2003;Vigerstol and Aukema, 2011;Qiu and Turner, 2013). Water supply is a provisioning service closely related to water demand in the domestic, industrial, agricultural, and environmental sectors (Haddeland et al., 2014), and is subject to the impacts of non-point water pollution such as total nitrogen (TN) and total phosphorus (TP), especially in the agricultural regions.
Global water use has increased sixfold over the past 100 yr and continues to grow at a rate of nearly 1% annually due to population growth and economic development (World Water Assessment Programme, 2020). In particular, the water used in agricultural activities to produce food for the increasing population accounts for nearly 86% of all water consumption across the world (D'Odorico et al., 2018). The long-term imbalance between water supply and demand has posed unprecedented pressure and challenges (Liu et al., 2017;Boretti and Rosa, 2019). For example, two-thirds of cities in China have experienced a water crisis, nearly 300 million rural residents lack safe drinking water, and over 40% of the country's river systems are polluted (Liu and Yang, 2012). On a global scale, Mekonnen and Hoekstra (2016) showed that nearly four billion people suffer severe water shortages, mainly in developing countries such as India and China.
Studies of the relationship between water supply and demand (also defined as water scarcity) have adopted different methods of assessing water scarcity, e.g., using a water scarcity index defined as the Water Demand-Supply Ratio (WDSR) and the green-blue water indicator (Rockströem et al., 2009). The calculation of water scarcity index involves the estimation of water supply and water demand. Water supply has been estimated us-ing land surface models with river runoff at discharge stations (Oki et al., 2001;Milly et al., 2005), a semi-distributed soil and water assessment tool (SWAT) and the distributed hydrology soil vegetation model (DHSVM) (Goeking and Tarboton, 2020), and the Integrated Valuation of Ecosystem Services and Tradeoffs (InVEST) water yield model with data on land use and land cover, precipitation, and potential evapotranspiration (Lang et al., 2017;Daneshi et al., 2021). Water demand has been simulated using the Water-global Assessment and Prognosis model (WaterGAP) (Schmied et al., 2021), econometric models (Blackhurst et al., 2010;Boero and Pasqualini, 2017), the Penman-Monteith formula (Wisser et al., 2008), the water footprint (Mekonnen and Hoekstra, 2011), and remote sensing techniques (Yousaf et al., 2021).
When water scarcity is examined in the context of water pollution, numerous studies have investigated the impact of pollution from point/non-point sources on water availability (Liu and Yang, 2012;Mateo-Sagasta et al., 2017). The results show that nearly 80% of domestic and industrial wastewater is discharged into water bodies globally, with total nitrogen (TN), total phosphorus (TP) and non-point pollution from agriculture overtaking point pollution as the major polluter (World Water Assessment Programme, 2017). Related researches showed that TN and TP pollutants were mainly from non-point resources such as agricultural fertilizer, livestock and poultry breeding, rural residents and atmospheric deposition (Xu et al., 2014;Wang et al., 2015). Therefore, more attention on the impacts of TN and TP pollution on water bodies has been focused mainly on detecting the sources of TN and TP pollution (Ma Enpu et al., 2021), spatio-temporal distribution of TN and TP (Zhong et al., 2021;Yu et al., 2022), and their impacts on water bodies at watershed scale (Gu et al., 2019). Popular methods involved in the above studies included the Eubolism model for estimating total TN and TP loads (Chen et al., 2010) and the SWAT and InVEST water purification methods for depicting spatial patterns of TN and TP loads (Xia et al., 2012;Cong et al., 2020). These studies have mainly focused on the relationship between TN/TP and water quality, but not directly aiming to water supply or water scarcity. Other reports have further conducted on the impacts of water-quality indicators on water scarcity such as the chemical oxygen de-mand (COD) and ammonium nitrogen  in China because these water-quality indicators were the key factors of water pollution control in China and monitoring data for COD and NH 3 -N could be obtained easily (Liu et al., 2017;Ma et al., 2020). However, few studies have quantified the impacts of TN and TP pollution on the spatio-temporal relationship between water supply and demand regionally and globally. The reason is that these two indicators are not among the key factors of water pollution control plans in the watersheds at all levels. So, measurements for TN and TP can not be easily available for quantifying their effects on water scarcity. Therefore, little information can be obtained on how much actual water is available and the spatiotemporal distribution of water scarcity, considering the impacts of their pollution on water bodies. This information is crucial for policymakers at all levels to allocate water resources scientifically.
In this study, we estimated TN and TP pollution from agricultural fertilizer, livestock and poultry breeding, and rural residents in the Huaihe River watershed (HRW) of China, and simulated the impacts of TN and TP pollution on the relationship between water supply and demand. The findings will provide scientific support to improve water use efficiency and water quality and allocate water resources more efficiently across regions over time. Theoretically, the findings of this study also provide important information on the impacts of TN and TP pollution on the water demand-supply relationship as well as on the TN and TP pollution loads for correcting the parameters in the InVEST water purification model.

Study area description
The HRW is located midway between the Yellow River and Yangtze River in northern China, ranging from 30°57′39″N-36°19′18″N and 111°53′27″E-121°22′55″E (Fig. 1), and it is an important grain-producing region. This region stretches approximately 700 km from east to west and 400 km from north to south and covers an area of approximately 26.88 × 10 4 km 2 , of which approximately 67% is cultivated land and 33% is mountainous and hilly. Wheat and corn are main crops in this watershed and more chemical fertilizer used in the agriculture sector is the main source of nitrogen and phosphorus pollution. Generally, the relief of this watershed is high in the west and low in the southeast and tilts from the northwest to southeast, with a maximum elevation of 2142 m in the west and a minimum elevation of -173 m in the southeast. Its climate ranges from humid monsoon in the southeast to semi-arid conditions in the northwest. Annual average precipitation is 920 mm and generally decreases from the southeast to northwest, mainly occurring from June to September, when 50%-80% of annual precipitation falls. Temperatures generally decrease from south to north with an annual mean temperature of 13.20-15. 70°C (Lu et al., 2022). The highest monthly temperature is approximately 27.00°C in July and the lowest is nearly 0 °C in January.
In 2018, the population in HRW was 219 141 000, with a population density of 824.30 people / km 2 , which was 5.60 times the national population density (National Bureau of Statistics of China, 2019). The GDP of the basin was 48 779.98 × 10 8 yuan (RMB), accounting for 6% of national GDP. The use of agricultural, industrial, and domestic water was 376.35 × 10 8 m 3 , 75.81 × 10 8 m 3 , and 76.00 × 10 8 m 3 , respectively, in 2018 (The Huaihe River Commission of the Ministry of Water Resources of China, 2019).

Data collection and methodology 2.2.1 Data collection
The meteorological data for 120 weather stations in this study were obtained from the National Meteorological Science Data Center (http://data.cma.cn) to interpolate the missing annual precipitation and reference crop evapotranspiration in 1980 and 2018 in the HRW. Data for the Advanced Spaceborne Thermal Emission and Reflection Radiometer Global Digital Elevation Model (ASTER GDEM) Version2 (1 km) and land use/cover (1 km) in this basin were obtained from the Geospatial Data Cloud (http://www.gscloud.cn) and the Resources and Environmental Science Data Center of the Chinese Academy of Sciences (http://www.resdc.cn), respectively. The evapotranspiration coefficient K c , rooting depth, the plant available water capacity, and an empirical constant essential for the InVEST water yield and water purification models were obtained from the World Soil Information Database (https://www.isric.org).
As for the TN & TP water-quality standards, the Na-tional Environmental Quality Standards for Surface Water of China (GB 3838-2002) defined the concentration limits of TN/TP for different types of surface water, but the Quality Standard for Ground Water (GB/T 14848-93) did not define them, and the standard has been invalidated. Additionally, the water demand data obtained at the county, municipal and provincial levels have not been further subdivided into ground and surface water demand, which makes it impossible to further subdivide the water supply into ground and surface water supply when studying the water demand-supply ratio in this paper. Therefore, this paper applied the TN & TP concentration limits in the GB 3838-2002 to all water bodies.

Potential water supply modeling
The potential water supply Y(x) can be simulated for each pixel x on the Budyko curve, annual average precipitation P(x), and actual evapotranspiration AET(x) using Eqs. (1) and (2) (Sharp et al., 2015): where PET(x) is potential evapotranspiration; ω is an empirical parameter and K c (l x ) is subject to the vegetative characteristics of the land use/land cover found on pixel x and ET 0 is the average annual reference evapotranspiration; AWC(x) is available soil water capacity; R n is the net radiation on the crop surface (MJ/(m 2 ·d)); G is the soil heat flux on the soil surface (MJ/(m 2 ·d); T is the atmospheric temperature at a height of 2 m (°C); μ 2 is the daily average wind speed at a height of 2 m (m/s); e s is the saturated water vapor pressure (kPa); e a is the actual water vapor pressure (kPa); Δ is the slope of the saturated water vapor pressure curve (kPa/°C); γ is the psychrometric constant (kPa/°C); and Z is an empirical constant, sometimes referred to as the 'seasonal factor', which captures the local precipitation pattern and additional hydro-geological characteristics. Theoretically, the value of seasonal factor (Z) equals to 20% of total days for daily precipitation (daily precip-itation > 1.00 mm) (Redhead et al., 2016). However, data for days with daily precipitation greater than 1.00 mm could not be obtained. The total days (daily precipitation > 0.01 mm) and corresponding Z in the HRW from 1980 to 2018 are shown in Fig. 2. It can be seen from Fig. 2 that total precipitation days (daily precipitation > 0.01 mm) ranged from 88 d to 119 d in the HRW during 1980-2018. The corresponding seasonal factor Z ranged from 17.5 to 23.7 with an average value of 20.7. However, it was difficult to obtain total days for daily precipitation (> 1.00 mm) in the HRW. According to the existing literature (Yuan et al., 2010;Wang and Wang, 2016) on recording total days for daily precipitation (> 1.00 mm) in Zhengzhou, Hefei, Henan, Jinan and Shandong provinces, total days for daily precipitation ranged from 45 to 111, and corresponding Z ranged from 9.0 to 22.0. Comparing the seasonal factors calculated according to the daily precipitation (0.01 mm) and the daily precipitation (1.00 mm), the maximum values of Z had little difference, but the minimum values had a large difference. In order to make the value of Z closer to 20% of total precipitation days (daily precipitation > 1.00 mm), this paper determined that the Z value is 9.0-23.7.

Simulation of TN and TP exports into river systems
TN and TP pollutants are discharged from non-point sources in the HRW, including from rural residents, livestock and poultry breeding, and agricultural fertilizer. In this study, following the discharge coefficient method used in previous studies (Song et al., 2018;Zhang, 2019), the loads of TN and TP pollution were simulated in the HRW during 1980-2018 as follows: where Load TN is the TN discharge; m is the number of administrative units; N (TN)i is the amount of chemical fertilizer, or total rural residents in the ith administrative unit; α (TN)i denotes the production coefficient of TN in the ith administrative unit; β (TN)i is the discharge coefficient of TN in the administrative unit, and t is time. TP load can be also simulated according to Eq. (3).
The InVEST water purification model can simulate how much TN and TP loads exported into river systems using the Eqs. (4-5) with more details in the paper by Sharp et al. (2015), and help to understand their impacts of on water quality and the relationship between water supply and demand.
where Load exp,i is the nutrient delivery on pixel i (kg/yr); Load surf,i and Load grou,i represent the surface and underground nutrient loads on pixel i (kg/yr), respectively; NDR surf,i and NDR grou,i represent the surface and underground nutrient delivery ratios on pixel i (kg/yr), respectively; Load modi (x, i) is the modified nutrient delivery on pixel i (kg/yr); RP i is the proxy for runoff on pixel i; RP aver,i is the average proxy for runoff in the studied watershed; prop grou is the proportion of underground nutrients to total nutrients.

Water demand modeling
The total water withdrawal is the sum of the volumes of agricultural, industrial, and domestic water. Total water demand was estimated at the municipal or county level and rasterized on various land use/land cover at a space of 1 km (Eq. 6): where Total(t) is the total quantity of water withdrawal (10 8 m 3 ); Area(t) is the effective irrigation area (in thousand hectares); α agr (t) represents the water withdrawal per thousand hectares (water coefficient); INDP ind (t) is GDP in the secondary industry (in 10 8 yuan); α equals to the proportion of industry in secondary industry; α ind (t) represents the water withdrawal per ten thousand yuan (water coefficient in 10 8 m 3 / 10 4 yuan). Further, Pop(t) is the total population (in 10 4 persons), and α dom (t) represents the water withdrawal per ten thousand persons (water coefficient in 10 8 m 3 / 10 4 persons). W i , a ij , S ij , and b i represent simulated water demand (m 3 ), the modified average water demand coefficients of various land use/land cover types (m 3 / km 2 ), area of land use/cover (km 2 ), and intercept of the equation, respectively. P ij is average water coefficient of the land type j in county i (m 3 / km 2 ). X i is the total population in county i (in 10 4 persons) and Total i is the total population in the HRW (in 10 4 persons). To ensure that when there is no land like reservoirs or lakes, there is no water demand, the intercept of the regression equation is set to 0. More information and methods for modeling water demand can be seen in the paper (Lu et al., 2022).

Water demand-supply ratio by considering TN and TP pollution
In this study, water demand-supply ratios (water demand-supply ratios, abbreviated to WDSR) can be defined as the ratio between water demand and supply (WDSR = water demand/water supply), and displayed at space of 1 km. In order to estimate the impacts of TN and TP pollution on WDSR, the TN and TP exports were divided by water supply to obtain their concentration at pixel of 1 km. WDSR was overlaid by TN and TP concentrations with ArcGIS spatial analysis to indicate the WDSR impacted by two scenarios of TN and TP concentrations.

Potential water supply in the Huaihe River watershed
In this study, we used time-series data for water resources published officially in the Water Resources Bulletin in the HRW as the reference values to calibrate the simulated potential water supply from InVEST water yield model (Compilation Group for Water Resources Bulletin in the Huaihe River watershed, 2022). The Morris sensitivity test (King and Perera, 2013) was performed to modify the parameters of the InVEST water yield model to simulate the potential water supply in the HRW in 1980-2018, as listed in Fig. 3. On the whole, the potential water supply showed a W-shaped trend during this period. From 1980 to 1995, the potential water supply decreased gradually. Then, from 1995 to 2005, it gradually increased. From 2005 to 2015, the water supply decreased again, before increasing in 2018. Fig. 4 shows the geographical distribution of potential water supply per 1 km during 1980-2018 in the HRW, which is similar to that of precipitation in China with higher precipitation in the southeast and lower precipitation in the northwest. However, the potential water supply varied spatially at different time spans in the HRW. The geographical distribution was similar from 1980 to 1995 (Figs. 4a−4c) in the south and north banks above Wangjiaba (E010200 and E020200) and gradually increased from the northwest to southeast. Sub-watersheds with a higher water supply were concentrated in the south bank above Wangjiaba (E010200), south bank of Wang-Beng (E020200), and Rigan region (E040500). However, from 2000 to 2005 (Figs 4d and 4e), a higher water supply was distributed from the northwest to southeast in a belt shape, whereas the water supply was relatively low in the upper reaches of the north bank of Wang-Beng (E020100), Huxi region (E040-200), and Hudong region(E040100). The water supply was higher in the north and south banks above Wangjiaba (E010100 and E010200) and downstream of the Yishuhe region (E040400), with an average maximum water supply of 1300 × 10 3 m 3 / km 2 and 1332 × 10 3 m 3 / km 2 , respectively. From 2010 to 2018 (Figs 4f−4h), the highest water supply in the HRW was relatively high, but gradually decreased from the southeast to northwest. Year Potential water supply Reference value Errors Fig. 3 Potential water supply and the errors between potential water supply and that published officially in the Huaihe River watershed, China during 1980China during -2018 The water supply was low in the north bank of the main stream of the Huaihe River and high in the south bank, which is mainly concentrated in the south bank of Wang-Beng (E020200) and Lixiahe region (E030200).

TN and TP exports into the river systems in the Huaihe River watershed
As shown in Table 1, the trends of TN and TP exports in the HRW over time were similar to those in other watersheds in China (Ma Lekuan, 2021). TN and TP exports increased from 1980 to 2005 but decreased from 2005 to 2018. Specifically, the exports of TN and TP into river systems gradually increased from 1980 to 2000 (33.28% increase for TN and 139.00% increase for TP), when the water quality of the other major watersheds in China continued to deteriorate. The TN and TP export then decreased slowly from 2000 to 2005. During this period, water quality improved in the major basins in China. From 2005 to 2018, the inflows of TN and TP into the HRW decreased significantly. During the same period, the water quality in the other major basins in China continued to improve. Table 2 shows the similar geographical distribution of the TN and TP exports at the sub-watersheds in 1980-2018. Generally, TN and TP converged into the river system along the river network, among which the annual TN and TP exports were high in the north bank of Wang-Beng (E020100), the south bank of Wang-Beng (E020200), the south bank above Wangjiaba (E010200), and theupper reaches of the Yishuhe and Lixiahe regions (E040400 and E030200). However, the geographical distribution of the annual TN and TP exports differed over time. Generally, the maximum annual TN and TP exports rose during 1980-1995, concentrating in  the south bank above Wangjiaba (E010200), the south bank of Wang-Beng (E020200), the south bank of Beng-Hong (E020400), and the Gaotian regions (E030100). The maximum annual TN and TP exports reduced from 1995 to 2018. For example, the maximum annual TN export continuously increased from 0.33 × 10 4 t/yr in 1980 to 242.32 kg/km 2 in 1995, and dropped to 0.36 × 10 4 t/yr in the north bank of Wang-Beng (E020100). At the same period, the maximum TP export showed the similar time change in this region. Fig. 5 show the water demand of each land use/land cover type and total water demand modeled with Eq. (6) in the HRW during 1980-2018. As shown in Fig. 5, total water demand rose from 1980 to 2010, but decreased slowly by 2018. Agricultural water demand showed similar time trend to that of total water demand at the same period, increasing slowly from 1980 to in 2010 and declining in 2018. With the development of the national economy, industrial water demand doubled between 1980 and 2018. Over the same period, the growth in domestic water demand was similar, increasing nearly four times in 1980-2018. Total water demand at county level has been estimated according to Eq. (6), as listed in Table S1 in the supplement file (http://egeoscien.neigae.ac.cn/index.htm). It can be seen in Table S1 that total water demand was highly correlated with the areas of paddy field, dry land, and urban and build-up with adjusted R 2 of 0.89 in 1980, and positively correlated with the areas of paddy field, dry land, urban and build-up, and rural settlements with adjusted R 2 of 0.78 in 2018. Then, total water demand has been rasterized on various land use/land cover at a space of 1 km in the HRW, that is, average water-demand coefficients under different land use/cov-   Fig. 6 shows the geographical distribution of the WD-SRs without considering the impacts of TN and TP pollution on water supply during 1980-2018. On the whole, water supply could meet the water demand at most regions in the HRW during 1980-2010. However, water supply could not meet water demand in more and more areas after 2010. WDSRs varied spatially across the whole watershed from 1980 to 2018. On the whole, there was 79.21% of the whole watershed where water supply was greater than water demand, and 20.79% of the watershed where water supply could not meet the water demand during 1980-2018. WDSR ranged from 0 to 1 in 1980 in most sub-basins of the HRW, meaning that the water supply was meeting society's water demand (Fig. 6a). The WDSR was between 1 and 5 in some counties of the Huxi region (E040200), resulting in local water shortages. It also ranged from 1 to 5 in some urban areas such as Zhengzhou and Kaifeng, upstream of the north bank of Wang-Beng (E020100), where the gap between water supply and demand was more marked. Fig. 6b shows that the range of the WDSR was lower in 1990 than in 1980, which bridged the gap between water supply and demand to a certain extent. The WDSR was between 0 and 1 in most areas of the HRW, while it ranged from 1 to 5 in other areas and from 15 to 18.25 in urban areas, which faced water shortages. Fig. 6c shows that the WDSR was between 0 and 1 in most areas of the HRW in 1995, meaning excess supply. However, it ranged from 1 to 5 in some areas with an increasing gap between water demand and supply, mainly in the southeast of the Huxi region (E040200), intersection between the south bank of Wang-Beng (E020200), south bank of Beng-Hong (E020400), Gaotian region (E030100), and most areas of the Lixiahe region (E030200). Regions with a WD-SR of 15.00-49.97 were distributed sporadically on the north bank of the main stream of the Huai River, such as the north bank of Wang-Beng (E020100), north bank of Beng-Hong (E020300), and Huxi region (E040100). In 2000, the geographical distribution of the WDSR was similar to that in 1990 (Fig. 6d), and there was a similar geographical distribution between 2005 and 2010, too (Fig. 6e). Compared with 2005, the WDSR decreased in the HRW in 2010 (Fig. 6f) and the maximum ratio fell to 17.19 in some areas. Areas with a WDSR between 1 and 5 expanded significantly and were concentrated in the southeast of the Huxi region (E040100) and areas bordering the north bank of Weng-Hong (E020300), Zhongyunhe region (E040300), and Yishuhe region (E040400). Compared with previous years, the WDSRs in the HRW showed geographical differences in 2015 and 2018. Overall, areas with WDSRs ranging from 0 to 1 gradually expanded and the gap between water demand and supply was more marked in some areas. In 2015, most areas of the HRW had WDSRs between 0 and 1 (Fig. 6g), mainly distributed in the south bank and downstream of the tributaries on the north bank of the Huai River. Areas with WDSRs ranging from 1 to 5 (excess demand) were mainly distributed upstream of the north bank of Wang-Beng (E020100), upstream of the Huxi region (E040200), in the Hudong region (E040100), in the Rigan region (E040500), and upstream of the Yishuhe region (E040400). The gap between water demand and supply intensified in areas where the WDSR was 5-10, mainly scattered in the urban areas on the north bank of the main stream of the Huaihe River. The gap between the water supply and demand was severe in some areas. For example, the WDSR ranged from 20 to 25 in Zhengzhou. In 2018, the geographical distribution of the WDSR in the HRW was similar to that in 2015 (Fig. 6h), but the relationship between water demand and supply was more intense upstream of the north bank of Wang-Beng (E020100) and in the Huxi region (E040200).

Relationship between water demand and supply considering TN pollution
Considering the impact of water-quality indicator TN on water demand-supply, water supply could be classified into different water bodies. Fig. 7 shows the trends of the water supply and water shortage in the HRW under China's TN water-quality standards during 1980-2018. Fig. 7a illustrates that total water supply showed a Wshaped trend over time. Grade-I water showed a Wshaped trend and Grade-III water showed an inverted Wshaped trend over time, whereas Grade-II water showed an inverted V-shaped trend.   (Fig. 7b). The amount of inferior Grade-V water was 0.01 × 10 7 m 3 in 1980, which increased to 0.23 × 10 8 m 3 in 1995, but decreased to 0.10 × 10 8 m 3 by 2018. More details were listed in Tables S2 and S3 in the supplement file (http://egeoscien.neigae.ac.cn/index.htm). Fig. 8 depicts spatial distribution of the WDSRs under the context of TN concentration during 1980-2018. In 1980, areas with Grade-I water and a WDSR between 0 and 1 were mainly distributed along the south bank and low reaches of north bank of the main stream of the Huaihe River (Fig. 8a). The water bodies mainly met the quality of Grade-II and Grade-III water in the upper and middle reaches of the tributaries on the north bank of the Huai River. There was excess supply and no water shortage pressure. Remaining areas had Grade-V and inferior Grade-V water bodies. Compared with the spatial distribution of WDSR in 1980, the water bodies met Grade-II water quality with the WDSR of 0-1 in the western part of the HRW in 1990 (Fig. 8b), mainly locating in the Rigan region (E040500), and the Yishuhe region (E040400). Major lake areas met Grade-I water quality and had a WDSR between 0 and 1. Here, the water bodies met the water quality standards of local water demand. The spatial pattern of WDSR in 1995 is similar to that in 1990 in the whole watershed (Fig. 8c), but the areas narrowed that met Grade-I water quality and with WDSRs between 0 and 1. Compared with the distribution of WDSR under the TN concentration limit in 1995, the spatial agglomeration of the WDSR gradually increased in 2000 (Fig. 8d), the scope for water bodies with Grade-I water quality and a WDSR of 0-1 expanded, mainly locating the north bank above Wangjiaba (E010100) and the Yishuhe region (E040400). Areas with Grade-III water quality and excess supply mainly included the Hudong region (E040100) and south bank of Beng-Hong (E020400).
In 2005, water quality had improved in the HRW (Fig. 8e) than that in 2000. Some water bodies met Grade-I water and had a WDSR between 0 and 1, mainly distributing in the north bank above Wangjiaba (E010100), downstream of the north bank of Wang-Beng (E020100), north bank of Beng-Hong (E020300), Zhongyunhe region (E040300), and downstream of the Yishuhe region (E040400). Water bodies in other areas met Grade-II water with a WDSR of 0-1. As shown in , the areas that water quality was classified as Grade-III water with the WDSR of 0-1 included the down stream of the north bank of Wang-Beng (E020100) and the north bank of Beng-Hong (E020300). Hence, there was excess supply.
As shown in Fig. 8g, the geographical distribution was more complex than ever before under the different water quality standards in 2015. Water bodies with Grade-I and a WDSR of 0-1 were mainly distributed in the south bank of Wang-Beng (E020200), south bank of Beng-Hong (E020400), Gaotian region (E030100), and Lixiahe region (E030200). Water bodies with Grade-II water quality and a WDSR of 0-1 mainly distributed on the south bank above Wangjiaba (E010200), downstream of the north bank of Wang-Beng (E020100), and downstream of the north bank of Beng-Hong (E020300). Water bodies with Grade-III water quality and a WDSR of 0-1 mainly distributed in the middle reaches of the north bank of Wang-Beng (E020100), Huxi region (E040200), and downstream of the Huxi region. Elsewhere, the water quality ranged from Grade-V to inferior Grade-V water and the WDSR was higher than 1; therefore, there was excess demand. By 2018, water quality improved throughout the HRW and the main water supply types were Grade-I and Grade-II (Fig. 8h) and the WDSRs ranged from 0 to 1. Grade-I water bod-   Table S4 in the supplement file (http:// egeoscien.neigae.ac.cn/index.htm) ies mainly distributed in downstream of the tributaries on the south and north banks of the main stream of the Huaihe River. Fig. 9 shows the trend of water-supply quantity and water shortages in 1980-2018 under the TP concentration limit, where is different from that under the TN concentration limit. Fig. 9a illustrates that water-supply quantity showed a W-shaped trend in 1980-2018. The water supply decreased from 973.43 × 10 8 m 3 in 1980 to 506.27 × 10 8 m 3 in 1995, increased to 1042.65 × 10 8 m 3 in 2005, and decreased to 788.37 × 10 8 m 3 in 2018. Grade-I water and Grade-II water were the main water supply types. Grade-I water quantity showed the same W-shaped trend as that of Grade-II water quantity. During the same period, the water shortage in the HRW increased from 12.45 × 10 8 m 3 in 1980 to 202.52 × 10 8 m 3 in 2005 and then decreased to 89.60 × 10 8 m 3 in 2018 (Fig. 9b). The overall trend was an inverted V-shape. Fig. 10 shows the geographical distribution of WD-SRs under the different TP concentration limits and relationship between water supply and demand during 1980-2018. The main water bodies that met Grade-I water quality and the WDSR ranged from 0 to 1 in most areas of the HRW in 1980 (Fig. 10a) changed into those meeting Grade-II water quality and with WDSRs over 0-1 and 1-50 (Fig. 10b), with gradually decreasing water quality throughout the HRW till 1990. During 1995-2000 10c and 10d), water pollution gradually deteriorated in some areas such as the upper reaches of Wang-Beng intersection (E020100) and Yishuhe region (E040-400), and most water bodies met Grade-II water quality, and the contradiction between water supply and demand began to rise. Compare with spatial patterns in 1995 and 2000, the obvious feature was that the areas narrowed with water supply not meeting water demand in some areas of the HRW from 2005 to 2010 (Figs. 10e-10d), and water quality improved at the same period. In 2015, the main water bodies were ranked Grade I and Grade III, the WDSRs ranged from 0 to 1, and there was excess supply (see Fig.10g). Grade-I water was mainly located upstream on the south bank of Wang-Beng (E020200), the Gaotian region (E030100), and the Lixiahe region (E030200). Areas with Grade-III water quality and a WDSR of 1-50 were distributed upstream of the north bank of Wang-Beng (E020100), upstream of the Huxi region (E040200) and in the Hudong region (E040100) and Rigan region (E040500). Fig. 10h shows Regions that met Grade-II and Grade-III water quality and had WDSRs of 0-1 mainly distributed in the upper and middle reaches of the north bank of Wang-Beng (E020100), upper reaches of the north bank of Beng-Hong (E020400), the Hudong region (E040100), the Zhongyunhe region (E040300), upper reaches of the Yishuhe region (E040400), and the Rigan region (E040500). Areas with Grade-III water quality were mainly distributed up streams of Wang-Beng (E020100 and E020200) and in the Huxi region (E040200).

Summaries for the water demand-supply ratios without/with considering the impacts of TN and TP
Without considering the impacts of TN and TP on the WDSRs, water supply could meet water demand in most regions of the HRW during 1980-2018. However, under the context of TN and TP pollutants, spatio-temporal patterns for the WDSRs varied, and water supply could not meet local water demand in some areas of the HRW. Time-series data for TN and TP showed that TN accounted for the majority of TN and TP pollutants

TN and TP pollutants influence the spatial distribution of the water demand-supply ratios
With population growth and economic development, human activities have affected the process of landform, land use/cover type, precipitation, and water supply, resulting in the destruction of watershed ecosystem services. To some extent, the stress of human activities in freshwater ecosystems has exceeded that of natural factors, such as the impacts of TN and TP pollutants on the WDSRs. There were a lot of literatures for examining the impacts of TN & TP on water bodies, but most researches have focused on modeling TN & TP or the spatio-temporal distribution of their concentrations by using data for precipitation, meteorological observations, land use/land cover, and so on (Huang and Guo, 2014;Cui et al., 2018;Xue et al., 2022). Another study showed that there existed a positive relationship between TN & TP and Grade-IV and above (Gu et al., 2019), however, few articles deeply detected the relationship between TN & TP pollutants and water supply or even the WDSRs at a watershed scale. That is, how much did water supply decrease under different TN or TP concentration limits? How did TN and TP effect spatial patterns for the WD-SRs? This paper attempted to model TN and TP loads and their impacts on the WDSRs under different TN and TP concentrations as an example of the HRW. Our findings showed that TN was the main pollutant among TN and TP pollution in the HRW during 1980-2018. Potential water supply, actual water supply, and water shortages have been modeled at the same period with/without considering TN and TP concentration limits. The spatial patterns of the WDSRs have been depicted without considering the contexts of TN and TP pollutants. On the average, there was 79.21% of the whole watershed with water supply greater than water demand, and 20.79% of the watershed with water supply not meeting the water demand during 1980-2018. However, the spatial distribution of the WDSRs varied differently across the HRW when TN and TP pollutants were considered.

The implication and limitation of this study
Although water demand-supply have been simulated under two contexts of considering and not considering TN and TP pollution in the HRW during 1980-2018, the implications of this study are as follows.
First, the implication is that water-supply simulation should consider the impact of water pollution discharged from human activities, e.g., agricultural chemical fertilizer. If the theoretical or potential maximum water supply needs to be estimated, the effect from water pollutants can be ignored. However, if we want to obtain the information of actual water supply under various water-quality standards at a regional-to-global scale, the impacts of TN and TP pollution on water supply should be considered. The water supply at these two scenarios is very important for water resource management and allocation.
Second, except for data on total amount of water resources, the information on spatio-temporal distribution for water supply-demand at more detailed space resolution are also needed when the policies for water resource allocation will be made at various administrative levels. These spatial data, particularly considering the impacts of point or non-point water pollution on water supply-demand, are very important for identifying the regions of water shortage and excess water supply under the context of big data management.
This study has limitations that suggest future research avenues. First, under the InVEST water yield model, the simulated potential water supply is not classified into surface water, underground water, and soil water and it is assumed that the simulated water supply can meet the water quality requirements needed for socioeconomic development without considering the impacts of TN/TP non-point water pollution. The gap between the simulated potential and actual water supply prevents us from formulating reasonable watershed management policies. Therefore, it is necessary to research the combination of the InVEST water yield model and water quality as well as develop the model into 1) a potential water supply simulation without the influence of non-point/point water pollution and 2) an actual water supply simulation with the influence of water pollution discharge to quantify how water pollution discharge affects the water supply. Second, we used the estimation coefficient/inventory analysis method to evaluate the quantities of TN and TP in the HRW to correct the parameters in the InVEST water purification model. However, only the TN and TP pollutants from agricultural fertilizer application, rural life, and livestock and poultry breeding were considered, and TN and TP pollutants from urban point source were not considered. Untreated urban industrial and domestic TN and TP pollutants, especially in areas with slow economic development and low sewage efficiency, enter the river systems along the urban drainage network with precipitation, which affects the TN and TP pollutants in this basin and thus the simulation of the pollution loads of different land use types. Therefore, it is necessary to further study the point/non-point source pollution loads of different land use/cover types to provide a scientific basis for the spatial simulation of pollution loads. Third, one problem is the lack of a standard for verifying the correct parameters in the InVEST water purification model. To simulate the inflow of TN and TP pollutants in this study, the Morris method was selected to verify the parameters. Using this method meant that the simulation results of the TN and TP pollutant loads were objectively close to the actual inflows of TN and TP in the HRW. This reduced the uncertainty in the simulation results of the water purification model and provided a reference for the simulation process of water purification in other basins. The solutions to solve the above limitations are to develop native software for modeling actual water supply and the WDSRs under the contexts of water pollutants such as TN, TP, and COD. Meanwhile, parameter calibration methods and standards for models are also involved in the software. Another prob-lem is to monitor the flows of TN and TP into river systems and include them in the National Water Pollution Prevention & Control Plan for decreasing the impacts of TN and TP on water bodies.

Conclusions
In this study, the geographical pattern of water demandsupply ratios in the HRW was modeled under the two scenarios of considering and not considering the impacts of non-point source pollution on water demand and supply. The main findings can be summarized as follows.
(1) Water supply and WDSR without considering the impacts of TN&TP pollution Not considering the impacts of TN&TP pollution in the HRW, the quantity of water supply showed a 'Wtype' time trend. On the whole, there was 79.21% of the areas where water supply could meet water demand, and 20.79% of the regions facing the pressure of water shortage in each sub-basins of the HRW from 1980 to 2018. From 2005 to 2018, the contradiction between water supply and demand intensified in the upper reaches of the north bank of the main stream of the Huai River.
(2) Impacts of TN and TP pollution on the WDSRs First, under the TN concentration limit, the water quantity presented a W shape over time, of which inferior Grade-V water that could not be used for any purpose increased first and then decreased over 1980-2018. In 2015, peak inferior Grade-V water bodies reached 0.48 × 10 8 m 3 . Grade-I, Grade-II, and Grade-III water were the main supply types in the HRW. Areas that met the Grade-I water quality and a WDSR of 0-1 were mainly concentrated in the Lixiahe region (E030200), south bank of Wang-Beng (E020200), north bank of Wang-Beng (E020100), and Yishuhe region (E040400). Regions with Grade-II water quality and excess supply (WDSR < 1) were concentrated in the north bank of Wang-Beng (E020100), Lixiahe region (E030200), Huxi region (E040200), and Yishuhe region (E040400). Areas with Grade-III water quality and excess supply (WDSR < 1) were mainly located in the north bank of Wang-Beng (E020100) and Yishuhe region (E040400). Sub-watersheds that met Grade-I water quality and had a WDSR over 1 were mainly concentrated in the north bank of Wang-Beng (E020100) and Yishuhe region (E040400). Regions that met the Grade-II and Grade-III water quality standards and had excess demand (WDSR ≥ 1) included the north bank of Wang-Beng (E020100) and Huxi region (E040200).
Second, under the context of TP concentration limit, Grade-I water and Grade-II water were the main water supply types. During the same period, the water shortage in the HRW increased from 12.45 × 10 8 m 3 in 1980 to 202.52 × 10 8 m 3 in 2005 and then decreased to 89.60 × 10 8 m 3 in 2018. Compared with the WDSRs under the TN concentration limit, spatial distribution of the WDSRs under the TP concentration limit varied slightly except for those in 1995, 2000 and 2015. The WDSRs over 1 (water supply < water demand) mainly located in the upstreams of the north bank of Wang-Beng (E020100), the Hudong region (E040100), the Huxi region (E040200) and the Yishuhe region (E040400).
(3) The academic suggestion from studying the impacts of TN and TP pollution on the WDSRs Reference values for actual water supply, TN and TP have been used to correct the parameters in InVEST water yield and water purification models when potential water supply and exports from TN and TP pollution have been modeled. Theoretically, this can make up for the lack of reference values to verify parameters in previous studies, which can make the simulation results more accurate. Additionally, water pollutants such as TN and TP must be considered when actual water supply or water demand-supply will be evaluated at various water scales. The findings of this study provide important information on the impacts of TN and TP pollution on the water demand-supply relationship as well as on the TN and TP pollution loads for correcting the parameters in the InVEST water purification model.