The research, conducted by scientists at CSIR-National Botanical Research Institute and the University of Lucknow, harnessed two of the most powerful Earth-observation tools available: Landsat 8’s thermal infrared sensor, which maps land surface temperature at 30-meter resolution, and Sentinel-2’s multispectral imager, which tracks vegetation health at 10 meters. The team selected cloud-free scenes from June and December of both 2019 and 2024, allowing them to compare warm-season and winter conditions across a five-year span of relentless urban expansion. Lucknow itself provided a compelling laboratory. Its built-up area has grown nearly fivefold in three decades, from roughly 54 square kilometers in 1991 to more than 260 square kilometers by 2021, while vegetated cover within and around the urban core has steadily eroded.

To quantify cooling, the researchers defined a deceptively simple metric: the temperature difference between each green patch and a 50-meter ring of built-up land immediately surrounding it. This patch-to-matrix contrast, expressed as delta LST, was calculated using median values to guard against outliers and mixed pixels, and a land-cover mask ensured that only genuine built-up surfaces contributed to the reference temperature, excluding water bodies and neighboring vegetation that could contaminate the signal. Positive values indicated that the patch surface was cooler than its immediate surroundings. The team then extended the analysis outward through concentric buffer rings reaching 500 meters, tracing how far the thermal contrast persisted into the urban fabric.

The headline result is that green patches in Lucknow are indeed consistently cooler than the concrete around them, but the magnitude is modest by global standards. Median warm-season contrasts hovered between 0.6 and 0.8 degrees Celsius, with local maxima exceeding 2 degrees. Winter contrasts shrank to roughly a quarter of that, around 0.2 degrees. This seasonal amplification fits a well-established pattern: park cool islands are strongest when evapotranspiration and canopy shading are working hardest, and weakest when vegetation is dormant. Compared with cities like Bengaluru, where mean cooling of 2.23 degrees has been reported extending nearly 350 meters beyond green-space boundaries, Lucknow’s humid subtropical climate, with its high background humidity and monsoon seasonality, appears to dampen the thermal gradients that green patches can generate.

Perhaps the most striking finding concerns patch size. The relationship between area and cooling proved distinctly non-linear. Cooling contrast climbed steadily from patches of about one hectare through the medium-size range, then flattened dramatically. Segmented regression identified approximate breakpoints near 8.4 hectares in summer scenes and 10.2 hectares in winter, beyond which additional area yielded almost nothing, with slopes hovering near zero or turning slightly negative. Generalized additive models confirmed the same shape: a broad maximum in the 5-to-10-hectare range, followed by a plateau. In practical terms, a 12-hectare park delivers roughly the same surface-temperature contrast as an 8-hectare one. The authors are careful to frame these figures as context-specific heuristics rather than universal design thresholds, noting that the exact breakpoints are sensitive to patch delineation choices and the native resolution of Landsat’s thermal sensor.

Size, however, was not the whole story. Vegetation condition emerged as a powerful and seasonally dependent control on cooling performance. In summer 2024, the temperature contrast correlated strongly with every vegetation index the team examined: NDVI, a proxy for green vegetation amount, explained 28 percent of the variance; leaf area index explained 27 percent; the moisture-sensitive NDMI explained 24 percent; and the chlorophyll vegetation index explained 14 percent. In winter, these relationships collapsed to near zero or vanished entirely. The implication is clear: a lush, dense, well-watered canopy cools far more effectively than a sparse or stressed one, and this advantage matters most precisely when heat exposure peaks. Larger patches consistently showed higher values across all vegetation indices, suggesting that size and quality reinforce one another, likely through more continuous canopy cover, reduced edge exposure and better moisture retention.

The spatial reach of cooling also proved limited and sharply defined. Buffer analyses showed that surface-temperature contrasts were detectable mainly within the first 100 to 200 meters of the surrounding built-up matrix, after which the distance-decay curves generally flattened toward 500 meters. Summer contrasts at the outermost ring reached 2.5 to 3.2 degrees, while winter values stayed below about 1.3 degrees. The researchers interpret this footprint as a neighborhood-scale indicator rather than a direct measurement of cooling propagation, since rising impervious cover at greater distances partly offsets the cooling that patches generate. Notably, peri-urban patches, embedded in less compact mosaics of agriculture and remnant vegetation, showed consistently stronger contrasts than their densely urban counterparts, with summer medians of 0.8 to 0.9 degrees versus 0.6 to 0.7 degrees in 2019, a pattern that echoes earlier findings that matrix context can sustain or erode a patch’s thermal influence.

Not every geometric variable lived up to expectations. Patch shape, measured by a shape index capturing boundary irregularity, showed only a weak positive association with cooling, explaining barely 2 percent of the variance. Euclidean nearest-neighbor distance, a measure of isolation, was not statistically significant. A multivariate regression restricted to area, shape and isolation explained just over 5 percent of the variation in temperature contrast, with patch area the strongest predictor. The team deliberately excluded vegetation indices from this model because variance inflation diagnostics revealed severe multicollinearity, with NDVI and LAI carrying nearly overlapping information, a methodological honesty that prevents unstable coefficients but also limits mechanistic attribution. The authors caution that shape metrics are particularly sensitive to how patch boundaries are drawn, and their delineation was performed by a single operator without inter-operator validation.

The study is equally candid about its limitations. The June scenes were not normalized for antecedent rainfall, humidity or heat-wave conditions, so year-to-year differences may partly reflect short-term weather rather than structural change. Field surveys at 45 sites verified patch presence and dominant species but did not quantitatively validate the satellite-derived temperatures. Patches smaller than one hectare were retained for completeness but treated as high-uncertainty because of mixed-pixel effects at Landsat’s thermal resolution. And crucially, the analysis measures surface temperature, not air temperature, human thermal comfort or health outcomes, so the findings should guide planning rather than serve as direct evidence of physiological heat relief.

Even with these caveats, the planning implications are concrete. The diminishing-returns pattern suggests that cities should prioritize protecting existing canopy-rich medium-to-large patches over chasing ever-larger parks, while simultaneously improving vegetation condition and moisture retention in smaller patches, where quality gains translate directly into cooling gains. Reducing impervious gaps between green spaces, through street trees, neighborhood gardens and low-heat surface treatments, could extend the reach of limited green infrastructure across the matrix. For land-constrained cities across South Asia and other humid subtropical regions, the study’s real contribution may be its workflow: estimate patch-to-matrix contrast, identify the size range where marginal cooling weakens, map the buffer distance over which contrast remains detectable, and read all of it against vegetation condition and surrounding context. In a warming world where every fraction of a degree matters, that is a blueprint worth copying.

Subject of Research: Patch-scale surface cooling effects of urban green spaces in a humid subtropical Indian city

Article Title: Patch size and vegetation condition influence surface-temperature contrasts around urban green patches in a humid subtropical city

Article References: Singh, S., Verma, P., Behera, S. K., Kumari, A., & Adhikari, D. (2026). Patch size and vegetation condition influence surface-temperature contrasts around urban green patches in a humid subtropical city. Discover Cities, 3(1), Article 170. https://doi.org/10.1007/s44327-026-00356-3

Image Credits: AI Generated

DOI: 10.1007/s44327-026-00356-3

Keywords: urban heat island, land surface temperature, urban green spaces, park cool island, remote sensing, Landsat 8, Sentinel-2, NDVI, patch size threshold, Lucknow, humid subtropical climate, urban planning