John Dinnall

Site Description

Catchment: Irvine Run is a forested headwater stream in Baltimore County, Maryland (39°25’58”N, 76°45’29”W), draining 0.56 square miles of mixed forest and institutional land cover within a partially forested channel. The catchment is situated within the Irvine Nature Center, a conservation and environmental education property, providing protected conditions and controlled land use relative to the surrounding suburban landscape.

Sensor Deployment: Three in-stream sensors were deployed at Downstream, Midstream, and Upstream positions with a fourth Air sensor mounted on a tree by the midstream point as a dry atmospheric reference. The Downstream position initially showed no flow and was eventually relocated further upstream. Midstream was fully submerged within a developed riparian zone. The Upstream sensor was positioned within a streambed that consisted of exposed rock and gravel with shallow surface water flowing through the gaps between cobble.

Goal: To document flow patterns in this watershed and assess whether low-cost thermal sensing is a viable approach for detecting subsurface flow activity and flow permanence.

Data Note: The Irvine Run Downstream sensor reached its maximum storage capacity of 8,192 readings in late November 2025, resulting in a data gap through the January 15th, 2026 collection date. This gap is visible across all plots and reflects a hardware logging limitation.

Study Area Map

Figure 1: Study area map of Irvine Run, Baltimore County, Maryland. Sensor points indicate Upstream, Midstream, and Downstream logger locations, including a relocated Downstream position following the September 9th collection. The inset map shows Irvine Run relative to all study sites within the Baltimore County area.

Figure 2: Daily precipitation recorded near Irvine Run from late June 2025 through May 2026.

Figure 3: Stacked bar chart showing the proportion of days classified as wet (blue), dry (red), and uncertain (gray) for each sensor position at Irvine Run across the full study period.

Figure 4: Daily wetness state classifications for all sensor positions across Irvine Run from July 2025 through May 2026. Each row represents a sensor position (Air, Downstream, Midstream, Upstream). Blue indicates wet, red indicates dry, and gray indicates uncertain classifications. There was a data gap from late November 2025, through January 15th, 2026 due the sensors reaching maximum capacity between collections.

Sensor Images

06/27/25

Figure 5: The late June deployment shows the site during the late summer. Midstream exhibited the most visible surface flow of the three sensor positions during the June deployment. Downstream is sitting in a shallow water pool. The Upstream sensor is positioned within an actively flowing reach with a coarse cobble bed.

09/09/25

Figure 6: September deployment shows the sensors during the late summer. Both the Upstream and Midstream sensors showcased reduced flow in the monitored areas. The downstream sensor is completely in a dry patch where the sun directly reflect off the ground.

01/15/26

Figure 7: January deployment shows how mid-winter affects the water surface flow. The relocateed downstream position is showing moderate flow. All sensor positions recorded wet classifications during this visit, with portions of the stream surface exhibiting partial ice formation. All vegetation have also been minimized due to the cold temperatures.

03/02/26

Figure 8: March deployment shows early spring behavior in the stream. All sensors including upstream are exhibiting steady flow.

05/29/26

Figure 9: May deployment shows the effect of late Spring season. All sensors remain wet along with various vegetation that have begun to sprout.

Thermal Overview Plots

Figure 10: Hourly temperature recorded by all four sensors at Irvine Run. The Air sensor reflects ambient atmospheric conditions while the three in-stream sensors reflect channel thermal conditions at each longitudinal position. In the earlier months, the Downstream sensor (green) exhibited the largest diurnal temperature swings of any in-stream sensor, consistent with reduced thermal buffering at that position due to losing stream conditions.

Figure 11: Daily mean temperature for all four sensors at Irvine Run. Averaging hourly readings to a daily mean smooths short-term fluctuations and reveals seasonal temperature trends across the monitoring period. Convergence between Air and stream sensor temperatures may reflect overcast, cool, or rainy conditions suppressing diurnal contrast rather than reduced surface water presence.

Case studies

Purpose: To illustrate sensor response during different atmospheric and seasonal conditions. Each case study presents a four-panel diagnostic plot followed by a dryness score table for a single selected day.

4-Panel Plots: A four-panel diagnostic figure showing each sensor’s thermal behavior over a 24-hour period.

July 14th, 2025 (Summer rainfall day)

Figure 12: Four-panel diagnostic plot for July 14th, 2025. Temperatures ranged from 75–89°F with 3.06 inches of rainfall. The Downstream sensor tracked the Air sensor’s thermal pattern closely in the raw panel, consistent with minimal water buffering at that position. In the normalized panel, Upstream and Midstream sensors rise less than 4°F above their morning minimums while Air and Downstream exceed 8°F, consistent with the losing stream behavior at the Downstream position.

Dryness Scores — July 14th, 2025 (Longitudinal Gradient)
Sensor warming_score cooling_score daily_score conflict_flag wetness_state
Air 0.4997 1.0000 0.7499 TRUE dry
Downstream 1.0000 0.7495 0.8748 FALSE dry
Midstream 0.0000 0.0000 0.0000 FALSE wet
Upstream 0.5008 0.2502 0.3755 FALSE uncertain

Table 1: A summary of July 14th’s sensor’s thermal classification. The most notable feature is that the Downstream sensor recorded a higher daily score than the Air sensor, suggesting the solar radiation that the downstream sensor experienced created the warmest sensor conditions compared to the shaded Air sensor.

September 20th, 2025 (Normal day)

Figure 13: Four-panel diagnostic plot for September 20th, 2025. Temperatures ranged from 69–82°F. Both the Midstream and Downstream sensors exhibit suppressed diurnal temperature amplitudes, with minimal deviation from their morning minimum values throughout the day. The Upstream sensor closely traces the Air sensor, producing an elevated dryness score at that position.

Dryness Scores — September 20th, 2025
Sensor warming_score cooling_score daily_score conflict_flag wetness_state
Air 1.0000 1.0000 1.0000 FALSE dry
Downstream 0.0000 0.0000 0.0000 FALSE wet
Midstream 0.0000 0.0000 0.0000 FALSE wet
Upstream 0.7142 0.6003 0.6573 FALSE uncertain

Table 2: A summary of September 20th’s sensor’s thermal classification. Midstream and downstream both exhibit daily scores of 0 confirming that flow exists at these sections of the stream. The Upstream sensor falls within the uncertain range with a daily score approaching the 0.7 dryness threshold.

January 31, 2026 (Normal day)

Figure 14: Four-panel diagnostic plot for January 31st, 2026. Temperatures ranged from 9–21°F with wind gusts steadily increasing throughout the day to 15 mph.

Dryness Scores — January 31st, 2026
Sensor warming_score cooling_score daily_score conflict_flag wetness_state
Air 1.0000 1.0000 1.0000 FALSE dry
Downstream 0.0000 0.0000 0.0000 FALSE wet
Midstream 0.4994 0.0000 0.2497 FALSE wet
Upstream 0.4983 0.9967 0.7475 FALSE dry

Table 3: A summary of January 31st’s sensor’s thermal classification. A longitudinal gradient is exhibited, with the Upstream sensor classified as the driest in-stream sensor (0.7475), followed by Midstream (0.2497) and Downstream (0.0).

February 22, 2026 (Normal winter day)

Figure 15: Four-panel diagnostic plot for February 22nd, 2026. Temperatures ranged from 32–39°F with overcast skies and wind gusts of 16 mph. All sensors exhibit similar trends in the raw panel with air leading as the sensor that exhibits the most diurnal temperature flux. The dryness score panel indicates that Downstream and Upstream fall within the uncertain range.

Dryness Scores — February 22nd, 2026
Sensor warming_score cooling_score daily_score conflict_flag wetness_state
Air 1.0000 1.0000 1.0000 FALSE dry
Downstream 0.0022 1.0000 0.5011 TRUE uncertain
Midstream 0.0011 0.0000 0.0006 FALSE wet
Upstream 0.0000 0.9956 0.4978 TRUE uncertain

Table 4: A summary of February 22nd sensor’s thermal classification. Midstream exhibits the lowest score of all the sensors (0.0006). But there are conflict flags that arise for both the upstream and downstream sensors. Both sensors’ warming score is extremely low and near 0 while the cooling scores are extremely high. This adverse divergence between the warming and cooling phases triggered conflict flags at both the upstream and downstream positions, indicating that the thermal signal at those locations could not be confidently resolved into a single classification for the day.

Analysis

Irvine Run exhibited spatially variable flow permanence across the study period, with distinct hydrological behavior observed at each sensor position. Midstream maintained near-perennial wet classifications across all windows, confirming persistent baseflow in the gaining reach throughout the year. Downstream classified dry 91% of days during the summer window which may be due to traversing through agricultural farmland with minimal riparian canopy cover that contributed to water loss. Following the September 9th collection, the sensor was repositioned to a transitional midreach location between the original Downstream and Midstream positions to better capture the longitudinal gradient between the forested gaining reach and the agricultural losing reach. Upstream exhibited seasonal intermittency, transitioning from 62% wet in summer to 17% wet in fall before recovering toward wet during winter groundwater recharge. The spatial abruptness of dry conditions at the original Downstream position while Midstream remained wet within the same reach reflects the geological control of carbonate bedrock on surface flow at this site. The repositioned Downstream sensor captured the transitional zone between the gaining and losing reaches, providing evidence of the longitudinal discontinuity in surface flow within the monitored network. These findings suggest losing stream at the monitored locations, while being fed through groundwater at the headwater position the loss of surface connectivity further downstream solidifies this hypothesis. The classification cannot be extended to the broader stream reach without additional spatial sampling.