Remote sensing technology in the study of lake carbon cycle: Opportunities and challenges

Changchun Huang, Ling Yao, Junsheng Li, Chenghu Zhou, Yulong Guo, Yunmei LiNational 2022 · DOI

This review paper, published in the National Remote Sensing Bulletin, provides a comprehensive look at how satellite and aerial imagery are transforming our understanding of the "active pipes" of the global carbon cycle: our lakes.

TL;DR

Lakes are critical but highly fragmented carbon processors that traditional field sampling cannot adequately monitor at scale. This paper reviews how remote sensing is evolving from simple water-quality monitoring to a sophisticated tool for quantifying global lake carbon budgets, especially when fused with AI and geochemical analysis.

What Problem Does This Paper Address?

For a long time, inland waters (lakes, rivers, reservoirs) were treated as simple "pipes" that moved carbon from land to sea. We now know they are "active pipes"—biogeochemical hotspots that bury carbon in sediment, emit greenhouse gases ($CO_2$ and $CH_4$), and transform organic matter.

However, there are approximately 117 million lakes larger than 0.002 $km^2$ globally. Their distribution is "fragmented" and their ecology is incredibly diverse. Traditional "point-based" sampling (going out in a boat) is too slow, expensive, and limited to provide a global picture. This creates a massive "uncertainty gap" in global carbon budget models. The authors explore how remote sensing (RS) can bridge this gap by providing synchronous, large-scale, and continuous observations.

Methods

The paper reviews a variety of remote sensing inversion techniques used to estimate carbon-related parameters:

  • Bio-optical Modeling: Using the spectral signatures of water to estimate Particulate Organic Carbon (POC), Dissolved Organic Carbon (DOC), and Chlorophyll-a (Chl-a).
  • Greenhouse Gas (GHG) Inversion: Leveraging thermal infrared data to estimate $pCO_2$ and Synthetic Aperture Radar (SAR) to detect methane ebullition (bubbles) in frozen northern lakes.
  • Geochemical Synergy: A novel focus on combining RS with "stable isotope" analysis. While RS provides the quantity of carbon, geochemistry provides the source (e.g., distinguishing between carbon from land-based plants vs. lake algae).
  • Landscape Analysis: Using RS to monitor the surrounding watershed (land use, vegetation productivity, soil erosion) to predict how much carbon is being "pushed" into the lake system.

Datasets

The paper synthesizes results from several decades of satellite missions, including:

  • International Missions: Landsat (long-term series), MODIS (high temporal frequency), MERIS, and the Sentinel-2/3 constellations (high spatial/spectral resolution).
  • Domestic Chinese Missions: HJ (Environment) and GF (High-resolution) series satellites.
  • Cloud Platforms: Extensive mention of Google Earth Engine (GEE) as the primary tool for handling "Remote Sensing Big Data."
  • Literature Benchmark: The authors conducted a bibliometric analysis of 525 core papers from 2000–2020 to track the field's evolution.

Key Insights and Ideas

  • The "Surface" Limitation: A major "aha" moment is the acknowledgment that RS only sees the surface layer. To calculate a true carbon pool, researchers must combine RS surface data with vertical distribution models and water depth data.
  • Source Identification: The authors argue that knowing the type of carbon is as important as the amount. Endogenous carbon (produced in the lake) turns over quickly, while terrestrial carbon (from land) stays longer. Integrating RS with isotopic geochemistry is the key to unlocking these "turnover times."
  • Black Carbon: The paper highlights "Black Carbon" (from incomplete combustion) as a rising parameter of interest. It has unique optical properties that make it a prime candidate for new RS inversion algorithms.
  • From Regression to AI: There is a clear shift from simple linear regressions to machine learning (Random Forest, Deep Learning) to handle the "Case 2" (optically complex) nature of inland waters.

Key Results

  • Research Explosion: The number of papers on RS for lake carbon has grown from ~7 per year in the early 2000s to over 60 per year by 2020.
  • Regional Leadership: China has become a major player, contributing 17.7% of global research in this sub-field, led by the Chinese Academy of Sciences (CAS).
  • Model Success: RS models for $pCO_2$ in lakes like Lake Taihu have shown strong agreement with field data, allowing for the creation of 15-year time-series maps of carbon concentration.
  • The "Active Pipe" Quantified: The paper cites that while land inputs ~5.7 PgC/yr to inland waters, about 68% is emitted as gas, 16% is buried, and only 16% reaches the ocean.

Limitations and Future Work

  • The Resolution Paradox: We currently face a trade-off: satellites with high spatial resolution often have low temporal frequency, and vice versa. The authors suggest multi-source data fusion (combining different satellites) as the only way to achieve "high-frequency, high-resolution" monitoring.
  • 3D Modeling: Future work must move beyond 2D surface maps to 3D volumetric models of lake carbon.
  • Virtual Sensors: The authors propose the development of "virtual sensors"—AI-driven models that can simulate high-quality data even when physical sensor coverage is lacking.
  • Standardization: There is a pressing need for a global, shared platform for lake carbon data to validate RS algorithms across different climatic zones.

Summary updated

Abstract

æ¹–æ³Šç¢³å¾ªçŽ¯æ˜¯å ¨çƒç¢³å¾ªçŽ¯è¿‡ç¨‹ä¸­çš„é‡è¦çŽ¯èŠ‚ï¼Œéšç€å ¨çƒç¢³å¾ªçŽ¯ç ”ç©¶çš„ä¸æ–­æ·±å ¥ï¼Œæ¹–æ³Šç¢³å¾ªçŽ¯å¯¹å ¨çƒç¢³å¾ªçŽ¯çš„å½±å“ï¼Œä»¥åŠå ¶å¯¹å ¨çƒæ°”å€™å˜åŒ–çš„è°ƒèŠ‚ä½œç”¨è¶Šæ¥è¶Šå—åˆ°å ³æ³¨ã€‚ç„¶è€Œï¼Œç”±äºŽæ¹–æ³Šåˆ†å¸ƒçš„ç ´ç¢Žæ€§ï¼ˆå¤§äºŽ0.002 km<sup>2</sup>的湖泊约有1.17×10<sup>8</sup>ä¸ªï¼Œå¹¶é›¶æ˜Ÿåœ°åˆ†å¸ƒåœ¨å ¨çƒï¼‰å’Œå¤šæ ·æ€§ï¼ˆæµåŸŸç”Ÿæ€å¤šæ ·æ€§ï¼Œæ¹–æ³Šç±»åž‹å¤šæ ·æ€§ï¼Œåˆ†å¸ƒçš„æ°”å€™å¸¦å¤šæ ·æ€§ç­‰ï¼‰ï¼Œä½¿å¾—å ¨é¢ç›‘æµ‹å’Œç ”ç©¶å ¨çƒæ¹–æ³Šç¢³å¾ªçŽ¯å ·æœ‰è¾ƒå¤§çš„æŒ‘æˆ˜æ€§ã€‚å ·æœ‰å¤§é¢ç§¯åŒæ­¥è¿žç»­è§‚æµ‹ä¼˜åŠ¿çš„é¥æ„ŸæŠ€æœ¯å¯ä»¥å ‹æœä¼ ç»Ÿè§‚æµ‹æ–¹æ³•çš„å±€é™ï¼Œå¯ä¸ºå ¨çƒæ¹–æ³Šç¢³å¾ªçŽ¯ç ”ç©¶æä¾›å¤§é¢ç§¯åŒæ­¥è§‚æµ‹æ•°æ®çš„æ”¯æ’‘ã€‚åŒæ—¶ï¼Œç”±äºŽå ‰è°±åœ¨ç‰©è´¨è¯†åˆ«å’ŒæŽ¢æµ‹æ–¹é¢çš„ä¼˜åŠ¿ï¼Œä½¿å¾—é¥æ„ŸæŠ€æœ¯åœ¨æœ‰æœºè´¨ç±»åž‹åæ¼”æ–¹é¢ä¸Žåœ°çƒåŒ–å­¦æ–¹æ³•å­˜åœ¨ç»“åˆçš„å¯èƒ½ã€‚æœ¬æ–‡å›žé¡¾äº†ç›®å‰æ°´çŽ¯å¢ƒé¥æ„Ÿç ”ç©¶ä¸­ä¸Žæ¹–æ³Šç¢³å¾ªçŽ¯ç›¸å ³çš„æ¹–æ³Šä¸åŒç±»åž‹ç¢³æµ“åº¦ã€æ°´ä½“ç†åŒ–å‚æ•°ç­‰é¥æ„Ÿåæ¼”ç®—æ³•åŠå ¶åº”ç”¨çš„çŽ°çŠ¶ï¼Œç»“åˆæ¹–æ³Šç¢³å¾ªçŽ¯ä¸­æœ‰æœºç¢³è¿ç§»è½¬åŒ–çš„ç”Ÿç‰©åœ°çƒåŒ–å­¦è¿‡ç¨‹ï¼Œä»¥åŠæ¹–æ³Šç¢³å¾ªçŽ¯ç ”ç©¶ã€é¥æ„Ÿå¤§æ•°æ®å’Œäººå·¥æ™ºèƒ½çš„å‘å±•ï¼ŒæŽ¢è®¨äº†æ¹–æ³Šç¢³å¾ªçŽ¯ç ”ç©¶ä¸­é¥æ„ŸæŠ€æœ¯åº”ç”¨çš„æœºé‡å’ŒæŒ‘æˆ˜ã€‚
BibTeX
@inproceedings{huang2022remote,
  author = {Changchun HUANG and Ling Yao and Junsheng Li and Chenghu Zhou and Yulong GUO and Yunmei LI},
  title = {{Remote sensing technology in the study of lake carbon cycle: Opportunities and challenges}},
  booktitle = {National},
  year = {2022},
  doi = {10.11834/jrs.20221220},
}
Figure