John Dinnall

Site Description

Catchment: The Conservation and Environmental Reasearch Area (CERA) is a suburban headwater stream in Baltimore County, Maryland (39°14’43”N, 76°42’46”W) on the campus of University of Maryland Baltimore County. This catchemtn drains approximately 0.45 square miles of mixed suburban and secondary forest land cover within a narrow, partially forested riparian corridor. The catchment is adjacent to an interstate highway spur constructed in the 1980s.

Sensor Deployment: Three in-stream sensors were deployed at Downstream, Midstream, and Upstream positions with a fourth Air sensor mounted as a dry atmospheric reference. The Downstream position showed active flow. Midstream was fully submerged within a developed riparian zone. The Upstream position exhibited saturated soils and dense skunk cabbage with the sensor at surface water level.

Goal: To prove that low-cost thermal sensing is a viable approach for detecting subsurface flow activity and assessing flow permanence under ambient weather conditions.

Study Area Map

Figure 1: Study area map of CERA Run, Baltimore County, Maryland. Sensor points indicate the locations of the Upstream, Midstream, and Downstream in-stream temperature loggers deployed along the longitudinal transect. The inset map shows the location of CERA Run relative to all study sites within the Baltimore County area.

Figure 2: This chart represents the amount of rainfall that CERA run experienced over the duration of the entire study from July 2025 to 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 aggregated across all eight study catchments.

Figure 4: Daily wetness state classifications for all sensor positions across eight headwater sites from July to October and beyond for some sites. Each row represents a sensor position (Air, Downstream, Midstream, Upstream) and each column represents a single day. Blue indicates wet, red indicates dry, and gray indicates uncertain classifications. A data gap at the Howard County Conservancy corresponds to a sensor error.

Sensor Images

09/22/25

Figure 5: September deployment shows the site during the late summer. All sections of the stream are showing wetness also exhibiting riparian cover throughout the catchment.

11/24/25

Figure 6: November deployment shows the sensor during the autumn and closer to the winter season. Down stream has lost the cover that it initially had during the summer. All part of the stream are wet but midstream has accumulated huge amounts of fall leaves within that its section of the stream.

01/06/26

Figure 7: January deployment shows how mid-winter affects the water surface flow. Here we see both upstream and midstream in frozen water, specifically midstream is clogged with leaves at the spot of the sensor.

02/27/26

Figure 7: February deployment shows how late winter affects the water surface flow. Like the January photos, upstream and midstream still exhibit frozen waters while downstream contiues to keep flow.

05/20/26

Figure 8: May deployment shows the effect of the Spring season. We see that the vegetation has grown since the last deployment along with decent baseflow in all parts of the stream.

Thermal Overview Plots

Figure 9: Hourly temperature recorded by all four sensors at CERA Run from July 2025 through May 2026. The Air sensor reflects ambient atmospheric conditions while the three in-stream sensors reflect channel thermal conditions at each longitudinal position. We see temperatures start in the 70’s but by January you’ll notice the temperature goes down to the 30’s along with the air sensor exhibiting larger temperature swings.

Figure 10: Daily mean temperature for all four sensors at Senior Run from July 2025 through May 2026. Averaging hourly readings to a daily mean smooths short-term fluctuations and reveals seasonal temperature trends across the monitoring period. Persistent separation between the Air sensor and in-stream sensors indicates thermal regulation by water, while merging suggests reduced surface water presence.

Case studies

Purpose: To illustrate sensor response during different atmospheric and seasonal conditions.

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

Dryness Score table: A summary of each sensor’s thermal classification for the selected day.

August 14th, 2025 (Precipitated summer day)

Figure 11: Four-panel diagnostic plot for August 14th, 2025. Temperatures ranged from 73–90°F with 1.06 inches of rainfall arriving between 3–4PM. The Upstream sensor warmed rapidly in the morning before the precipitation event abruptly cooled the atmosphere, triggering a conflict flag at that position. Midstream and Downstream remained thermally stable and classified wet throughout the day.

Dryness Scores — August 14th, 2025
Sensor warming_score cooling_score daily_score conflict_flag wetness_state
Air 1e+00 1e+00 1e+00 FALSE dry
Downstream 4e-04 2e-04 3e-04 FALSE wet
Midstream 0e+00 0e+00 0e+00 FALSE wet
Upstream 4e-04 2e-04 3e-04 FALSE wet

Table 1: A summary of August 14th’s sensor’s thermal classification. The Upstream sensor warmed rapidly in the morning before the precipitation event abruptly cooled the atmosphere, triggering a conflict flag at that position. Midstream and Downstream remained thermally stable and classified wet throughout the day.

July 30th, 2025 (Normal summer day)

Figure 12: Four-panel diagnostic plot for July 30th, 2025. Temperatures ranged from 94°F to 75°F with up to 10mph winds.

Dryness Scores — July 30th, 2025 (Longitudinal Gradient)
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 0.5006 0.0000 0.2503 TRUE wet
Upstream 0.5006 0.5003 0.5004 FALSE uncertain

Table 2: A summary of July 30th’s sensor’s thermal classification. This reading reflects a longitudinal gradient, with upstream being classified as uncertain (the dryest sensor) All the way to Downstream which exhibited no daily score on this date (Wettest).

Decemeber 14th, 2025 (Precipitated winter day)

Figure 13: Four-panel diagnostic plot for December 14th, 2025. Temperatures ranged from 17–36°F with 1/2 a inch of snow. The temperatures of all stream sensors are relatively stagnant through the day. compared to the air sensor which will drop well below freezing according to the raw graph.

Dryness Scores — December 14th, 2025
Sensor warming_score cooling_score daily_score conflict_flag wetness_state
Air 1e+00 1e+00 1e+00 FALSE dry
Downstream 3e-04 2e-04 2e-04 FALSE wet
Midstream 6e-04 2e-04 4e-04 FALSE wet
Upstream 0e+00 0e+00 0e+00 FALSE wet

Table 3: A summary of December 14th’s sensor’s thermal classification. All three stream sensors scored within thousandths of zero, indicating near-complete thermal buffering by surface water at every in-stream position. The Air sensor scored 1.0 as expected. No conflict flags are exhibited here.

December 10, 2025 (Normal winter day)

Figure 14: Four-panel diagnostic plot for December 10th, 2025. Temperatures ranged from 30–45°F with overcast skies and wind gusts of 13 mph. All three stream sensors classified wet with no conflict flags, displaying a clean longitudinal gradient. The Air sensor rose approximately 12°F above its morning minimum while stream sensors rose only 3–5°F — a clear example of winter thermal decoupling between atmospheric and in-stream conditions under cloudy skies.

Dryness Scores — December 10th, 2025 (Longitudinal Gradient)
Sensor warming_score cooling_score daily_score conflict_flag wetness_state
Air 1.0000 1.0000 1.0000 FALSE dry
Downstream 0.0006 0.2000 0.1003 FALSE wet
Midstream 0.0011 0.0000 0.0006 FALSE wet
Upstream 0.0000 0.1997 0.0999 FALSE wet

Table 4: A summary of December 10th’s sensor’s thermal classification. All three stream sensors classified wet with no conflict flags. during these cold temperatures all sensors remained wet and air still acquired a score of 1 due to the diurnal temperature swings being more aligned with the atmosphere.

Analysis

CERA maintained consistent surface water presence across all deployment windows, with Downstream and Midstream sensors classifying wet above 85% of days throughout the study period. The overall dryness heat map confirms persistent flow at all three in-stream positions from July 2025 through May 2026, with no dry periods observed at any sensor location. Conflict flags occurred primarily on precipitation days and thermally complex days. Spring conditions introduced increased precipitation frequency and elevated temperatures that produced higher uncertainty counts, specifically looking at the latter end of the downstream sensor on the heat map.Based on the wetness patterns, CERA is classified as a perennial stream. All three in-stream sensor positions maintained surface water presence across summer, fall, winter, and spring conditions, with no seasonal drying pattern observed at any position.