Skip to main content
geofurkan logo
SPATIAL TIME-SERIES RESEARCHGoogle Earth EngineArcGIS ProLandsat 5 TM & Sentinel-2

40-Year Spatial Change in Akgöl Wetland: The Role of Climatic Factors and Anthropogenic Pressures

Multi-spectral satellite reconstruction (1984–2025) of the ecological collapse of Ereğli Akgöl Reeds in the Konya Closed Basin. Combining Landsat & Sentinel-2 archives on Google Earth Engine with dynamic Otsu thresholding to dissect how upstream dam construction, deep aquifer overdraft, and meteorological drought eliminated 99.47% of the open water surface.

Furkan Davut DEMİR·Geographer
1984–2025 Time Series·Konya Closed Basin
11.667 ha
1984 Baseline Water
-%71,55
1984–90 Shock Loss
-15,9 m
Groundwater Drop (SGI)
61,35 ha
2025 Remaining Core
01. QUANTITATIVE TIME-SERIES MODEL

40-Year Water Mirror Contraction Dynamics (1984–2025)

Multi-temporal quantitative time-series derived on Google Earth Engine using 30x30m Landsat surface reflectance and dynamic Otsu thresholding. Hover across years to inspect hydrological shock points and satellite sensor archives.

INTERACTIVE TIME-SERIES VISUALIZATION

1984–2025 Multi-Decadal Contraction Dynamics

Selected Year
2025
Water Surface
61,4 ha
Loss vs 1984
-%99.5
Satellite Archive
Sentinel-2 MSI & Landsat 8
0 3.000 6.000 9.000 12.000 198419901995200020052010201520202025
2025 Hydrological Milestones·Sentinel-2 MSI & Landsat 8

Ecological Collapse: Lake confined to 6,700m artificial levee impoundment remnant (DSİ)

Milestones:
02. CARTOGRAPHIC & SPECTRAL VISUALS

High-Resolution Spatial Documentation

1984–2025 Akgöl Multi-Temporal Change Map
1984–2025 Multi-temporal water masks: Blue (1984: 11,667 ha) → Yellow (2000: 751 ha) → Orange (2010: 86 ha) → Red (2025: 61 ha).
03. ECOLOGICAL HERITAGE & BIODIVERSITY

Ramsar-Grade Biodiversity Lost to Anthropogenic Desiccation

Historically commanding an expansive watershed influence of approximately 21,500 hectares, Akgöl Wetland lost almost all of its surface hydrological recharge following the construction of upstream storage dams (İvriz, Ayrancı, and Gödet) and massive agricultural groundwater overdraft, transforming an open water mirror into a dry, salt-crusted depression (Tapur, 1998; Ceyhun et al., 2001; Zeybek & Gür, 2006).

From a biodiversity perspective, the Akgöl Reedbeds harbored an extraordinarily rich flora and fauna, serving as a genetic refuge of global ecological standing (Zeybek & Gür, 2008). Situated directly along one of the two major avian migratory flyways crossing Turkey, the wetland functioned as an indispensable stopover, staging, breeding, and feeding haven for migratory waterfowl (ÇŞDTVKGM, 2014; WWF, 2016).

Official Statutory Conservation Chronology
1992
1st Degree Natural Site

Designated as a strict natural reserve due to unique landscape geomorphology.

1995
Strict Nature Reserve

6,800-hectare contiguous core reedbed ecosystem granted absolute protection status.

Ramsar Criteria
#1 Bird Sanctuary in Turkey

Ranked #1 among 78 critical avian sanctuaries, meeting full Ramsar criteria.

Threatened Endemic Freshwater Fish of the Basin (Güçlü et al., 2023)
Anatolian MinnowEndangered (Endemic)
Pseudophoxinus anatolicus

Critically endemic species restricted to shallow, fresh-to-brackish habitats across the Konya Basin.

Ereğli Stone LoachCritically Endangered (CR)
Garra kemali

Local ichthyological heritage documented exclusively in Akgöl and its connected drainage channels.

Central Anatolian MinnowVulnerable (VU)
Pseudophoxinus stymphalicus

Breeds in littoral reed beds and benthic substrates, severely threatened by persistent water retreat.

04. TIME-SERIES CHRONOLOGY

Quantitative Dynamics: 11,667 ha Down to 61.35 ha

Surface reflectance processing from Landsat 5 TM and Landsat 8 OLI summer archives (July 01 - August 31) using MNDWI and dynamic Otsu thresholding reveals that wetland loss occurred via sudden engineering threshold collapses rather than gradual linear drying:

EpochWater Area (Hectares)Period ChangeLoss vs 1984Ecological Status
198411.667,61 haBaseline-%0Expansive open water surface, 21,500 ha watershed impact, contiguous reedbeds.
19903.322,76 ha-%71,55-%71,55Immediate sharp desiccation following the completion of İvriz (1985) & Gödet (1988) dams.
2000751,05 ha-%77,40-%93,57Severe reed ecotone fragmentation; open water retreats into shallow disconnected micro-basins.
201086,76 ha-%88,45-%99,26Historic minimum dry bed coinciding with intense regional drought and unmonitored deep wells.
2020246,77 ha+%184,43-%97,89Transient anomaly driven by wet spring rains and irrigation drainage channel discharges.
202561,35 ha-%75,14-%99,47Complete ecological collapse. Lake confined to a 6,700 m artificial levee impoundment (DSİ).
05. ROOT CAUSES

The Lethal Combination: Upstream Dams & Groundwater Overdraft

1. Upstream Storage Dams (Surface Inflow Cutoff)

Akgöl was a terminal closed-basin wetland with no natural outlet, replenished exclusively by runoff descending from the Bolkar Mountains. Construction of the İvriz Dam (1985) and Gödet Dam (1988) impounded hundreds of millions of cubic meters of water for intensive irrigation, severing the lake’s primary lifelines. The catastrophic 71.55% surface collapse between 1984 and 1990 was the direct outcome of these civil engineering interventions.

2. Deep Aquifer Overdraft & Sinkhole Acceleration

Once surface streams were impounded, regional agriculture pivoted toward water-intensive monocultures (maize and sugar beet), causing an explosion in groundwater exploitation. Across tens of thousands of registered and illegal deep wells, the regional static water table plummeted by an average of 15.9 meters (56.8 cm/year). Deprived of upward baseflow recharge, the desiccated lake floor collapsed, accelerating regional sinkhole formation.

Major Upstream Hydraulic Barriers in the Basin

İvriz Dam1985
River: İvriz Stream
Capacity: 83,0 hm³

Retained the primary surface recharge of Akgöl for agricultural irrigation, triggering an immediate 71.55% surface collapse within 5 years.

Gödet Dam1988
River: Gödet Stream
Capacity: 158,0 hm³

Sealed western sub-basin runoff and prevented flash-flood recharge from reaching the southern marshlands.

Ayrancı Dam1958
River: Kocadere River
Capacity: 30,9 hm³

Early hydrological barrier that retained seasonal runoff from the Karaman-Ayrancı southern slopes.

06. REMOTE SENSING METHODOLOGY

Why MNDWI Instead of Conventional NDWI in Shallow Marshes?

Employing a single spectral index in complex wetland ecotones introduces severe misclassification errors. In shallow basins such as Akgöl with dense reed canopy (Phragmites australis) and summer salt-crusted mud, the McFeeters (1996) NDWI misclassifies saturated soil as open water while mislabeling emergent reeds as dry land. Xu’s (2006) MNDWI replaces NIR with the SWIR1 band, which exhibits near-total water absorption, cleanly suppressing vegetation and soil reflectance to achieve superior delineation.

MNDWI (Xu, 2006)
MNDWI = (Green - SWIR1) / (Green + SWIR1)

Water pixels exhibit maximum absorption in SWIR wavelengths, effectively suppressing reed shadows and dry soil noise.

Otsu Dynamic Thresholding (1979)
σ²_B(t) = ω₀(t) · ω₁(t) · [μ₀(t) - μ₁(t)]²

Automated inter-class variance maximization across image histograms instead of arbitrary static index cutoffs (optimum 0.175 for 1985).

AWEI (Feyisa et al., 2014)
AWEI = 4 × (Green - SWIR1) - (0.25 × NIR + 2.75 × SWIR2)

Multi-band weighted formulation optimized to suppress dark mountain shadow surfaces and built-up urban noise.

NDMI (Wilson & Sader, 2002)
NDMI = (NIR - SWIR1) / (NIR + SWIR1)

Normalized Difference Moisture Index used as a cross-verification layer to distinguish saturated mudflats from true open water bodies.

07. SPATIAL ACTION PLAN

4 Steps to Prevent Irreversible Desertification

1. Automated Sentinel-2 GEE Early Warning System

Establish a cloud-based automated early warning framework on Google Earth Engine that calculates wetland water surface and moisture budget over each 15-day Sentinel-2 revisit cycle.

2. Environmental Baseflow & DEM Bathymetric Routing

Enforce minimum environmental flow releases from İvriz Dam, routed via Digital Elevation Model (DEM) micro-topography into optimum shaded depressions to minimize evaporative losses.

3. Machine Learning Crop Classification & Well Auditing

Deploy Random Forest parcel-level crop classification to detect illegal water-thirsty monocultures (corn, sugar beet), enforcing the closure of unpermitted deep extraction wells.

4. Halophyte Vegetative Windbreaks Against Dust Storms

Plant indigenous halophytic and xerophytic vegetative shelterbelts across the 11,000-hectare dried lake bed to prevent severe regional saline dust storms and wind erosion.

Have Questions, Observations or Feedback?

Feel free to get in touch if you have observations on this study, inquiries about the spatial methodology, or questions about the research findings.

Get in Touch