Assessment of Water and Soil Contamination and Land Cover Changes in the Spring Creek Bayou Watershed in Houston, Texas

Abstract

Stormwater runoff and nutrient pollution are significant sources of water contamination that continue to grow in rural and suburban watersheds. The goal of this research is to analyze and evaluate the impact of urbanization and industrialization on suburban watersheds in southeast Texas. The objectives are to: (1) determine nutrient and heavy metal concentrations in soil and water samples along Spring Creek Bayou (SC), (2) analyze land cover changes over the last 30 years and (3) assess and evaluate socio-economic data within the watershed. The soil and water samples were collected from upstream, midstream and downstream locations in triplicate during the spring and fall seasons along the bayou. The samples were analyzed to determine chemical concentrations and Landsat 5, and eight imageries were used to derive thematic land cover maps. The soil and water chemical concentrations were interpolated to spatial maps for distribution analysis. The chemical analysis of water samples collected from SC Bayou revealed that N and P concentrations were at elevated levels that can pose a threat to water quality and aquatic organisms. Heavy metal concentrations of Zn were at elevated levels in water samples from the SC Bayou watershed. Land cover change patterns showed that high-vegetation surfaces decreased while low-vegetation surfaces increased slightly over the past three decades. The watershed experienced an increase in total population from 129,629 residents in 1990 to 389,977 residents in 2020. This research is important in improving our understanding on the impact of natural and human activities on suburban watersheds in the Greater Houston metropolitan region.

1. Introduction

The urbanization rate in Texas has steadily increased over the last three decades. In 1980, approximately 79.7% of the total population lived in urban areas in the United States [1]. By 2020, the total urban population had increased to 88.4% [2]. As urban centers increase in size, human populations and community assets become increasingly concentrated in areas that are vulnerable to environmental hazards. Frequent flooding events in the Greater Houston region have become a major hazard resulting in extensive property damage and loss of life, especially in densely populated areas [3,4,5]. According to the Federal Emergency Management Agency (FEMA), the Greater Houston region has experienced 12 major flood events, five hurricanes, and five tropical storms since 1980 [6]. Extensive research has been conducted on the relationships between land use and land cover changes at various scales and environmental hazards [7,8]. However, fewer comprehensive studies have been conducted to test the direct effects of land use and land cover changes on environmental pollution, such as nutrient and heavy metal contamination.

Heavy metals, including Pb, Cd, Hg, and Cr, and the metalloid arsenic (As) are listed as priority pollutants by the Environmental Protection Agency (EPA) [9]. These metals and metalloids are released into the environment through industrial and domestic waste, or through acidic rain resulting in the leaching of metals from soil, and then enter streams, rivers, lakes and groundwater [3,10]. It is, therefore, important to monitor and assess the concentrations of potentially toxic heavy metals and metalloids in the various environmental media of soil, water and air.

Land cover is important in protecting soil and water quality in urban watersheds. Land cover change (LCC) can influence environmental conditions such as water quality and hydrology, infiltration rate, pollutant transport, surface runoff patterns and flood buffering in watersheds [11]. Anthropogenic changes have already transformed more than 75% of the Earth’s surface through expansion in urban and agricultural land use, resulting in significant losses of natural land cover and vegetation [12,13,14]. Despite some net increase in forested land and wetland areas in some regions, the urbanization trend will continue to expand into the future, with significant impacts on several natural ecosystems [15,16].

Impervious surfaces such as paved roads, parking lots, roofs, and compacted soils prevent rainwater infiltration, which causes excess stormwater runoff, higher stream flow rates, and poor water quality [17]. Higher proportions of impervious surface in urban regions result in reduced infiltration, shorter residence time, increased overland flow rates, and reduced baseflow [14,18]. This results in increased impacts on water pollution rates in urban areas. The reduced baseflow of streams in the urban environment usually increases the relative concentration of pollutants in the water [19,20,21].

Surface runoff has cumulative effects on humans and the environment including flooding, streambank erosion, public safety, and increased cost of water and wastewater treatment [22]. Surface runoff also transports different pollutants that are found on paved surfaces, such as sediment, nitrogen, phosphorus, bacteria, pesticides, and metals [4,5,23]. However, vegetated surfaces play a vital role in the management of storm water runoff by intercepting precipitation and reducing the volume and rate of stormwater runoff. Vegetation also reduces erosion and the occurrence of flash flooding. The increased fragmentation of natural vegetation in urban regions in recent years is reducing its ability to intercept, assimilate, and reduce water pollution [24,25]. Therefore, assessing land cover change is crucial in understanding the changes and interactions between human activities and natural phenomena within watersheds [26]. This research aims to analyze the effect of land cover change on environmental pollution using GIS and remote sensing technologies. The specific objectives are to: (1) determine nutrient and heavy metal concentrations in soil and water samples along Spring Creek Bayou (SC), (2) analyze land cover changes over the last three decades and (3) evaluate socio-economic characteristics within the watershed.

2. Methodology

2.1. Description of Study Areas

The Spring Creek watershed (SCW) is a suburban watershed located in northern Harris County and extending across Montgomery, Waller and Grimes counties in southeast Texas as shown in Figure 1. Spring Creek is its primary stream, which flows east into the San Jacinto River and ultimately into Lake Houston [27]. This is a relatively large watershed covering approximately 997.0 km². The SCW is mostly undeveloped with large heavily wooded areas and several suburban communities scattered throughout. The undeveloped conditions and natural features of the watershed provides a diverse ecosystem and an expansive floodplain, making the region highly environmentally sensitive. The western region of the watershed is primarily rural, the central region is slowly transitioning into urban, and the eastern region is densely suburban/urban [28].

2.2. Data Collection and Analysis

Water and soil samples were collected in the fall and spring of 2020 and 2021, respectively, from three different sampling locations along Spring Creek Bayou. The soil and water samples were collected in triplicate and given an abbreviated name identified with the name of the bayou followed by the distance (given in km) from the mouth of the bayou as follows: SC3.4, SC45.2, SC88.6. The geographic positions were recorded using a portable Global Positioning System (GPS) device for each sampling location. Sample preparation and analysis was conducted by the following procedures described in EPA methods 3051A and 3015A [29]. Total C and N concentrations were analyzed using the Total Carbon and Nitrogen (TCN) Analyzer [5]. Results were compared with the Guidance for Assessing and Reporting Surface Water Quality in Texas [30] and Nutrient Criteria of Rivers and Streams [31] to identify elements of concern.

Remote sensing imagery was used to detect land cover changes in the SCW over the last three decades. Landsat satellite imagery was obtained from the USGS Global Visualization Viewer (GloVis) for the period from 1984 to 2020 [32,33]. The attributes and access periods for the satellite imagery are summarized in

Table 1. Landsat images used in this study.

Sensor	File Name	Acquisition Date	Path/Row
TM	LT50250391984342XXX02	7 December 1984	25/39
TM	LT50260391984333XXX08	28 November1984	26/39
TM	LT50250391994353XXX02	19 December1994	25/39
TM	LT50260391994008XXX01	8 January 1994	26/39
TM	LT50250392004349EDC00	14 December 2004	25/39
TM	LT50260392004052LGS01	21 February 2004	26/39
OLI	LC80250392014328LGN01	24 November 2014	25/39
OLI	LC80260392014287LGN01	14 October 2014	26/39
OLI	LC80250392020361LGN00	26 December 2020	25/39
OLI	LC80260392020336LGN00	1 December 2020	26/39

. Prior to classification, satellite images were orthorectified to a Universal Transverse Mercator projection using datum WGS (World Geodetic System) 84 zone 15N using ERDAS Imagine v16.5 software.

Landsat imagery was categorized into land cover types according to the Normalized Difference Vegetation Index (NDVI) threshold value by supervised classification. This ratio index is helpful in the determination of the spatial changes in vegetation coverage over time. For the extraction of vegetation features, NDVI method was used for four threshold values as shown in

Table 2. Land cover classes along with the NDVI values used for classification of Landsat TM and OLS imagery.

Class Name	NDVI Value	Description
Water	−1 to −0.1	Water body
Built-up area	−0.1 to 0.099	No vegetation
Low vegetation	0.1 to 0.199	Sparse vegetation
Tree Stand	0.2 to >1	Dense vegetation

. The delineated land cover classes were water, built-up areas, low vegetation, and dense vegetation or tree stands. The maximum likelihood supervised classification algorithm was used to generate land cover maps. To assess map accuracy, classification results were compared to reference data including Google Earth, high-resolution aerial images, and ground-based surveys of the study areas. The land cover maps were verified for classification accuracy using the confusion matrix methodology and assessed for overall accuracy.

Social factors such as economic status and population density play a very important role in amplifying the vulnerability of urban communities. Therefore, it is important to evaluate the various demands people place on land that lead to rapid changes in land cover. Census data were collected for the years 1990 and 2020 [34] to assess the socio-environmental interactions related to urbanization.

3. Results

Nutrient and metal elemental analysis for water and soil samples collected from Spring Creek Bayou (SC) are shown in

Table 3. Metal concentrations in water (μg L⁻¹) and soil (mg kg⁻¹) samples (n = 18) for Spring Creek Bayou. TN = Total Nitrogen.

Element	Media	SC3.4	SC45.2	SC88.6	ALP	HHP	EPA	Plant	BG
Cd	water	BDL	0.12	0.08	1.10	5	−	−	−
	soil	0.20	0.28	0.31	−	−	−	32	1
Cr	water	BDL	BDL	0.57	10.6	62	−	−	−
	soil	8.2	13.4	12.6	−	−	−	1	30
Cu	water	BDL	1.35	1.23	0.96	−	−	−	−
	soil	16.5	3.33	7.48	−	−	−	70	15
Ni	water	BDL	BDL	1.60	1.00	332	−	−	−
	soil	4.41	3.95	3.35	−	−	−	38	10
Pb	water	BDL	BDL	0.77	1.46	1.15	−	−	−
	soil	3.56	4.11	6.34	−	−	−	120	15
Zn	water	5.87	11.1	14.8	0.99	−	−	−	−
	soil	24.7	2.63	12.9	−	−	−	160	30
P	water	121	29.1	32.0	−	−	36.6	−	−
	soil	131	48.9	171	−	−	−	−	−
TN	water	539	816	2767	−	−	690	−	−
	soil	1092	323	1383	−	−	−	−	−

Note: Below Detectable Level (BDL); Chronic = Chronic levels for aquatic life protection (ALP); human health protection (HHP); EPA criteria for aggregate nutrients (EPA); toxicological benchmarks for screening contaminants of potential concern for effects on terrestrial plants (Plant); Texas-specific soil background concentrations (BG).

. For water samples, the concentrations of Ni, Zn, P and Total Nitrogen (TN) were at levels that can affect aquatic life. The concentrations of Cu and TN were higher at the mid- and upstream locations SC45.2 and SC88.6 compared to the downstream location SC3.4. The concentration of Zn remained higher than the critical level that would affect aquatic life all along the bayou. However, the concentration of Zn decreased from upstream to downstream along the bayou. The concentration of P remained higher at the downstream location SC3.4 compared to the upstream locations. At the downstream (SC3.4) location, mean P concentrations (121 μg L⁻¹) in water were nearly 3.5 times higher than nutrient criteria levels (36.6 μg L⁻¹). The concentrations of Cr, Cu, Ni, Pb, Zn and TN decreased from upstream to downstream along the bayou (Table 3).

The concentrations of nitrogen and phosphorus in water samples were significantly higher than the EPA criteria for aggregate nutrients (Table 3). The mean N concentrations (816–2767 μg L⁻¹) in water at the midstream (SC45.2) and upstream (SC88.6) locations were 1.1 to 4.0 times higher than nutrient criteria levels (690 μg L⁻¹) while soil N concentrations were highest at the downstream (SC3.4) location, as shown in Figure 2. The Cu concentration in the soil samples at SC3.4 exceeded the background levels (Table 3). There was no significant spatial correlation between the soil and water nutrient concentrations.

3.1. Land Cover Change Analysis

The SCBW was classified into four major land cover classes using Landsat imagery. Results of urban vegetation mapping for the years 1984 to 2020 using the NDVI threshold approach are shown in Figure 3, where green indicates high (dense) vegetation, brown low (sparse) vegetation and red non-vegetation (built-up) areas. The classification results were then evaluated to determine the quality of information obtained from remotely sensed data (

Table 4. Accuracy assessment of vegetation maps in 1984 and 2020 for Spring Creek Bayou.

Land Cover Class	Land Cover Map 1984		Land Cover Map 1994		Land Cover Map 2004		Land Cover Map 2014		Land Cover Map 2020
	PA%	UA%	PA%	UA%	PA%	UA%	PA%	UA%	PA%	UA%
1	100%	65%	93%	70%	95%	90%	95%	100%	100%	100%
2	78%	86%	77%	87%	80%	80%	95%	91%	96%	93%
3	71%	93%	93%	100%	72%	90%	94%	94%	83%	92%
4	99%	98%	99%	98%	98%	92%	99%	99%	97%	93%
Overall Accuracy	94%		95%		91%		98%		93%
Kappa Coefficient	0.86		0.92		0.85		0.96		0.88

Note: Class: (1) Water, (2) No veg., (3) Low veg., (4) High veg.; Producer Accuracy (PA) and User’s Accuracy (UA).

). The overall accuracy for the land cover maps ranged from 91% to 95% while the Kappa coefficient was >0.80, indicating substantial accuracy (Table 4).

The land cover maps for 1984 and 2020 (Figure 3) shows that high or dense vegetation surfaces did not show any significant decrease over the past three decades. Land cover analysis revealed that the total land area of the SCBW was approximately 997 km². The change detection statistics (

Table 5. Land cover transformation matrix for 1984 to 2020 (km²).

	Land Cover Classes	1984 Map				Total Area in 2020
		Water	No Veg.	Low Veg.	High Veg.
2020 Map	Water	0.70	1.01	0.33	4.48	6.53
	No veg.	0.48	3.88	5.61	40.9	51.1
	Low veg.	0.34	3.11	17.1	226.4	247.5
	High veg.	0.22	2.26	19.5	669.1	691.9
	Total Area in 1984	1.77	10.3	42.8	943.1	997.0
	Area Change	4.76	40.78	204.7	−251.1
	Area Change %	269%	397%	479%	−27%

) were applied to identify where changes occurred and the class into which the pixels changed. High-vegetation surfaces accounted for 94.6% of total land area in 1984 but decreased to 69.4% by 2020, decreasing by an estimated area of 251.1 km². Low or sparse vegetation surfaces increased significantly from 4.3% (42.8 km²) of total area in 1984 to 24.8% (247.5 km²) in 2020. The non-vegetation surfaces increased from 10.3 to 51.1 km² in 1984 and 2020, respectively.

3.2. Socio-Economic Dynamics

The SCBW experienced substantial population growth over the 30 years from 1990 to 2020 increasing from 129,629 residents to 389,977 residents (Figure 4). Densely populated areas are found in the eastern region of the watershed in Harris County, while most of the watershed has remained undeveloped. Census years 1990 and 2020 did not reveal a significant shift in the racial composition of the SCBW. In 1990, the watershed comprised 86.9% white residents, 7.0% Hispanic residents, 4.8% Black residents, 0.9% Asian residents, and 0.3% all other residents. By census year 2020, the percent of white residents had decreased to 58.5% but remained in the majority. All other racial groups increased, with Hispanic residents accounting for 24.0% of the total population, Black residents 7.8%, Asian residents 4.7%, and all other race residents 4.9% (Figure 4). The median household income for the watershed increased from $36,636 in 1990 to $86,621 in 2020, which was above the Texas median income for both years [33].

4. Discussion

4.1. Heavy Metal Pollution of Water and Soil

Chemical analysis of water samples collected from SCB revealed that P and TN were the elements of concern that could most affect water quality and aquatic organisms. Mean concentrations of P at the downstream (SC3.4) location and TDN at the midstream (SC45.2) and upstream (SC88.6) locations were above the EPA criteria for aggregate nutrients. The water quality degradation in urban areas typically occurs when urban impervious cover in a catchment reaches between 10 and 15% [35,36]. Although urban regions are often associated with point-source pollution such as wastewater treatment plants, sewage outfalls, industries, hospitals, etc., built-up areas can also be a significant source of diffuse pollution [14,37]. Elevated concentrations of nutrients, which can originate from a variety of urban sources, are often found in runoff derived from built-up areas [18].

The higher nutrient concentrations coincide with the various agricultural areas within the watershed. Phosphorus contributions from urban areas have been found to be even higher than those from agricultural land [18]. High suspended solid and sediment loads are also common in urban runoff, especially from construction activities or road debris [14]. Excess phosphorus and nitrogen can cause excess algal growth, impaired water quality, and decreases in dissolved oxygen levels that are needed for aquatic life survival [31]. Algal blooms can also produce elevated toxins and bacterial growth that are harmful to humans who consume tainted fish, shellfish, or drink contaminated water [31]. The nutrient and heavy metal concentrations in soil samples did not exceed background levels or the criteria for aggregate nutrients set by the EPA.

4.2. Land Cover Change Analysis and Socio-Economic Dynamics

The assessment of land cover change revealed that urban areas slightly increased in the central and western region of the watershed over the last three decades. There was a slight decrease in vegetation surface (4.7%) and a slight increase (4.1%) in built-up (impervious) areas over the last three decades. The more natural features and less impervious surface conditions of the watershed can help to curtail various levels of environmental degradation experienced in urban watersheds similar to the SCB watershed. U.S. Census data [32] revealed a significant growth in population in the eastern region of the watershed while the western region remained undeveloped. The racial dynamics did not change significantly over the three decades, with white residents remaining in the majority. Other racial groups remained under 50% of the total population. Median household income was above the Texas median income for both 1990 and 2020. Land use change patterns are disturbing natural systems, leading to serious ecological and environmental problems, especially in the urban regions of China [38]. The relationship between land cover changes and their drivers in the Greater Bay Area of China were analyzed and found that urbanization, transportation infrastructure, and agricultural practices were the primary drivers of land cover changes in the region [39].

5. Conclusions

Our chemical analysis revealed that P and TN in the water samples were found to be at levels that could pose a threat to water quality and aquatic organisms. Heavy metal concentrations of Zn were at elevated levels in water samples from the SCB watershed. Vegetation surfaces did not decrease significantly over the last three decades, indicating slower urbanization in the SCB watershed. Slight increases in impervious surface conversion correlated with the population growth seen in the western region of the watershed. The SCB watershed experienced extensive population growth, adding over 8500 residents per year over the 30-year period from 1990 to 2020. The median household income remained above the Texas median income. The SCB watershed transitioned from undeveloped to suburban/urban in the western region of the watershed but remained densely suburban for most of the remaining eastern portion of the watershed.