Advertisement

Earth Systems and Environment

, Volume 2, Issue 1, pp 67–84 | Cite as

Long-Term Hydrologic Impact Assessment of Non-point Source Pollution Measured Through Land Use/Land Cover (LULC) Changes in a Tropical Complex Catchment

  • Jabir Haruna Abdulkareem
  • Wan Nor Azmin Sulaiman
  • Biswajeet Pradhan
  • Nor Rohaizah Jamil
Original Article

Abstract

The contribution of non-point source pollution (NPS) to the contamination of surface water is an issue of growing concern. Non-point source (NPS) pollutants are of various types and altered by several site-specific factors making them difficult to control due to complex uncertainties involve in their behavior. Kelantan River basin, Malaysia is a tropical catchment receiving heavy monsoon rainfall coupled with intense land use/land cover (LULC) changes making the area consistently flood prone thereby deteriorating the surface water quality in the area. This study was conducted to determine the spatio-temporal variation of NPS pollutant loads among different LULC changes and to establish a NPS pollutant loads relationships among LULC conditions and sub-basins in each catchment. Four pollutants parameters such as total suspended solids (TSS), total phosphorus (TP), total nitrogen (TN) and ammonia nitrogen (AN) were chosen with their corresponding event mean concentration values (EMC). Soil map and LULC change maps corresponding to 1984, 2002 and 2013 were used for the calculation of runoff and NPS pollutant loads using numeric integration in a GIS environment. Analysis of Variance (ANOVA) was conducted for the comparison of NPS pollutant loads among the three LULC conditions used and the sub-basins in each catchment. The results showed that the spatio-temporal variation of pollutant loads in almost all the catchments increased with changes in LULC condition as one moves from 1984 to 2013, with 2013 LULC condition found as the dominant in almost all cases. NPS pollutant loads among different LULC changes also increased with changes in LULC condition from 1984 to 2013. While urbanization was found to be the dominant LULC change with the highest pollutant load in all the catchments. Results from ANOVA reveals that statistically most significant (p < 0.05) pollutant loads were obtained from 2013 LULC conditions, while statistically least significant (p < 0.05) pollutant loads were obtained under 1984 LULC condition. This reveals the clear effect of LULC changes on NPS pollution. The findings of this study may be useful to water resource planners in controlling water pollution for future planning.

Keywords

Hydrologic impact Remote sensing Non-point source pollution GIS Malaysia 

1 Introduction

Water balance of a watershed is influenced by land use/land cover changes (LULC) (Fohrer et al. 2001). LULC change is the single most prominent element responsible for changes at both regional and global scale which results to several types of natural and biological changes (Bai, et al. 2008; Wang and Wang 2013; Li et al. 2016). A watershed becomes more hydrologically active when it develops with changing runoff components, stream flow and flood volume. The effect of LULC on storm runoff generation is very complicated. Several studies in the past have identified that LULC having a strong impact on water quality (Thanapakpawin et al. 2007; Zaimes et al. 2008; Shen et al. 2010) predominantly because of non-point source pollution (NPS) which are known to have a direct relationship with LULC change (Girmay et al. 2009). LULC changes adversely affect NPS pollutant loads by altering their sources and the way they are being transferred into water bodies.

As water is being drained from the land surface, it has the tendency to carry along with it residues of several types to other water bodies. Under the first flush phenomena, surface run off is a major source of non-point source pollution (NPS). The type of contaminant depends on the run off that is associated with the LULC and the Event Mean Concentrations (EMC) values of the pollutant load (Engel 2001; Novotny 2003). EMC quantifies the volume of pollutants conveyed per unit volume of runoff. For example, the major contaminants from run off that will pollute agricultural land use will be nutrients (mostly nitrogen and phosphorus) and sediments. Runoff from highly urbanized areas, on the other hand, may be polluted with rubber fragments, heavy metals, in addition to sodium and sulfate from road (Tong and Chen 2002). The problem of NPS pollution is an issue of great concern as it poses a great risk to water quality in developed countries (US EPA 2009).

Unlike point source pollution whose sources are known, NPS pollution is peculiar to complex mechanisms and techniques that are random and sporadic in occurrence. In addition to this, NPS pollution poses uncertainty with regards to discharge in channels and amounts, variability in both temporal pollution loads which results to difficulties in monitoring, simulation, treatment as well as control. To tackle the risk of NPS pollution, it is vital to have precise simulations and estimations of NPS (Shen et al. 2012).

Quantification of several kinds of NPS pollution is a key issue for successful land use planning as well as in alleviating the hazard (Engel et al. 2003; Zhang et al. 2011). Pollutant load estimation carried out through monitoring activities is a complex process that involves precise computation of both pollutant concentration (EMC) as well as runoff and accurate calculations that are mostly based on statistical methods. Therefore, it is important to set up initial monitoring activities of NPS pollutants for good load estimation (Meals et al. 2013). Comprehensive knowledge of the areas’ topography and NPS sources is a prerequisite when characterizing pollutants. Identification and location of NPSs of pollution is desirable for pollutant loads and should be fully evaluated.

Several studies have been conducted in the past in Malaysia based on spot field evaluation on NPS pollution (e.g., Yusop et al. 2005; Chow and Yusop 2006; Eisakhani et al. 2009; Nazahiyah et al. 2007; Chow et al. 2011). It should, however, be noted that, the spatio-temporal variation of NPS pollution cannot be attained through spot field investigation and short-term monitoring, therefore, researches should inevitably be conducted using mathematical models (Li et al. 2016). The Kelantan River basin in the north-eastern part of Peninsular Malaysia was chosen as the study area due to its constant and frequent incidences of flooding which leads to build up of NPS pollution load in the area. The watershed is under illegal and unrestricted land cover conversion without giving attention to the environmental consequences which has altered the natural hydrologic system of the basin giving rise to several incidences of flood. This research has three main objectives; to determine the spatio-temporal variation of NPS pollutant loads among different catchments in Kelantan river basin, to determine the temporal variation of NPS pollutant loads among different LULC changes and to establish NPS pollutant loads relationship among different sub-basins in each catchment and LULC conditions.

2 Study Area

Kelantan River basin is positioned in the north-eastern part of Peninsular Malaysia between latitudes 4° 40′ and 6° 12′ north, and longitudes 101° 20′ and 102° 20′ east. The capital city of Kelantan is Kota Bharu which is situated at the Northern part of the state. Kelantan state occupies 4.40% of Malaysia’s total area with a total of 15,099 km2. The state has an estimated population of 1.539 million. The maximum length and breadth of the catchment are 150 and 140 km, respectively. The length of the main river is about 248 km long which drains an area of about 13,100 km2, occupying more than 85% of the Kelantan state. The estimated quantity of the annual precipitation in the basin is about 2383 ± 120 mm, a large amount of which occurs during the north-east monsoon between mid-October and mid-January. The basin has an estimated runoff discharge of 500 m3 s−1 (DID, 2000). The average annual temperature at Kota Bharu is 27.5 °C with mean relative humidity of 81%. The average flow of the Kelantan River measured at Guillemard Bridge is 557 m3 s−1.

The Kelantan River divides into Galas and Lebir Rivers near Kuala Krai about 100 km from the river mouth. Galas River is formed by the junction of Nenggiri and Pergau Rivers. The origin of Nenggiri River is from the south-western part of the main mountain range. While the origin of Lebir River is from Tahan mountain range. The flow direction of Kelantan River system is northward where it passes along major towns like Kuala Krai, Tanah Merah, Pasir Mas and Kota Bharu, before finally discharging into the South China Sea. The majority of the catchment is steep mountainous area rising to a height of 2135 m, occupying 95% of the area while the rest is undulating land. The mountainous areas are covered primary with virgin forest while rubber and some paddy are cultivated in the lowlands. The eastern and western parts of the watershed, consists of mountains of various ranges, while the soil cover is granitic comprising a combination of fine to coarse sand and clay. The soil cover is approximately a meter deep on average but depths of more than 18 m may be encountered in localized areas. In the extreme of the southern half of the basin, the major soil type is fine sandy loam, which has its depth rarely exceeding a few meters. The other part consisting of almost one-third of the basin is covered by a variable soil cover that varies in depth, which is about 1 m to more than 9 m. The forested areas, mostly in the Lojing highlands, are experiencing serious logging activities which some people believe is the major cause of recent floods in the basin. The map of the study area is shown in Fig. 1.
Fig. 1

Map of Kelantan river basin showing Galas, Pergau, Lebir, and Nenggiri catchments with rain gauge and discharge stations

3 Methodology

The flow chart adopted in this study is shown in Fig. 2. The framework is divided into four components. First, digital elevation model, LULC maps, remote sensing images and soil maps were prepared in a GIS environment using ArcMap 10.3. Second, ArcCN within the ArcMap environment was used to calculate curve numbers (CN) and runoff amount. The third component involves the calculation of pollutant load using numeric integration and, lastly the relationship among LULC condition based on NPS pollution loads were statistically analyzed.
Fig. 2

Methodological framework adopted in this study

In this paper, pollutant loads were estimated using numeric integration in a geographic information system (GIS) environment. Land use maps corresponding to 1984, 2002 and 2013 and SPOT 5 images were used to obtain the LULC change of the watershed. The digital elevation model (DEM) was used to extract the physiographic characteristics in the area, soil maps were used to derive the soil properties, long-term daily hydrological gauged data from 1984 to 2014 were obtained. The findings of this study will of serious help to water resource planners in controlling water pollution for future planning.

3.1 Data Sources and Preparation

Advanced Spaceborne Thermal Emission and Reflection Radiometer (ASTER) Global Digital Elevation Model (GDEM) with 30 m spatial resolution was used as the Digital Elevation Model (DEM) source in this research for extracting the physiographic features of the basin. HEC-GeoHMS which is an extension of ArcMap software was used in delineating the basin. Four major sub-basins were extracted from the DEM, e.g., Galas, Nenggiri, Lebir, and Pergau. These basins were further divided into smaller sub-basins considering geomorphological similarities to increase the precision of runoff and NPS load estimation. LULC maps corresponding to the three LULC conditions (1984, 2002 and 2013) as well as soil maps are shown in Fig. S1, while the DEM for the four basins is shown in Fig. S2. Table S1 summarizes the description of datasets used in this study.

3.2 Land Use Classes, Rainfall and Soil Type

Three land use maps corresponding to years 1984, 2002 and 2013 and soil series maps of the basin were obtained from the Department of Agriculture (DOA), Malaysia. Land use classes in the LULC maps were categorized into forest, paddy, agriculture, grassland, urbanization, cleared land, mangrove swamp, secondary forest, rivers, ponds and lakes, and mining for ease of analysis. Daily rainfall and runoff data were obtained from Department of Irrigation and Drainage, Malaysia for the period of 1984–2014 and annual average rainfall were used for calibration and validation of ArcCN (Zhan and Huang 2004). Rainfall and discharge stations are shown in Fig. 1. These stations were chosen in this study based on their complete records and data availability.

3.3 Runoff Estimation using ArcCN

The Soil Conservation Service Curve Number (SCS-CN) method was used in this study. It was based on a water balance and two fundamental hypotheses (SCS 1956).
$$Q = \frac{{(P + I_{\text{a}} )^{2} }}{{P - I_{\text{a}} + S}}, \quad P \ge 0.2S$$
(1)
$$S = \frac{25400}{\text{CN}} - 254,$$
(2)
where Q is the direct runoff (mm), P is the rainfall (mm), Ia is the initial abstraction (mm), S is the potential maximum retention after runoff begins (mm). CN values were determined by intersecting each of the LULC map with the soil map using ArcCN script in ArcMap from the land/soil intersect file (Zhan and Huang 2004). To determine runoff for the year 1984, available rainfall stations were used to determine average rainfall for the year 1984 LULC condition. For the 2002 LULC condition, averages calculated from 1984 to 2002 were used for computing runoff for that year whereas, for 2013 LULC condition runoff was determined using averages calculated from 1984 to 2013. These averages were further divided by 12 keeping in mind 12 large rainfall events corresponding to each LULC condition. Fig. S3 shows rainfall trends in Kelantan River. More detailed description of the ArcCN method can be obtained from Zhan and Huang (2004).

3.4 Pollutant Load Estimation

Pollutant loads are expressed as mass or weight of a pollutant that is transferred through a cross-sectional area of water body (rivers, streams) at a specified time. It is expressed in mass units (kilograms, tons), although the time interval is inherent when pollutant loads are formed, it should, however, be distinct from context. The NPS pollutant load studied in this research are; total suspended solids (TSS), total phosphorus (TP), total nitrogen (TN) and ammonia nitrogen (AN). The load of each pollutant was calculated using numeric integration (Meals et al. 2013) carried out in a GIS environment with the formula given below;
$${\text{Load}} = \mathop \sum \limits_{i = 1}^{n} c_{i} q_{i} t_{i} ,$$
(3)
where c i is the event mean concentration in the ith sample, q i is the corresponding runoff values, and t i is the time interval represented by the ith sample, calculated using Eq. 4 below;
$$\frac{1}{2}\left( {t_{i + 1} - t_{i - 1} } \right).$$
(4)

3.4.1 Event Mean Concentration (EMC)

Event mean concentration is the mean concentration of an urban pollutant measured during a storm runoff event. It can also be defined as the total mass of total constituents discharged expressed over the total runoff volume (Huber 1993; Adams and Papa 2001). It can be expressed in the equation below;
$${\text{EMC}} = C = \frac{M}{V} = \frac{{\mathop \smallint \nolimits Q\left( t \right)C\left( t \right){\text{d}}t}}{{\mathop \smallint \nolimits Q\left( t \right){\text{d}}t}},$$
(5)
where M is total mass of pollutant during the entire runoff (kg), V is total volume of runoff (m 3 ), C(t) is time varying pollutant concentration (mg L−1), Q(t) is time variable flow (L s−1), and t is total duration of runoff (s). It should, however, be noted that EMC results from a flow-weighted average does not represent the time average of the concentration. Table S2 shows the EMC values used in this study which is adopted from DID (2012).

3.5 Model Validation and Statistical analysis

Runoff values calculated from SCS-CN method were validated prior to pollutants load estimation. This procedure was carried out by comparing measured runoff values from 1984, 2002 and 2013; predicted LULC conditions using correlation coefficient (R) and Nash–Sutcliffe Efficiency (NSE) (Nash and Sutcliffe 1970). The procedures are presented in Eqs. (6) and (7) below.
$$R = \frac{{\mathop \sum \nolimits_{i = 1}^{N} \left( {Q_{\text{Obs}} - \overline{Q}_{\text{Obs}} } \right)\left( {Q_{\text{sim}} - \overline{Q}_{\text{sim}} } \right)}}{{\sqrt {\mathop \sum \nolimits_{i = 1}^{N} \left( {Q_{\text{Obs}} - \overline{Q}_{\text{Obs}} } \right)^{2} \left( {Q_{\text{sim}} - \overline{Q}_{\text{sim}} } \right)^{2} } }}$$
(6)
$$E = 1 - \frac{{\mathop \sum \nolimits_{i = 1}^{N} (Q_{{i{\text{Obs}}}} - Q_{{i{\text{Sim}}}} )^{2} }}{{\mathop \sum \nolimits_{i = 1}^{N} (Q_{{i{\text{Obs}}}} - \overline{Q}_{{i{\text{Obs}}}} )^{2} }},$$
(7)
where, QSim is the simulated discharge at time t = i, QObs is the observed discharge at time t = i, \(\overline{Q}_{\text{sim}}\) is the average simulated discharge \(\overline{Q}_{\text{Obs}}\) is the average observed discharge; N is the number of observations.

Measured average annual rainfall data were used to calculate average annual runoff of each of the stations used in the study area for the year 1984 which represents 1984 LULC condition. For the 2002 LULC condition, the average annual runoff data from 1984 to 2002 was used, whereas for the 2013 LULC condition the average annual runoff data from 1984 to 2013 was calculated from rainfall during that period. These averages were further divided by 12 keeping in mind that there are 12 large runoff annual events corresponding to each LULC condition.

Statistical Analytical System (SAS) version 9.4 was used to carry out Analysis Of Variance (ANOVA) for the comparison of NPS pollutant loads among the three LULC conditions in this study for the entire watershed. Mean separation was carried out using least significance difference (LSD) and significant means were grouped using Tukey’s range test. This comparison was aimed at exploring statistical similarities or differences existing among sub-basins in each catchment and different LULC conditions.

4 Results and Discussion

4.1 Results of Model Validation and Statistical Analysis

The results of the validation are expressed graphically and are shown in Fig. 3. Values of R ranged from 0.7198 to 0.9018 indicating strong positive correlation between the measured and estimated model values in all the catchments. Changes in R may be attributed to changes in LULC condition and its quantity which was seen visible in all the years as well as in the catchments (Ozdemir and Elbaşı 2015). NSE values in this study ranged from 0.45 to 0.65 indicating moderate to higher model performance in all the catchments across different LULC conditions. Generally, NSE statistic ranged from − ∞ and 1.0, where 1 is considered as the optimum. Values between 0.00 and 1.00 are considered satisfactory, while values ≤ 0.00 implied that the measured hydrological values produced better results than the simulated values, thus, indicating unacceptable model performance (Moriasi et al. 2007; Ozdemir and Elbaşı 2015).
Fig. 3

Relationship between observed and predicted runoff values a Galas 1984 b Galas 2002 c Galas 2013 d Pergau 1984 e Pergau 2002 f Pergau 2013 g Lebir 1984 h Lebir 2002 i Lebir 2013 j Nenggiri 1984 k Nenggiri 2002 l Nenggiri 2013

4.2 Land Use Analyses of Past and Present LULC Changes

LULC changes were carefully analyzed in the major basins of Kelantan river basin from 1984 to 2014. The area (both in km2 and %) occupied by each LULC change with its corresponding change according to 1984, 2002 and 2013 and is shown in Fig. 7. Deforestation was observed as the dominant LULC change across the watershed while agricultural activities were observed to increase non-uniformly across the catchments due to increase in urbanization in the area.

In Galas basin (Fig. S4), an estimated deforestation of about 51% was observed, while in Pergau (Fig. S4), it was found to be around 19.32% during 1984–2002. Unlike deforestation, grassland had increased by 15.77% in Galas and 7.11% in Pergau whereas secondary forest has undergone little or no changes in both locations. From 2002 to 2013 smaller area of the forested lands were lost to deforestation for both Galas basin (3.40%) and Pergau basin (7.11%). The major LULC changes observed during this period, were decreased in grassland from 15.85 to 0.46% in Galas and from 12.33 to 0.53% in Pergau. On the other hand, an increase in secondary forest can be seen (from 0.35 to 8.68%) in Galas and from 0.00 to 11.19% in Pergau.

The past and present LULC changes in Lebir and Nenggiri catchments are presented in Fig. 7c, d. In these basins deforestation was also observed during 1984–2002. In Lebir catchment, 24.91% deforestation was observed, whereas in Nenggiri it accounts for 17.69% of the total LULC change from 1984 to 2002. Agricultural activities and grassland both increased (15.77 and 10.92%, respectively) in Lebir and 2.49% and 14.84%, respectively, in Nenggiri. Deforestation occurred slightly in Lebir (2.91%) whereas afforestation was recorded in Nenggiri (6.98%) from 2002 to 2013. The major LULC change was decreased in grassland for both Lebir and Nenggiri while urbanization and cleared land all witnessed slight increase during the same period under study.

In all the catchments deforestation was found to be the foremost LULC change observed during 1984 to 2014. Deforestation can be attributed mainly due to intense logging and agricultural activities in the area that is believed to be a source of concurrent runoff activities as reported in previous studies in the same location by (Wan 1996; Jamaliah 2007; Adnan and Atkinson 2011). This increase in runoff has glaring effect on NPS pollution in the area.

4.3 Effect of Runoff on NPS Pollution in Kelantan River Basin

In all the catchments, runoff values increased with changes in LULC condition which also lead to build up in NPS pollution. Urbanization on the other hand was found to have increased as the number of years’ increase is one of the major sources of surface runoff as well as NPS pollution due to increase in impervious surfaces that lead to increase in total pollutant load in the watershed. This increase contributed in altering the runoff dynamics of the basin as well as pollutant transport and delivery although not as rapid as deforestation. Previous surfaces that include all water bodies which have the tendency not to cause runoff were categorized as rivers, ponds and lakes. They were observed across the catchments predominantly under 2002 and 2013 LULC changes probably making the basins to have lower runoff and pollutant load deposition values under these conditions when compared with 1984 LULC condition where the catchment is dominated by forest.

For pollutant loads like TN, AN and TP that are associated with runoff from agricultural activities and are commonly considered as non-urban pollutants, it will be anticipated that transformation of non-urban land use types to urban land use type will lead to decrease of these pollutants in this watershed (Bhaduri et al. 2000). However, this is not the case in this study. The results showed that the transformation of non-urban land use types from 1984 to 2002 had caused an increase of about 0.54 and 0.43% in Pergau and Lebir respectively. This has led to an increase in TP by about 36% and TN by about 0.17% in Lebir and increase in TP by about 24% and TN by about 24% in Pergau. The rapid increase in agricultural activities from 7.20 to 25.72% in Pergau and that in Lebir from 8.29 to 24.06% is attributed to cause this increase of TP, TN, AN and TP during this period. Even though pollutant loads in urbanized land uses are produced at a slower rate than in non-urbanized land use types, one of the major consequences of urbanization is increase in runoff, which contribute enormously to NPS pollution build up. Another factor that may have influenced runoff, which directly affect the NPS load in the watershed, is the presence of high sloping areas. High values of pollutant loads recorded across the watershed may be due to undulating areas. Since the bigger the slope, the more the runoff as well the more the NPS pollution.

4.4 Spatio-Temporal Variation of NPS Pollutant Loads

The results of the spatio-temporal variation of NPS pollutant loads are presented in Fig. 4, 5, 6 and 7.
Fig. 4

TSS loads (kg year−1) in Galas, Pergau, Lebir and Nenggiri a 1984 b 2002 c 2013

Fig. 5

TP loads (kg year−1) in Galas, Pergau, Lebir and Nenggiri a 1984 b 2002 c 2013

Fig. 6

TN loads (kg year−1) in Galas, Pergau, Lebir and Nenggiri a 1984 b 2002 c 2013

Fig. 7

AN loads (kg year−1) in Galas, Pergau, Lebir and Nenggiri a 1984 b 2002 c 2013

4.4.1 TSS

TSS pollutants are derived from litters from agricultural and forested land uses, oil and grease from urbanized land use during storm runoff. Finer particles present high potential to block drainage system. These pollutants, therefore, need to be eliminated from time to time, if possible after each rain storm before the next to avoid NPS pollution build up (DID 2012). Since TSS loads were derived during storm events by involving multifaceted process such as build up as well as washing away from impervious surfaces, transfer in the sewers, sedimentation and re-suspension of sediments among others, etc. (Rossi et al 2005). Therefore, EMC values used for estimating TSS loads did not give accurate estimate, but rather an approximation of the loads in the area. Figure 4 shows the spatio-temporal variation of TSS NPS pollutant loads from 1984 to 2013 in Kelantan river basin. TSS was increased with changing LULC condition from 1984 to 2013 across the watershed. The total TSS loads in Galas (Fig. 4) increased from 106,959 kg year−1 in 1984 to 134,498 in 2013, while in Pergau it increased from 18,933 kg year−1 to 289,181 kg year−1 during the same period. In both Lebir and Nenggiri, highest TSS pollutant loads were obtained in sub-basins W160 and W200, respectively, probably due to their locations at the outlet, where high runoff may have favored the accumulation of TSS pollutant loads. By comparison, highest total TSS loads change from 1984 to 2013 were obtained in Lebir with 75%, followed by 62% from Nenggiri, 52% in Pergau and 26% in Galas.

4.4.2 TP

The sources of TP are both from agricultural activities and urbanization, which usually binds to soil particles. Figure 5 shows TP NPS pollutant load with its corresponding spatio-temporal variation from 1984 to 2013. The spatio-temporal variation of TP is distributed in accordance to changes in LULC condition. Total TP loads in both Galas and Pergau (Fig. 5) were increased by about 24% from 1984 to 2002 and by about 23% from 2002 to 2013. In Lebir, total TP loads ranged from 679 to 926 kg year−1 from 1984 to 2013, whereas in Nenggiri it ranged from 468 to 718 kg year−1. Although the highest TP was not obtained from 1984 LULC condition as obtained in other catchments, but rather under 2002 LULC condition. This may be attributed to higher pervious surfaces obtained under 2002 LULC condition as compared to that of 1984 LULC condition (Fig. S4). In all the catchments, except for Galas, highest total TP loads were recorded in sub-basins closest to the outlets. It could be affirmed that runoff at the outlet resulting from all the sub-basins leads to the accumulation of high TP loads in these catchments.

4.4.3 TN

The term TN encompasses AN, nitrate nitrogen, nitrite-nitrogen as well as organically bonded nitrogen. Agriculture and sewage have been identified as some of the most important sources of TN load in water. Tables 1, 2, 3 and 4 and Fig. 6 shows the trend of TN load in Kelantan river basin. Total TN load in Galas ranged from 2116 to 2757 kg year−1 and 3425–5232 kg year−1 from 1984 to 2013 (Fig. 6). Highest TN in Galas under 1984 LULC condition was found in sub-basin W80, W90 and W100 (452 kg year−1), whereas under 2002 and 2013 LULC condition, W80 dominated with 510 and 587 kg year−1, respectively. This dominance of sub-basin W80 over the others may be because of rapid deforestation in that area (from 1984 to 2002 and from 2002 to 2013) that may have resulted in increase in runoff, which in turns leads to TN loads accumulation. TN in Lebir increased by about 36% from 1984 to 2013 corresponding to increase in agriculture by about 67% (Fig. 6). Whereas, in Nenggiri, TN load was estimated at about 48% corresponding to over 300% increase in agricultural activities in the area (Fig. 6). Although, the percentage increase in agriculture in Nenggiri was much higher than Lebir, higher total TN loads are recorded in Lebir due to higher runoff activities when compared to Nenggiri.
Table 1

NPS pollutant load (kg year−1) in Galas according to land use changes from 1984 to 2013

Land use type

TSS

TP

TN

AN

1984

2002

2013

1984

2002

2013

1984

2002

2013

1984

2002

2013

Forest

2819

3197

3595

14.10

15.97

17.97

93.98

107

120

4.71

5.32

5.99

Paddy

15,240

16,917

19,156

42.28

47.95

53.91

376

426

479

4.71

5.32

5.99

Agriculture

15,240

16,917

19,156

42.28

47.95

53.91

376

426

479

4.71

5.32

5.99

Grassland

15,240

16,917

19,156

42.28

47.95

53.91

376

426

479

4.71

5.32

5.99

Urbanization

43,180

47,931

54,276

61.10

69.28

77.91

518

639

719

46.99

53.29

59.94

Cleared land

15,240

16,917

19,156

42.28

47.95

53.91

376

426

479

4.71

5.32

5.99

Total

106,959

118,796

134,495

244.

277

311

2116

2450

2755

70.54

79.89

89.89

Table 2

NPS pollutant loads (kg year−1) in Pergau according to land use changes from 1984 to 2013

Land use type

TSS

TP

TN

AN

1984

2002

2013

1984

2002

2013

1984

2002

2013

1984

2002

2013

Forest

4014

4968

6131

20.07

24.85

30.67

134

166

204

6.70

8.30

10.24

Paddy

20,071

24,840

30,655

60.21

74.50

91.97

535

662

817

6.70

8.30

10.24

Agriculture

20,071

24,840

30,655

60.21

74.50

91.97

535

662

817

6.70

8.30

10.24

Grassland

20,071

24,840

30,655

60.21

74.50

91.97

535

662

817

6.70

8.30

10.24

Urbanization

52,677

70,379

86,856

86.96

107.65

132.83

803

994

1226

66.91

83.85

102

Cleared land

24,262

24,840

30,655

60.21

74.50

91.97

535

662

817

6.70

8.30

10.24

Mangrove swamp

4014

4968

6131

20.07

24.85

30.67

134

166

204

6.70

8.30

10.24

Secondary forest

4014

4968

6131

20.07

24.85

30.67

134

166

204

6.70

8.30

10.24

Mining

40,142

49,679

61,310

80.28

99.37

123

Total

189,336

234,322

289,179

388

480

593

3425

4239

5229

114

142

174

Table 3

NPS pollutant loads (kg year−1) in Lebir according to land use changes from 1984 to 2013

Land use type

TSS

TP

TN

AN

1984

2002

2013

1984

2002

2013

1984

2002

2013

1984

2002

2013

Forest

8580

8855

12,084

42.88

44.28

60.42

298

295

403

14.29

14.63

20.11

Agriculture

42,901

44,273

60,422

186

133

181

1191

1181

1611

14.29

14.63

20.11

Grassland

42,901

44,273

60,422

186

133

181

1191

1181

1611

14.29

14.63

20.11

Urbanization

121,552

125,440

171,195

186

192

262

1768

1771

2417

144

146

201

Cleared land

42,901

44,273

533,423

186

133

181

1191

1181

1611

14.29

14.63

20.11

Secondary forest

8580

8855

12084

42.88

44.28

60.42

298

295

403

14.29

14.63

20.11

Mining

85,801

88,546

1,208,434

179

177

242

Total

353,216

364,515

2,058,064

830

680

926

6116

6081

8298

215

219

301

Table 4

NPS pollutant loads (kg year−1) in Nenggiri according to land use changes from 1984 to 2013

Land use type

TSS

TP

TN

AN

1984

2002

2013

1984

2002

2013

1984

2002

2013

1984

2002

2013

Forest

6017

9095

9793

30.50

44.75

46.86

197

303

308

9.85

14.78

15.38

Agriculture

30,247

45,476

48,964

91.60

134

141

789

1213

1232

9.85

14.78

15.38

Grassland

1714

3071

2880

6.44

7.25

8.64

45.72

81.89

76.80

0.57

1.02

0.96

Urbanization

85,700

128,846

138,739

132

193

203

1433

1819

1848

99

148

154

Cleared land

30,247

30,583

48,964

91.60

134

141

740

1213

1232

9.85

14.78

15.38

Secondary forest

6017

9095

9793

30.50

44.75

46.86

197

303

308

9.85

14.78

15.38

Mining

60,493

90,953

97,933

137

182

185

Total

220,435

317,119

357,066

383

558

587

3539

5115

5190

139

208

216

4.4.4 AN

AN load was increased with changes in LULC condition where higher loads were observed during 1984–2013 (Fig. 7). In Galas and Pergau (Fig. 7c), 2013 LULC condition recorded the highest load of 89.89 and 174 kg year−1, respectively. Although Galas has the higher percentage of agriculture (which is the major source of NPS in the area) compared to Pergau, higher runoff activities were recorded in Pergau due to the presence of more forested areas in Galas. In Lebir (Fig. 7), about 2% increase in AN was observed between 1984 and 2002 and 37% was observed from 2002 to 2013. Whereas, in Nenggiri (Fig. 7), huge difference was recorded from both 1984–2002 and 2002–2013 LULC (52 and 56%, respectively). This large difference may be attributed to inconsistency in the runoff volume in the area.

4.5 NPS Pollutant Load Variation Among Different LULC Changes

The results of NPS Loads according to LULC changes are presented in Tables 1, 2, 3 and 4. LULC types in Kelantan river basin were classified into forest, paddy, agriculture, grassland, urbanization, cleared land, mangrove swamp, secondary forest, rivers, ponds and lakes and mining for ease of analysis as well as for increase in precision of the results. NPS pollutant loads among different LULC changes also increases with changes in LULC condition. In Galas (Table 1), estimated TSS load under 1984 LULC condition outweigh that of 2002 LULC condition with total values of 106,959.40 and 118,796.80 kg year−1, respectively. In addition, TSS loads under 2013 LULC condition was recorded to supersede that of 2002 LULC condition. In Pergau total estimated TSS ranged from 189,335.50 to 289,180.7 kg year−1 from 1984 to 2013. In all the catchments, urbanization was recorded to give the highest supply of TSS under all LULC conditions. In Lebir and Nenggiri (Tables 3 and 4), urbanization recorded the highest TSS loads of 121,551.70 and 85,700.35 kg year−1 under 1984 LULC condition, while forest recorded the lowest in both catchments (8580.12 and 6016.74 kg year−1). Higher values of TSS recorded in urbanized areas may be due to large runoff events caused by impervious surfaces. The relative changes of NPS in Kelantan river basin are not only governed by the type of LULC and nature of the pollutant load but also by the amount of annual rainfall high enough to cause runoff. This is why changes in pollutants loads from 1984 to 2002 and from 2002 to 2013 were not as notable as the LULC changes. For instance, from 1984 to 2002 where massive deforestation of 45% was observed in Galas, these changes were not observed to reflect in the amount of TSS (about 11%) recorded in that year. This may be because only the average rainfall in 1984 was used in computing runoff under 1984 LULC condition. (Fig. S3). Whereas, for 2002 and 2013 LULC conditions, the average values during 1984–2002 and 1984–2013 were used, respectively.

The temporal distribution of TP loads with regards to LULC changes was also found to be regular with changes in LULC condition (Tables 1, 2, 3 and 4). Urbanization was recorded as the LULC change with the highest TP load in all the catchments. Even though agriculture is expected to have a significant contribution of TP loads due to addition from fertilizers and animal manure, higher runoff activities are more likely to occur in urbanized land compared to agricultural land. Thus, making TP loads higher in all catchments in this study. TP loads in Galas (Table 1) ranged from 829 to 926 kg year−1 from 1984 to 2013 with urbanization recording the highest load and forest recording the lowest load. Higher TP loads were recorded in Lebir more than all the catchments with estimates of 829, 679 and 926 kg year−1 for 1984, 2002 and 2013 LULC conditions, respectively.

TN and AN (Tables 1, 2, 3 and 4) have increased with changes in LULC condition as seen in other pollutants. Similarly, TP urbanization was found to be the LULC change with the highest contribution to both TN and AN load. Even though, agriculture has been reported as one of the major contributors of TN as well as AN load in most NPS studies. The reason for this is that, higher runoff activities were recorded from 1984 to 2013 LULC condition in urbanized areas and, therefore, more NPS loads (TN and AN) are likely to be transferred under these land uses compared to others. From 1984 to 2002 an increase in TN load of 15% was recoded in Galas and 24% in Pergau, while from 2002 to 2013 an increase of 13% and 23% for Galas and Pergau, respectively. Total TN values in Lebir ranged from 6114.03 to 8297.87 kg year−1 from 1984 to 2013 and from 3539.47 to 5190.74 kg year−1 in Nenggiri. Total AN load for the 2013 LULC which recorded the highest compared to other LULC conditions in all the catchments are 89.89 kg year−1 for Galas, 174 kg year−1 for Pergau, 302 kg year−1 for Lebir and 216 kg year−1 for Nenggiri.

4.6 NPS Pollutant Load Relationships Among Different LULC Conditions

4.6.1 Galas

Results of the NPS pollutant loads relationships among sub-basins and LULC conditions in Galas and Pergau are shown in Table 5. The comparison was done between sub-basins W60, W70, W80, W90 and W100 and 1984, 2002 and 2013 LULC conditions. In Galas, sub-basin W80 (26947a) under 2013 LULC condition was ranked statistically most significant (p < 0.05) when compared to other sub-basins and LULC conditions with regards to TSS. While sub-basin W70 (21285k) under 1984 LULC condition was classified as least significant using the same comparison. Other sub-basins such as W80 (21438i), W90 (21,437i) and W100 (21438i) all under 1984 LULC conditions were grouped as statistically the same by the Tukey’s range test. The results obtained from TSS, TP, TN and AN all indicated that sub-basin W80 under 2013 LULC as the statistically most significant (p < 0.05) when compared to other sub-basins and other LULC conditions. The statistically most significant results obtained from TP, TN and AN are 66.39a, 587.58a and 19.06a, respectively. Mean separation using LSD carried out on TP grouped some sub-basins under different LULC conditions are statistically the same, for example, W80 (52.88e), W90 (52.92e), W100 (52.88e) all under 1984 LULC conditions and W70 (53.30e) under 2002 LULC condition. Whereas under AN statistical grouping was done up to 2 and 3 orders, for example, sub-basins W60 (17.55ab) and W90 (17.82ab) all under 2013 LULC condition were ranked as the most significant and at the second most significant. While sub-basin W90 (15.28cd) under 1984 LULC condition and sub-basin W60 (15.38cd) under 2002 LULC condition were ranked with two same orders even though from different LULC conditions. This makes them statistically similar to other pollutant loads with the same letter in that group. Another noticeable example is where sub-basins W90 (16.30bcd) and W100 (16.45bcd) all under 2002 LULC condition were ranked up to three orders. The statistically most significant pollutant loads obtained from 2013 LULC conditions and that of statistically least significant obtained under 1984 LULC condition is a clear indication of the effect LULC changes on NPS pollution.
Table 5

NPS pollutant loads relationships among sub-basins and LULC conditions in Galas and Pergau

 

Year

Sub-basin

W60

W70

W80

W90

W100

Galas

 TSS (kg year−1)

1984

21361j

21285k

21438i

21,437 i

21438i

2002

23734g

23713h

23797e

23777f

23777f

2013

26891c

26851d

26947a

26891c

26919b

 TP (kg year−1)

1984

46.17f

39.50g

52.88e

52.92e

52.88e

2002

53.30e

51.66e

58.53d

56.83d

56.93d

2013

61.58c

58.19d

66.39a

61.57c

63.95b

 TN (kg year−1)

1984

409.12j

350.43 k

452.21i

452.19i

451.99i

2002

470.73g

457.21 h

510.39e

497.92f

497.89f

2013

544.62c

514.04d

587.58a

544.14c

566.07b

 AN (kg year−1)

1984

13.40e

14.78de

15.43cd

15.28 cd

15.01cde

2002

15.38cd

14.78d

16.64bc

16.30bcd

16.45bcd

2013

17.55ab

16.64bc

19.06a

17.82ab

18.67a

 

Year

Sub-basin

W300

W310

W320

W330

W340

W350

W360

W390

Pergau

 TSS (kg year−1)

1984

24,871j

24,871i

23,147k

24,871i

23,147k

23,147k

21,565l

24,871i

 

2002

28,838g

29,823f

29,823f

30,644e

28,838g

27,697h

29,823f

28,838g

 

2013

36,660b

36,660b

35,942c

37,235a

35,942c

36,660b

34,144d

35,942c

 TP (kg year−1)

1984

132.08j

139.42i

129.55k

139.67i

129.56k

129.62k

120.78l

139.50i

2002

161.82g

167.16f

166.98f

171.48e

161.63g

154.78h

167.22f

161.05g

2013

205.56b

205.53b

201.15c

208.69a

200.81c

204.94b

191.09d

201.47c

 TN (kg year−1)

1984

429.67j

450.54i

419.01k

450.57i

419.06k

419.44k

390.70l

450.44i

2002

522.35g

539.86f

540.26f

553.21e

522.48g

501.46h

539.63f

522.35g

2013

663.93b

663.43b

650.76c

674.42a

650.32c

663.32b

618.23d

650.64c

 AN (kg year−1)

1984

14.58de

15.30d

14.87de

15.34d

14.11de

14.13de

13.43e

15.10de

2002

17.77c

18.64c

18.63c

18.85c

18.07c

18.21c

18.91c

18.16c

2013

22.31ab

22.40ab

22.19ab

23.16a

22.01ab

22.38ab

20.70b

22.03ab

Least significant means with the same letter are not significantly different at p < 0.05

4.6.2 Pergau

ANOVA was conducted for the comparison of NPS pollutant loads among the three LULC conditions (1984, 2002 and 2013) and eight sub-basins (W300, W310, W320, W330, W340, W350, W360 and W390) in Pergau and are shown in Table 5. For all the pollutant loads in this catchment, sub-basin W330 under 2013 LULC condition and sub-basin W360 were ranked as the statistically most significant and least significant (p < 0.05) when compared with other sub-basins and other LULC conditions. Pollutants loads that were ranked statistically most significant are 37235a for TSS, 208.69a for TP, 674.42a for TN and 23.16a for AN. While pollutant loads that were grouped as statistically least significant (p < 0.05) for TSS, TP, TN and AN are 21565l, 120.78l, 390.70l and 13.43e, respectively. For both TSS, TP and AN, the mean separation carried out using LSD grouped the pollutant loads at just one order. While for AN, grouping was done for up to two orders, for example, most of the sub-basins under 1984 and 2013 LULC conditions were ranked for up to two orders except for W310 (15.30d), W330 (15.34d) and W360 (13.43e) under 1984 LULC condition. While under 2013 LULC condition W330 (23.16a) and W360 (20.70b) were grouped with one order while the rest were grouped with two orders. The statistically most significant (p < 0.05) pollutant loads obtained under 2013 LULC condition and statistically least significant (p < 0.05) pollutant loads obtained under 1984 LULC condition clearly indicated the influence of LULC change on NPS pollution. The 1984 LULC condition that is characterized with low runoff due to high percentage of forested areas will result to build up of low NPS pollutant loads. While the reverse is the case for the 2013 LULC condition which has undergone deforestation mostly agricultural activities, illegal logging and urbanization will result to high NPS pollution due to high rate of runoff.

4.6.3 Lebir

Results of NPS pollutant loads relationships among LULC conditions and sub-basins in Lebir are shown in Table 6. The comparison involved three LULC conditions (1984, 2002 and 2013) and fifteen sub-basins. In all the pollutant loads, 2013 LULC conditions was found to be statistically the most significant (p < 0.05) when compared to other sub-basins and other LULC conditions. Although the sub-basins involved are not the same for all the pollutants, this could be due to LULC changes from one sub-basin to another that may have altered runoff which in turn influences pollutant load. Under TSS sub-basin W220 (512339a) was statistically the most significant (p < 0.05), while for both TP, TN and AN sub-basins W160, W180, W220, W250 and W260 were grouped as statistically most significant (p < 0.05) compared to other sub-basins and LULC conditions. In the case of statistically least significance (p < 0.05), sub-basin W240 under 1984 LULC conditions was recorded for both TSS (19450l), TP (46.31h), TN (324.91k) and AN (13.03qr). Ranking of means with regards to statistical significance was done using one order for both TSS and TN where means with the same letters were grouped as statistically the same. While for TP ranking of means was done for up two orders, for example; sub-basin W190 (53.53ef) under 1984 LULC condition and for three orders; sub-basin W160 (51.90efg) under 2013 LULC condition. A more complex ranking was obtained from AN pollutant where ranking was done for up to four orders; sub-basin W160 (19.13efgh) and up to five orders; sub-basin W180 (16.48ijklm) all under 1984 LULC condition. This complex ranking is due to statistical similarities involving the pollutant loads where any two or more means carrying the same letters are grouped as statistically the same. From the result, it could be observed that, 2013 LULC condition was the statistically most significant (p < 0.05) compared to 2002 and 1984 LULC condition. Thus, indicating the influence of LULC changes on NPS pollution in the watershed.
Table 6

NPS pollutant loads relationships among LULC conditions and sub-basins in Lebir

 

Year

Sub-basin

W160

W170

W180

W190

W200

W210

W220

W230

W240

W250

W260

W270

W280

W290

W300

TSS (kg year−1)

1984

28,860e

22,586i

24,468h

22,586i

24,468h

24,468h

22,586i

22,586i

19,450l

24,468h

24,468h

28,860e

19,449l

24,469h

19,449l

2002

27,543f

24,594g

27,544f

24,594g

27,543f

24,594g

24,594g

22,273j

22,273j

27,543f

27,544f

22,273j

22,273j

22,273j

17,066m

2013

39,338b

33,822c

39,338b

30,735d

33,822c

33,822c

5,12,339a

30,735d

30,735d

39,338b

39,338b

30,735d

21,282 k

33,822c

2128k

TP (kg year−1)

1984

68.04b

53.40efg

57.73d

53.53ef

58.26d

57.97d

53.76e

53.36efg

46.31h

57.60d

58.00d

68.08b

46.50h

58.00d

46.36h

2002

51.90efg

46.43h

51.54 g

46.46h

51.55 g

46.12h

46.16h

41.96ij

42.29i

51.91efg

51.82 fg

41.76ij

41.90ij

42.13ij

32.17k

2013

73.76a

63.58c

73.76a

57.74d

63.45c

63.29c

73.73a

57.87d

57.91d

73.47a

73.84a

57.63d

40.31j

63.27c

40.34j

TN (kg year−1)

1984

399.41g

377.20h

481.71d

377.19 h

481.87d

481.65d

376.96h

481.91d

324.91k

481.68d

324.95 k

399.39 g

325.26 k

481.84d

324.94k

2002

371.88i

410.57f

459.91e

410.91f

459.75e

410.65f

410.77f

460.05e

372.20i

459.71e

459.91e

372.30i

372.07f

371.92i

285.28l

2013

656.63a

564.36b

656.51a

512.88c

565.03b

564.70b

656.71a

512.85c

512.96c

656.48a

656.76a

513.16c

355.65j

564.47b

355.44j

AN (kg year−1)

1984

19.13efgh

15.87mno

16.48ijklm

16.11mno

16.37jklmn

16.17mno

15.36mno

15.04mnop

13.03qr

16.53ijklm

16.28klmno

18.91efgh

13.17qpr

16.48ijklm

13.03qr

2002

16.58ijklm

16.16mno

18.21hijkl

16.51ijklm

18.31ghij

16.26klmno

16.23lmno

15.17mnop

15.05mnop

18.44fghi

18.23ghijk

14.91mnopq

15.08mnop

14.85mnopq

11.68r

2013

25.76a

22.51b

26.22a

20.22defg

22.47b

22.09bcd

25.89a

20.60bcde

20.38cdef

25.90a

26.11a

20.46cde

14.44nopq

22.34bc

14.34opq

Least significant means with the same letter are not significantly different at p < 0.05

4.6.4 Nenggiri

NPS pollutant loads relationships were compared among three LULC conditions and different sub-basins in Nenggiri and the results are presented in Table 7. The results indicated for TSS pollutants sub-basin W200 (26225a) under 2002 LULC condition was ranked as the statistically most significant (p < 0.05) compared to other sub-basins and LULC conditions. While for TP, sub-basins W200 (49.30a), W280 (49.20a), W290 (49.42a) and W320 (49.65a) all under 2002 LULC condition were grouped as the statistically most significant (p < 0.05) sub-basins when compared to other sub-basins and LULC conditions. For TN, sub-basins W200 (438.05a) and W220 (438.05a) all under 2002 LULC condition were ranked as the statistically most significant (p < 0.05) compared to other sub-basins and LULC conditions. For AN loads, sub-basins W280 (17.21a) was found to be the statistically most significant (p < 0.05) compared to other sub-basins and LULC conditions. Although, other sub-basins such as, W270 (16.92ab) and W290 (16.99ab) under 2002 LULC condition and sub-basins W280 (15.17abcdefg) and W290 (15.23abcdefg) under 2013 LULC condition are grouped as statistically similar to W280, because they carry the same letter in their ranking. The ranking is therefore not limited to W280 alone, but to all sub-basins with similar letters used for ranking in all the sub-basins and across the LULC conditions. For all the pollutant loads in Nenggiri, sub-basin W180 under 1984 LULC condition was grouped as the statistically least significant (p < 0.05) when compared with other sub-basins and LULC conditions. Mean values of pollutant loads for sub-basin W180 are as follows; TSS (10853 l), TP (20.20i), TN (181.82m) and AN (7.02n). Mean separation carried out in this catchment ranked some pollutants up to just one order such as what is obtained in TSS and TN. While a more complex way of ranking was observe for AN where means were grouped up to 7 orders, for example; W280 (15.17abcdefg) and W290 (15.23abcdefg) all under 2013 LULC condition. This complex way of ranking reveals the statistical similarities existing among the means carrying the same letters. It can be inferred from the result that NPS pollution in Nenggiri is influenced by long-term LULC changes. Although, 2002 LULC condition and not 2013 LULC condition was reported to observe the statistically most significant (p < 0.05) sub-basin unlike in other catchments. High percentage of forested areas were observed under 2013 LULC condition compared to 2002 LULC condition. Therefore, 2002 LULC condition will likely to favor large runoff which in turns carries along with it NPS pollutants as compare to 2013 LULC condition.
Table 7

NPS pollutant loads relationships among LULC conditions and sub-basins in Nenggiri

 

Year

Sub-basin

W180

W190

W200

W210

W220

W230

W240

W250

W260

TSS (kg year−1)

1984

10,853l

14,117k

16,313j

14,117k

14,117k

16,313j

16,313j

16,313j

14,117k

2002

17,127i

17,127i

26,225a

17,127i

19,770g

19,770 g

17,127i

19,770g

19,770g

2013

23,715d

20,492f

25,536b

25,535b

25,535b

22,049e

23,715d

20,492f

20,491f

TP (kg year−1)

1984

20.20i

26.78h

31.18fg

31.07g

31.20efg

20.97i

30.91g

31.10g

27.21h

2002

32.61efg

32.59efg

49.30a

32.49efg

41.82c

41.64c

32.77efg

37.57d

41.37c

2013

44.95b

38.85d

48.28a

42.02c

41.74c

41.60c

44.53b

38.90d

38.75d

TN (kg year−1)

1984

181.82m

255.18k

292.82i

255.37

254.93 k

181.61m

293.07i

292.81i

255.23k

2002

285.95j

286.11j

438.05a

286.01j

438.05a

366.42fg

286.12j

365.85g

365.63g

2013

342.42h

342.04h

426.70b

342.37h

368.51ef

368.90e

342.34h

368.84e

368.61e

AN (kg year−1)

1984

7.02n

9.30jklm

10.40jkl

9.14klmn

9.18jklmn

7.28nm

10.88ijkl

10.59jkl

9.52jklm

2002

14.10cdefg

10.66jkl

11.12hijkl

11.27hijk

14.20cdefg

13.87efg

10.68lkj

13.74gf

11.00hijkl

2013

13.25fgh

13.19fgh

16.09abcde

13.16ghf

14.09cdefg

14.15cdefg

13.01ghi

14.22cdefg

13.91efg

 

Year

Sub-basin

W270

W280

W290

W300

W310

W320

W330

W340

TSS (kg year−1)

1984

14,117k

14,117k

14,117k

14,117k

14,117k

17,596h

14,117k

14,117k

2002

25,284c

25284c

25,284c

19,769g

19,770g

25,284c

19,770g

25,284c

2013

25,535b

25,536b

25,536b

25,535b

20,492f

25,536b

23,715d

23,715d

TP (kg year−1)

1984

27.16h

27.21h

27.03h

27.07h

26.90h

33.48e

26.79h

33.40ef

2002

49.79a

49.20a

49.42a

37.56d

41.69c

49.65a

37.85d

37.84d

2013

48.34a

41.39c

41.60c

41.67c

38.40d

47.94a

45.09b

44.97b

TN (kg year−1)

1984

255.33k

254.74k

255.18k

254.90k

255.10k

246.95 l

254.72k

254.82k

2002

422.14c

422.23c

422.37c

365.71g

366.05g

422.09c

366.04g

422.43c

2013

396.48d

395.85d

396.21d

368.72e

368.25ef

396.32d

368.71e

396.06d

AN (kg year−1)

1984

9.60jkl

9.22jklmn

9.57jkl

9.27jklmn

9.13klmn

11.45hij

8.98lmn

9.11klmn

2002

16.92ab

17.21a

16.99ab

13.74gf

14.03cdefg

16.21abcd

13.96defg

16.28abc

2013

14.85bcdefg

15.17abcdefg

15.23abcdefg

13.79gf

14.19cdefg

15.29abcdef

14.02cdefg

14.98abcdefg

Least significant means with the same letter are not significantly different at p < 0.05

5 Conclusion

With the aid of GIS tool, numeric integration was carried out to estimate spatio-temporal changes on NPS in Kelantan river basin from 1984 to 2013. The following conclusions were drawn. All the pollutant loads estimated in this study were observed to be affected by LULC changes, runoff and EMC values. The extensive LULC change mostly involves deforestation for logging and agricultural purposes. In all the catchments, pollutant loads mostly increase with changes in LULC condition as one moves from 1984 to 2013. NPS pollutant loads among different LULC changes also increase with changes in LULC condition from 1984 to 2013. Urbanization was found to be the dominant LULC change with the highest pollutant load in all the catchments. Higher values of pollutant loads recorded in urbanized areas may be due to large runoff events caused by huge impervious surfaces which when combined with EMC values give figures that are far above those of other LULC changes given rise to high-level TSS in urbanized places.

Analysis of variance (ANOVA) was conducted for the comparison of NPS pollutant loads among the three LULC conditions used and the sub-basins in each catchment in this study for the entire watershed. Mean separation was carried out using least significance difference (LSD) and significant means were grouped using Tukey’s range test. The results revealed that 2013 LULC condition was the statistically most significant (p < 0.05) LULC condition in Galas, Pergau and Lebir, while in Nenggiri it was found to be the 2002 LULC condition. In all the catchments 1984 LULC condition was found to be the statistically least significant (p < 0.05) LULC condition compared to other LULC conditions. The findings of this study will of serious help to water resource planners in controlling water pollution for future planning.

Notes

Acknowledgements

This research was funded by the Fundamental Research Grant Scheme (FRGS) 2015-1 from the Ministry of Higher Education (MOHE), Malaysia. The authors wish to thank Department of Irrigation and Drainage (DID) Ampang, Malaysia for the hydrological data and Department of Agriculture, Malaysia for the soil map as well as land use maps.

Compliance with Ethical Standards

Conflict of interest

On behalf of all authors, the corresponding author states that there is no conflict of interest. Authors further declare that there is no financial or personal relationship with a third party whose interests could be positively or negatively influenced by this article’s content.

Supplementary material

41748_2018_42_MOESM1_ESM.docx (1.6 mb)
Supplementary material 1 (DOCX 1641 kb)

References

  1. Adams BJ, Papa F (2001) Urban storm water management planning with analytical probabilistic models. Can J Civ Eng 28:545CrossRefGoogle Scholar
  2. Adnan NA, Atkinson PM (2011) Exploring the impact of climate and land use changes on streamflow trends in a monsoon catchment. Int J Climatol 31:815–831.  https://doi.org/10.1002/joc.2112 CrossRefGoogle Scholar
  3. Bai X, Ma KM, Yang L, Zhang XL (2008) Simulating the impacts of land-use changes on non-point source pollution in Lugu Lake watershed. Int J Sust Dev World 15:18–27.  https://doi.org/10.1080/13504500809469764 CrossRefGoogle Scholar
  4. Bhaduri B, Harbor J, Engel B, Grove M (2000) Assessing watershed-scale, long-term hydrologic impacts of land-use change using a GIS-NPS model. J Environ Manag 26:643–658CrossRefGoogle Scholar
  5. Chow MF, Yusop Z (2006) Storm runoff quality in a residential catchment in Malaysia. J Environ Hydrol 14:1–7Google Scholar
  6. Chow MF, Yusop Z, Mohamed M (2011) Quality and first flush analysis of storm water runoff from a tropical commercial catchment. Water Sci Technol 63:1211–1216.  https://doi.org/10.2166/wst.2011.360 CrossRefGoogle Scholar
  7. DID (Drainage and Irrigation Department) (2000) Annual flood report of DID for Peninsular Malaysia. Kuala LumpurGoogle Scholar
  8. DID (Drainage and Irrigation Department) (2012) Urban Stormwater Management Manual for MalaysiaGoogle Scholar
  9. Eisakhani M, Pauzi A, Karim O, Malakahmad A, Kutty SRM, Isa MH (2009) GIS-based non-point Sources of pollution simulation in Cameron Highlands, Malaysia. Int J Civ Environ Eng 3:3–7Google Scholar
  10. Engel B (2001) L-THIA NPS Long-Term Hydrologic Impact Assessment and Non-Point Source Pollutant Model, version 2.1. Purdue University and US Environmental Protection AgencyGoogle Scholar
  11. Engel BA, Choi JY, Harbor J, Pandey S (2003) Web-based DSS for hydrologic impact evaluation of small watershed land use changes. Comput Electron Agric 39:241–249CrossRefGoogle Scholar
  12. Fohrer N, Haverkamp S, Eckhardt K, Frede HG (2001) Hydrologic response to land use changes on the catchment scale. Phys Chem Earth Pt B 26:577–582.  https://doi.org/10.1016/S1464-1909(01)00052-1 CrossRefGoogle Scholar
  13. Girmay G, Singh BR, Nyssen J, Borrosen T (2009) Runoff and sediment-associated nutrient losses under different land uses in Tigray. Northern Ethiopia. J Hydrol. 376:70–80.  https://doi.org/10.1016/j.jhydrol.2009.07.066 CrossRefGoogle Scholar
  14. Huber WC (1993) Contaminant transport in surface water. In: Handbook of hydrology. Springer, Dordrecht, pp 11–14Google Scholar
  15. Jamaliah J (2007) Emerging Trends of Urbanization in Malaysia. J Dept Stat, Malaysia. 1:43–54. http://www.statistics.gov.my/eng/images/stories/files/journalDOSM/V104ArticleJamaliah.pdf
  16. Li T, Bai F, Han P, Zhang Y (2016) Non-point source pollutant load variation in rapid urbanization areas by remote sensing, GIS and the L-THIA Model: a Case in Bao’an District, Shenzhen, China. Environ Manag 58:873–888.  https://doi.org/10.1007/s00267-016-0743-x CrossRefGoogle Scholar
  17. Meals DW, Richards RP, Steven AD (2013) Pollutant load estimation for water quality monitoring projects. National Nonpoint Source Monitoring Program, pp. 1–21Google Scholar
  18. Moriasi DN, Arnold JG, Van Liew MW, Bingner RL, Harmel RD, Veith TL (2007) Model evaluation guidelines for systematic quantification of accuracy in watershed simulations. Trans ASABE 50:885–900CrossRefGoogle Scholar
  19. Nash JE, Sutcliffe JV (1970) River flow forecasting through conceptual models part I—a discussion of principles. J Hydrol 10:282–290.  https://doi.org/10.1016/0022-1694(70)90255-6 CrossRefGoogle Scholar
  20. Nazahiyah R, Yusop Z, Abustan I (2007) Storm water quality and pollution loading from an urban residential catchment in Johor, Malaysia. Water Sci Technol 56:1–9.  https://doi.org/10.2166/wst.2007.692 CrossRefGoogle Scholar
  21. Novotny V (2003) Water quality: diffuse pollution and watershed management. Wiley, HobokenGoogle Scholar
  22. Ozdemir H, Elbaşı E (2015) Benchmarking land use change impacts on direct runoff in ungauged urban watersheds. Phys Chem Earth Pt A/B/C 79–82:100–107.  https://doi.org/10.1016/j.pce.2014.08.001 CrossRefGoogle Scholar
  23. Rossi L, Krejci V, Rauch W, Kreikenbaum S, Fankhauser R, Gujer W (2005) Stochastic modeling of total suspended solids (TSS) in urban areas during rain events. Water Res 39:4188–4196.  https://doi.org/10.1016/j.watres.2005.07.041 CrossRefGoogle Scholar
  24. Shen Z, Hong Q, Yu H, Niu J (2010) Parameter uncertainty analysis of non-point source pollution from different land use types. Sci Total Environ 408:1971–1978.  https://doi.org/10.1016/j.scitotenv.2009.12.007 CrossRefGoogle Scholar
  25. Shen Z, Liao Q, Hong Q, Gong Y (2012) An overview of research on agricultural non-point source pollution modelling in China. Sep Purif Technol 84:104–111.  https://doi.org/10.1016/j.seppur.2011.01.018 CrossRefGoogle Scholar
  26. Thanapakpawin P, Richey J, Thomas D, Rodda S, Campbell B, Logsdon M (2007) Effects of landuse change on the hydrologic regime of the Mae Chaem river basin, (NW) Thailand. J Hydrol 334:215–230.  https://doi.org/10.1016/j.jhydrol.2006.10.012 CrossRefGoogle Scholar
  27. Tong STY, Chen W (2002) Modeling the relationship between land use and surface water quality. J Environ Manag 66:377–393.  https://doi.org/10.1006/jema.2002.0593 CrossRefGoogle Scholar
  28. United States Environmental Protection Agency (2009) National Water Quality Inventory: report to Congress. Water, 1:43. http://www.epa.gov/owow/305b/2004report/
  29. Wan I (1996) Urban growth determinants for the state of Kelantano of the state’s policy makers. Penerbitan Akademik Fakulti Kejuruteraan dan Sains Geoinformasi. Buletin Ukur 7:176–189Google Scholar
  30. Wang S, Wang S (2013) Land use/land cover change and their effects on landscape patterns in the Yanqi Basin, Xinjiang (China). Environ Monit Assess 185:9729–9742.  https://doi.org/10.1007/s10661-013-3286-0 CrossRefGoogle Scholar
  31. Yusop Z, Tan LW, Ujang Z, Mohamed M, Nasir KA (2005) Runoff quality and pollution loadings from a tropical urban catchment. Water Sci Technol 52:125–132Google Scholar
  32. Zaimes GN, Schultz RC, Isenhart TM (2008) Total phosphorus concentrations and compaction in riparian areas under different riparian land-uses of Iowa. Agric Ecosyst Environ 127:22–30.  https://doi.org/10.1016/j.agee.2008.02.008 CrossRefGoogle Scholar
  33. Zhan X, Huang ML (2004) ArcCN-runoff: an ArcGIS tool for generating curve number and runoff maps. Environ Model Softw 19:875–879.  https://doi.org/10.1016/j.envsoft.2004.03.001 CrossRefGoogle Scholar
  34. Zhang J, Shen T, Liu M, Wan Y, Liu J, Li J (2011) Research on non-point source pollution spatial distribution of Qingdao based on L-THIA model. Math Comput Model 54:1151–1159.  https://doi.org/10.1016/j.mcm.2010.11.048 CrossRefGoogle Scholar

Copyright information

© Springer International Publishing AG, part of Springer Nature 2018

Authors and Affiliations

  1. 1.Department of Environmental Science, Faculty of Environmental StudiesUniversiti Putra MalaysiaSerdangMalaysia
  2. 2.Department of Soil Science, Institute for Agricultural Research/Faculty of AgricultureAhmadu Bello UniversityZariaNigeria
  3. 3.School of Systems, Management and Leadership, Faculty of Engineering and Information TechnologyUniversity of Technology SydneySydneyAustralia

Personalised recommendations