A Spatially Distributed Investigation of Stream Water Temperature in a Contemporary Mixed-Land-Use Watershed

: Stream water temperature ( ◦ C) is an important physical variable that inﬂuences many biological and abiotic water quality processes. The intermingled mosaic of land-use / land-cover (LULC) types and corresponding variability in stream water temperature (Tw) processes in contemporary mixed-land-use watersheds necessitate research to advance management and policy decisions. Water temperature was analyzed from 21 gauging sites using a nested-scale experimental watershed study design. Results showed that forested land use was negatively correlated ( α = 0.05) with mean and maximum Tw. Agricultural land use was signiﬁcantly positively correlated ( α = 0.05) with maximum Tw except during the spring season. Mixed development and Tw were signiﬁcantly correlated ( α = 0.05) at quarterly and monthly timescales. Correlation trends in some reaches were reversed between the winter and summer seasons, contradicting previous research. During the winter season, mixed development showed a negative relationship with minimum Tw and mean Tw. During the summer season, higher minimum, maximum, and mean Tw correlations were observed. Advanced understanding generated through this high-resolution investigation improves land managers’ ability to improve conservation strategies in freshwater aquatic ecosystems of contemporary watersheds. impact that maximum Tw can have on stream ecosystems, a preliminary investigation of thermal surge events was conducted. This investigation showed thermal surges in the study watershed, and therefore presents future research opportunities into the investigation of Tw surge dynamics. Both results and ﬁndings of this study will advance the decision-making success of land managers and policy makers concerned with the health of aquatic ecosystems. In particular, the high-resolution (n = 22) study design presented in this work facilitates identiﬁcation of upland mitigation sites and corresponding greater certainty in ﬁscal investment outcomes.


Introduction
Stream water temperature (Tw) affects abiotic and biotic processes in aquatic ecosystems [1,2]. Abiotic variables influenced by Tw include dissolved oxygen concentration, chemical reaction rates, viscosity, density, and surface tension [3]. Biological processes influenced by Tw include the growth rate of fish [4] and rates of primary production in some autotrophic species [5]. Water temperature therefore impacts multiple trophic levels of the aquatic food web, including periphyton, benthic macroinvertebrates, and fishes [6]. This is of further relevance given that many aquatic organisms (e.g., mottled sculpin (Cottus bairdii), Escherichia coli, and flathead mayfly (Heptageniidae sp.)) have specific tolerance ranges for Tw [4]. Water temperature tolerances/preferences also directly affect species distributions and population densities, thus affecting many important aquatic industries (i.e., fishing) [7][8][9]. Globally, water temperature has traditionally received less attention relative to other water quality parameters, such as suspended sediment and water chemistry [1]. However, water temperature's ecohydrological importance and susceptibility to anthropogenic disturbance make it a critical variable of concern for resource managers [4,10]. also been shown to differentially affect Tw [39]. Rice et al. [34] showed that increasing watershed urbanization causes the Tw and Ta relationship to break down. This is offset, however, given that higher channel volumes downstream are often sufficient enough to attenuate thermal inputs [2,40]. Thus, stream inputs (e.g., groundwater, surface runoff, and confluence) from surrounding land-use types may have a greater influence upstream near headwaters relative to downstream and during periods of low(er) flow relative to periods of flooding [6,9]. Ultimately, water volume and Ta are not the only influencing factors affecting Tw, and the impacts of complex LULC must be considered [6,14,27,32,34].
The Appalachian region of the United States is an example of a physiographically complex region with high relief relative to surrounding areas [41]. Factors such as higher relief more greatly alter the rate of runoff from surrounding land-use types relative to flatter terrain [42]. Thus, a physiographically complex region, such as Appalachia, is well suited to advance existing knowledge gaps in the relationship between land-use types and Tw for many regions. In recent years, Appalachia, in particular the state of West Virginia, has been impacted by increased flood frequency, largely attributed to an increase in percent mixed development land-use types [43]. As mixed development land-use types expand, they often replace forested land-use types [44]. This is important because previous research showed these two land-use classifications have distinct effects on Tw. Furthermore, the interactions between competing mixed land-use types are largely unknown. Thus, studies in a contemporary mixed-land-use watershed designed to assess land-use impacts on hydrologic variables, including Tw, are needed in general (globally) and in the region (specifically) [45][46][47][48].
There is an ongoing need for high-resolution studies in contemporary watersheds that include multiple land-use practices [45][46][47][48]. Such studies will advance spatial and temporal understanding of Tw regimes, and therefore improve management decision making in contemporary (municipal) watersheds [13,[45][46][47][48]. An effective method used to assess hydrologic processes in contemporary mixed-land-use watersheds is the experimental watershed study design [13,25,32,45,[47][48][49][50][51]. The experimental watershed study design is effective at addressing both site-specific management questions and assisting in predictive model development, calibration, and validation [48][49][50]. Using this study design, researchers can partition a larger catchment into individual sub-catchments, enabling quantification of specific land-use impacts on variables of interest [13,[48][49][50][51]. There is a lack of municipal watershed-scale studies that utilize this broadly accepted method (experimental watershed study design) in Appalachia, or elsewhere. The lack of such studies may be in part attributable to preconceptions of historic high cost of instrumentation, time-consuming data collection, and challenges with the transferability of results [48][49][50][51]. However, advantages of successful outcomes of the experimental watershed study design (e.g., high-temporal and spatial-resolution data) have been repeatedly shown to far exceed disadvantages [48][49][50][51].
The overall objective of this study was to use a highly instrumented (n = 22 gauging sites) experimental watershed study design to investigate land-use practice impacts on Tw, particularly maximum Tw, spatially and temporally in a representative contemporary mixed-land-use watershed. Sub-objectives included quantifying (a) annual, (b) quarterly, and (c) monthly relationships between LULC and Tw. Increased understanding generated through work such as this, better informs land managers wishing to improve the conservation and preservation of aquatic ecosystems [1,2,7-9,47-51].

Study Site
West Run Watershed (WRW) is located in the Monongahela Watershed and is categorized as a Hydrologic Group D watershed (HUC #05020003) located near Morgantown, West Virginia, USA [12]. The watershed spans 23 km 2 , and the main drainage, West Run Creek, drains directly into the Monongahela River [27,35]. Anecdotally observed by the project managers of the watershed-based plan for West Run of the Monongahela River [52], the increase in urban sprawl stemming from the surrounding city of Morgantown continues to increase the severity and frequency of flooding. The channel of West Run, according to habitat surveys, lacks sinuosity in many reaches and possesses a low channel slope of 1.1%, with back watering (flooding) near the terminus [12,52]. For this reason, site 22 ( Figure 1) was excluded from the current study due to potential backwatering with the Monongahela River, and subsequent confounding analyses at that location. The average net radiation in Morgantown, measured during a study by Arguez et al. [53], from 1981 to 2010 was 130.68 W/m 2 . The average recorded precipitation depth in Morgantown is approximately 1096 mm/year . Precipitation falls throughout the year but increases in quantity during the spring and summer months [54]. Previous studies showed that precipitation has significantly (p = 0.01) increased in the Appalachian region by 2.2% over the past 111 years [27,54,55]. Precipitation is frequently generated via frontal storm convergence systems, but during the summer months in particular, precipitation also occurs via orographic and/or convective processes [12,[54][55][56]. Dynamic water parameters (i.e., base flow, Tw, and storm flow) were first monitored in the WRW in 2016 when stilling wells were installed [12], using a paired and nested-scale experimental watershed study design. Each site was equipped with a Solinst Levelogger Gold pressure transducer that logged and stored Tw ( • C) data, with an accuracy of ±0.05 • C, and stage (water depth, cm), with an accuracy of ±0.3 cm, at five-minute intervals. During the study period, climate data were recorded using research-grade climate instrumentation located within approximately 100 m of site 13 ( Figure 1). Climate variables (recorded at a height of 3 m) included precipitation (TE525 Tipping Bucket Rain Gauge), average air temperature and relative humidity (Campbell Scientific HC2S3 Temperature and Relative Humidity Probe), average wind speed (Met One 034B Wind Set instrument), and net radiation (Campbell Scientific NR01 Four-Component Net Radiation Sensor).

LULC Data
LULC data were derived from National Agricultural Imaging Program (NAIP) 2016 data. Initially, LULC types included 16 different assignations. For the current work, each of the original 16 LULC types was further grouped (lumped) into one of four LULC categories mixed-developed, agriculture, forested, and open water (Table 1). Using Arc GIS, watershed and sub-catchment boundaries were delineated, and LULC data incorporated. Each pixel representing 5 m 2 was counted and converted to km 2 and used to estimate percent LULC for each sub-catchment draining to each monitoring site; see Table 2 and Figure 2. Overall land use in West Run Watershed assessed using Arc GIS projected a composition of 19.4% agriculture, 42.7% forest, and 37.7% urban/suburban, with the

LULC Data
LULC data were derived from National Agricultural Imaging Program (NAIP) 2016 data. Initially, LULC types included 16 different assignations. For the current work, each of the original 16 LULC types was further grouped (lumped) into one of four LULC categories mixed-developed, agriculture, forested, and open water (Table 1). Using Arc GIS, watershed and sub-catchment boundaries were delineated, and LULC data incorporated. Each pixel representing 5 m 2 was counted and converted to km 2 and used to estimate percent LULC for each sub-catchment draining to each monitoring site; see Table 2 and Figure 2. Overall land use in West Run Watershed assessed using Arc GIS projected a composition of 19.4% agriculture, 42.7% forest, and 37.7% urban/suburban, with the portion of agricultural land use mainly comprised of animal husbandry (i.e., cattle) and crop fields (i.e., corn, soybeans, and cover crops) [12,27]. The presence of different land-use types coupled with regular land-use manipulation via human development justifies WRW categorization as a representative contemporary mixed-land-use watershed [12]. A combination of agricultural, mining, and industrial land use in and surrounding WRW has contributed to ongoing land and water resource degradation [12]. As per The United States Census Bureau report from 2019, the current population of Morgantown is 30,539 [57]. Water, river floodplains, and wetlands PEM associated with early successional stages (i.e., low vegetation) or agriculturally maintained pastures and fields. The original classification grouped as mixed development land-use practices included mixed-development and all other LULC classifications associated with urban areas and/or impervious surfaces (e.g., barren, roads, and impervious) ( Table 1) [34,35,60,61]. Figure 2 shows LULC proportions separated into percent area, and the LULC area of both the individual sites and each individual sub-basin.  Forested land-use practices are the most abundant land-use classification in the WRW. For the current work, land-use/land-cover types grouped in the forested category included mature vegetation but were separated from mine grass classification, which is assumed to have succeeded to an intermediate successional stage [58,59]. Agricultural land-use practices were those classifications associated with early successional stages (i.e., low vegetation) or agriculturally maintained pastures and fields. The original classification grouped as mixed development land-use practices included mixed-development and all other LULC classifications associated with urban areas and/or impervious surfaces (e.g., barren, roads, and impervious) ( Table 1) [34,35,60,61]. Figure 2 shows LULC proportions separated into percent area, and the LULC area of both the individual sites and each individual sub-basin.

Data Analysis
Data analyses included descriptive statistics (5 min data) of annual, quarterly, and monthly data in 2018 [47]. Quarterly time steps were delineated as 1 January-31 March (quarter 1), 1 April-30 June (quarter 2), 1 July-30 September (quarter 3), and 1 October-31 December (quarter 4). Data postprocessing included estimation of erroneous data or missing points (<0.2% of total data) by averaging between data points on either side of a gap, or by linear interpolation [62]. The Tw data were shown to be non-normally distributed using the Anderson-Darling test [62]. Therefore, land-use areas corresponding to each site were compared to maximum Tw using the Spearman rank correlation coefficient test (α = 0.05). The open water land-use type was excluded from analyses due to its negligible areal coverage relative to other land-use types in the watershed. Three separate analyses were run including (a) all sites (b) tributaries only (i.e., sites 1, 2, 5, 7, 8, 9, 11, 12, 14, 15, 16, 17, and 20), and (c) mainstem West Run Creek only (i.e., sites 3, 4, 6, 10, 13, 18, and 21) to assess the varying effects of surrounding land-use practices on tributaries vs. mainstem sites. Due to the detrimental influence of maximum Tw on aquatic biological/geochemical processes, a pairwise comparison of daily maximum Tw was conducted on a site by site basis using a Kruskal-Wallis ANOVA [62]. Multiple principle component analyses (PCAs) were developed to illustrate the relationship between LULC types at an annual, quarterly, and monthly timescale, again excluding open water [63]. Using OriginPro Pro 9b Academic (OriginLab Corporation, Northampton, MA, USA), correlation biplots were generated with standardized data [63]. Due to the use of observed Tw data, no autoscaling preprocessing was needed to compare Tw and LULC data in the PCA analysis. However, due to the use of differing units (i.e., proportions (%) and temperature ( • C)), a correlation matrix was used rather than a covariate matrix [63]. Following the Kaiser-Guttman criterion, eigenvalues greater than one were accepted as principle components above the threshold of importance [63].

Climate during Study
Total precipitation recorded in West Run Watershed in 2018 was 1378 mm, which was 282 mm higher than average annual precipitation (1096 mm) over the previous 111 years [54]. The largest precipitation event (22.9 mm) occurred in a 5 min window on 9 September. The largest continuous precipitation event (83.2 mm) began on 8 September at 18:00 and lasted until 9 September at 16:30; see Figure 3b. Mean air temperature (Ta) in 2018 was 11.6 • C, which was 0.2 • C lower than the average annual temperature (11.4 • C) in West Virginia between 1990 and 2016 [54]. The coldest (−24.8 • C) and warmest (34.6 • C) recorded temperature occurred on January 1st at 8:00 and on 1 July at 17:00, respectively. The maximum net radiation was 1100 W/m 2 , recorded on 7 May at 12:00. The mean near surface (1.   (rs = 0.6) between maximum Tw and agriculture LULC area, and a negative correlation (rs = −0.5) between maximum Tw and Forest LULCs (p = 0.03). Table 3. Descriptive statistics of stream water temperature ( • C), annual by site, quarterly (all sites), and monthly (all sites) collected in West Run Watershed, West Virginia, USA, in 2018.

Sites/Season/Month
Mean

Quarterly Stream Water Temperature
Seasonal Tw regimes showed that quarter 1 (1 January-31 March) had the lowest minimum Tw    (Table 3). LULC and Tw analysis (n = 21 sites) showed 12 significant (α = 0.05) correlations with Tw variables. Results from the Kruskal-Wallis ANOVA showed that all 21 sites were significantly different from each other during at least one of the twelve months based on daily maximum Tw. The most significant differences were shown in July, with 108 significant differences (p ≤ 0.04). Both February and October had no significant differences between sites. Significant Spearman correlation coefficients (α = 0.05), between LULC types and maximum Tw,

PCAs
Annual timeseries comparisons between LULC areas and Tw variables at site 1-21 all showed two eigenvalues above the threshold of importance. If all variables (n = 4) exerted equal influence for the annual PCA, the values would be 0.5 (i.e.,

Climate during Study
Climate variables (e.g., Ta, precipitation, and net radiation) recorded in West Run Watershed in 2018 were average relative to historic climatic trends of West Virginia (i.e., 1900-2016) [54]. The average mean temperature (11.6 • C) differed only slightly (1.7%) from historic (1900-2016) averages observed in West Virginia from 1900 to 2016 (11.4 • C). During the study period, there was above average (20.5% higher) total precipitation relative to the historic (1900-2016) average (1096 mm) [54,64]. WRW did not include a dry season in 2018 and, a majority of the total precipitation fell during quarters 2 and 3 [46]. The overall climate in WRW during 2018 was predictably variable and consistent with historic climate trends ( Figure 3) [46].

Stream Water Temperature
The highest maximum (27.4 • C) Tw was recorded during quarter 3, specifically for July. This is expected given the seasonal climate of WRW, which has the highest recorded Ta (34.6 • C) during July. While the highest Ta and Tw were recorded during quarter 3, the highest mean Ta was recorded during July, and the highest mean Tw was recorded during September (19.6 • C). In quarter 2, May had the highest Tw standard deviation (3.3 • C) (Figure 4) [64]. Further analysis into the Tw and Ta trends ( Figure 3) showed that Tw followed but lagged behind Ta during May and across the entire study period [65][66][67][68]. This trend is constant with results of past studies that showed a strong relationship between Tw and Ta [39,66,67]. In May, Ta had a high standard deviation compared to other months (6.2 • C). The high Tw standard deviation in quarter 2 and May is likely attributable to the close relationship between Ta and Tw, as shown in many previous studies [38,39,66,67].

Stream Water Temperature LULC Relations
In general, results from both the PCA ( Figure 5) and the Spearman rank correlation coefficient test (Table 5) showed that an increase in the proportion of forested LULC types is negatively correlated with all Tw variables, as confirmed in previous studies [14,22,65]. These results follow the same conclusions made by previous researchers that showed that forest harvest (e.g., clear cuts/canopy removal) increases Tw [6,14,16,22]. Additionally, although not surprising, during the winter season, a positive correlation was observed between the proportion of forest LULC and minimum Tw. Moore et al. [14] suggested that riparian vegetation insolates Tw by lowering convective heat loss to the above atmosphere, thereby cooling Tw in the summer and warming Tw in the winter via latent heat gain [15,23,68]. In the current work, positive correlations between maximum Tw and percent forest LULC were observed in specific tributaries (Figure 6), thus contradicting findings of previous research [6,65,68]. However, when sites 8 and 9 were removed from the data pool, correlations trends reverted to the expected negative correlation [6,65]. For sites 8 and 9 of the current work, the positive correlation may be attributed to the high proportion of directly adjacent mixeddevelopment LULC types at both sites 8 and 9 prior to monitoring sites.  Moore et al. [14] suggested that riparian vegetation insolates Tw by lowering convective heat loss to the above atmosphere, thereby cooling Tw in the summer and warming Tw in the winter via latent heat gain [15,23,68]. In the current work, positive correlations between maximum Tw and percent forest LULC were observed in specific tributaries (Figure 6), thus contradicting findings of previous research [6,65,68]. However, when sites 8 and 9 were removed from the data pool, correlations trends reverted to the expected negative correlation [6,65]. For sites 8 and 9 of the current work, the positive correlation may be attributed to the high proportion of directly adjacent mixed-development LULC types at both sites 8 and 9 prior to monitoring sites.
temperatures [15,27,29]. Furthermore, during these periods, water contacting or infiltrating into the soil is heated and transported to surrounding streams via runoff or subsurface lateral flows [27]. Younus et al. [27] found that drainage tiles exacerbated transport of heated subsurface lateral flow into surrounding streams during irrigation or precipitation events. The agricultural fields in the current study watershed have drainage tiles installed. Thus, subsurface lateral flows to West Run Creek may be increased, further explaining the significant relationship between maximum Tw and agricultural LULC types.

LULC Tw Tipping Points
Forested LULC types influenced Tw most significantly in the current investigation ( Figure 6). As the percent of forested LULC types decreased below 74.2%, associated maximum Tw began to increase (thus a potential tipping point). As the percent of forested LULC decreased below 61.1% mean Tw began to either increase or decrease depending on the time of year. Below 52.2% forested LULC, the minimum Tw of associated streams began to increase or decrease again depending on the time of year. Mixed-development LULC types had next greatest impacts on Tw.  Results from both the PCA ( Figure 5) analyses and the Spearman rank correlation coefficient test (Table 5) showed that mixed-development LULC types were significantly correlated with mean, minimum, and maximum Tw, with variable effects throughout the year [30][31][32]34,35,60,61]. While results of the mixed-development Tw analysis were similar to results of previous research, the current work provides important validation by means of the high number of sampling sites and high frequency sampling [32,34,47]. Interestingly, PCA results showed a negative correlation between minimum, maximum, and mean Tw during the winter and early spring months (i.e., January, February, March, April, November, and December), indicating overall lower Tw. These findings contradict results of other research analyzing the relationship between mixed-development LULC and Tw. Rice et al. [34] showed that mean Tw increased during the winter season in heavily urbanized catchments of Boone, North Carolina. Alternatively, to Rice et al. [34], lower Tw correlated with mixed development during cooler months (i.e., January, February, March, April, November, and December) could be explained by runoff from impervious surfaces, which during these months is often lower in temperature than the Tw of surrounding streams. Qun et al. [69] showed a negative correlation between urban impervious surface and land surface temperatures (i.e., lower surface temperatures) during winter daytimes. Conflicting results might be further explained by the complex physiographic mosaic of the study watershed in which mixed development LULC types are broken up by other forested and agricultural LULC types at varying relative positions on the landscape, thus leaving room for future investigations.
Agricultural LULC types had a positive Spearman correlation, with maximum Tw in every quarter and month, except January (Table 5). These findings may be, at least in part, due to the removal of riparian vegetation, increased subsurface lateral flow rates through drainage tiles, and/or increased soil shortwave radiation exposure during periods when fields are bare as per findings of previous literature [27,70]. Interestingly, a relationship appeared between maximum Tw and crop absence. Typically, in the study watershed, crops are planted in spring and harvested in middle to late July following the observed trend in Table 5. Therefore, the positive relationship between crop absence and maximum Tw could be explained, at least in part, by increased shortwave interception and reduce runoff volumes through evapotranspiration and interception of precipitation by crops [15,23]. Conversely, after harvest, the relationship becomes significant because runoff volumes increase and the soil is exposed to greater amounts of shortwave radiation thereby reaching higher temperatures [15,27,29]. Furthermore, during these periods, water contacting or infiltrating into the soil is heated and transported to surrounding streams via runoff or subsurface lateral flows [27]. Younus et al. [27] found that drainage tiles exacerbated transport of heated subsurface lateral flow into surrounding streams during irrigation or precipitation events. The agricultural fields in the current study watershed have drainage tiles installed. Thus, subsurface lateral flows to West Run Creek may be increased, further explaining the significant relationship between maximum Tw and agricultural LULC types.

LULC Tw Tipping Points
Forested LULC types influenced Tw most significantly in the current investigation ( Figure 6). As the percent of forested LULC types decreased below 74.2%, associated maximum Tw began to increase (thus a potential tipping point). As the percent of forested LULC decreased below 61.1% mean Tw began to either increase or decrease depending on the time of year. Below 52.2% forested LULC, the minimum Tw of associated streams began to increase or decrease again depending on the time of year. Mixed-development LULC types had next greatest impacts on Tw.
As the percent of mixed-development LULC types increased above 14%, associated mean Tw increased (thus a potential mixed-development tipping point). Above 24.1% mixed development, maximum Tw began to either increase or decrease depending on the time of year, and above 26.8% mixed development, the minimum Tw of streams began to increase or decrease (depending on the time of year). Agricultural LULC types also influenced Tw significantly. As the percent of agriculture LULC types increased above 14.9%, associated maximum Tw began to increase. Above 16.0% agriculture LULC, mean Tw began to increase, and above 26.4% agriculture LULC, the minimum stream Tw began to increase or decrease (time of year dependent).

Study Implications and Future Directions
Implementing the nested experimental watershed study design coupled with the high-temporal and spatial-sampling regime used in this work allowed for a more comprehensive evaluation of surrounding LULC effects on associated Tw than is normally found in the primary literature. Forested LULC types were associated with overall lower Tw, whereas both mixed-development and agriculture LULC types had higher overall Tw. The finding of LULC tipping points (thresholds) for all three of the analyzed LULC classifications provides valuable information for both land managers and policy makers. For example, Tw tipping points emerged as forested LULC types dropped below 74.2%, due to the conversion to mixed development or agricultural LULC types. These tipping points can be used to guide management decisions in terms of development limits. Given the importance of maximum Tw for stream biota, a preliminary analysis of thermal surges was conducted. Select summer precipitation events were assessed as per Zeiger and Hubbart [32], Rice et al. [34], and Anderson et al. [71], where a Tw surge is defined as a greater than 1.0 • C increase within a 15 min time interval. Figure 7 is a stacked comparison of Tw on the y-axis and time on the x-axis showing time lag from the related precipitation event. Successive sites were added to Figure 7, showing the thermal plume moving through the watershed and eventually dissipating. Although this analysis was preliminary, it showed the existence of these events, thus conveying the need for future research into thermal surge dynamics. Future investigations may benefit from additional monitoring years to better understand the importance of climate and antecedent conditions with regard to Tw processes. In addition, future studies could focus on minimum Tw and mean Tw, and perhaps Tw variance, to understand relationships between land use, climate change and Tw regimes.

Conclusions
No previous research investigating stream water temperature (Tw) in contemporary watersheds has used such a high-temporal and spatial-sampling regime as that included in the current investigation. Implementation of the experimental design used in the current research is necessary to provide both validation for previous results and the discovery of temporal variation in LULC characteristics influencing Tw. In the current work, the relationship between LULC types and Tw was investigated in the Appalachian region of the eastern United States. The analysis used five-minute Tw timeseries data collected at 22-site nested sites using an experimental watershed study design. Results indicated that LULC has varying effects on Tw both spatially and temporally. PCA results showed that forested LULC types typically lowered maximum and mean Tw, particularly in the late summer months, whereas Spearman correlation results showed significant (p = 0.01) negative correlations with maximum Tw (−0.9) during August and September. PCA results indicated that mixed-development LULC types typically increased maximum and mean Tw during the summer months, whereas Spearman correlation results showed significant (p = 0.00) positive correlation with maximum Tw (0.9) during August and September. PCA results showed that agriculture LULC types were correlated with maximum Tw in every month except January. Although results are specific to the study watershed, the finding of tipping points shows LULC thresholds that, when exceeded, may begin to impact associated Tw. These relationships likely exist in all watersheds, particularly contemporary (municipal) watersheds. Given the impact that maximum Tw can have on stream ecosystems, a preliminary investigation of thermal surge events was conducted. This investigation showed thermal surges in the study watershed, and therefore presents future research opportunities into the investigation of Tw surge dynamics. Both results and findings of this study will advance the decision-making success of land managers and policy makers concerned with the health of aquatic ecosystems. In particular, the high-resolution (n = 22) study design presented in this work facilitates identification of upland mitigation sites and corresponding greater certainty in fiscal investment outcomes.