Groundwater Recharge Assessment and Recharge Zonation of the Intermontane Groundwater Basin, Chiang Mai, Thailand, Using a Groundwater Flow Model and Stable Isotopes

Abstract

Urbanization, escalating agriculture, tourism, and industrial development in the Chiang Mai–Lamphun groundwater basin in northern Thailand have increased water demand, causing widespread groundwater extraction. Over the past few decades, there has been a rapid, unrecoverable steady drop in groundwater levels in several areas in Chiang Mai and Lamphun provinces. This study employed hydrogeological investigations, hydrometeorological data analyses, stable isotopic analysis (δ¹⁸O and δ²H), and groundwater flow modeling using a 3D groundwater flow model (MODFLOW) to quantify groundwater recharge and delineate important groundwater recharge zones within the basin. The results showed that floodplain deposits exhibited the highest recharge rate, 104.4 mm/y, due to their proximity to rivers and high infiltration capacity. In contrast, younger terrain deposits, covering the largest area of 1314 km², contributed the most to total recharge volume with an average recharge rate of 99.8 mm/y. Seven significant recharge zones within the basin, where annual recharge rates exceeded 105 mm/y (average recharge of the entire basin), were also delineated. Zone 4, covering parts of densely populated Muaeng Lamphun, Ban Thi, and Saraphi districts, had the largest area of 330 km² and a recharge rate of 130.2 mm/y. Zone 6, encompassing Wiang Nong Long, Bai Hong, and Pa Sang districts, exhibited the highest recharge rate of 134.6 mm/y but covered a smaller area of 67 km². Stable isotopic data verified that recent precipitation predominantly recharged shallow groundwater, with minimal evaporation or isotopic exchange. The basin-wide average recharge rate was 104 mm/y, reflecting the combined influence of geology, permeability, and spatial distribution. These findings provide critical insights for sustainable groundwater management in the region, particularly in the context of climate change and increasing water demand.

1. Introduction

The main source of fresh water below the Earth’s surface is groundwater, which is essential for domestic consumption, industrial supplies, and agricultural purposes in northern Thailand. However, excessive groundwater consumption is causing groundwater heads to decline quickly, particularly at locations where surface water is not available. To protect future groundwater supplies, it is crucial to assess groundwater recharge and protect important recharge zones. Chiang Mai and Lamphun provinces experienced a notable increase in the urbanized areas between 1998 and 2018, primarily due to the city’s rapid urbanization brought on by industrialization, tourism, and migration. Previously, the city had a single central hub, but now it has many urban centers [1].

The Chiang Mai–Lamphun groundwater basin (Figure 1) is an intermontane valley situated between mountains, sharing parts of Chiang Mai and Lamphun provinces in northern Thailand. Outside the cities, most people in the area mainly rely on groundwater for their water needs [2]. The Chiang Mai–Lamphun basin is the largest Cenozoic basin in northern Thailand. The main river in this basin, the Ping River, is a key tributary of the Chao Phraya River and generally flows from north to south. The region receives an average of 1800 mm/y rainfall annually, with the rainy season usually occurring between June and October.

Chiang Mai, located in northern Thailand, is the country’s second-largest province, with a population of 2 million people [2]. This population increase is anticipated to result in a decline in groundwater storage and the corresponding hydraulic heads in the region. Groundwater levels have significantly dropped, particularly in the southern areas (Hang Dong and San Pa Tong districts) and the northern and northeastern regions (San Kamphaeng and Mae Rim districts), due to agricultural expansion and population growth [3,4,5]. Recently, land use in the Chiang Mai–Lamphun basin has shifted, with more areas being developed for urban and agricultural purposes as the population has grown, leading to increased groundwater usage. Additionally, the Royal Thai Government’s nationwide groundwater exploration and development projects commenced during 2005–2010 that aimed at combating drought have resulted in an increase in groundwater usage from 18.7 Mm³/y in 2005 to 24.0 and 33.6 Mm³/y in 2015 and 2020, respectively [3,4,5,6].

Protecting groundwater recharge zones is crucial for keeping groundwater levels stable, especially in areas where water demand is high due to population growth and agricultural activities [7]. Few studies have been conducted to evaluate the groundwater potential and the effects of groundwater exploitation in the Chiang Mai–Lamphun basin [8,9]. According to forecasts from Thailand’s Department of Groundwater Resources [5], groundwater demand of the Chiang Mai–Lamphun basin is expected to increase from the current usage rate of 44 Mm³/y in 2024 to 55 and 65 Mm³/y as of 2028 and 2035, respectively. This increase in water demand corresponds to an increase in population, which has been projected to reach 2.4 and 3.6 million in 2028 and 2035, respectively [2].

Groundwater flow models have increasingly been used as tools for advancing sustainable groundwater management practices [10]. Groundwater flow mechanics through porous materials is represented by partial differential equations, which can be approximated by using groundwater models [11]. These models are essential resources for assessing the capacity of an aquifer and modeling its response to pumping stress [12]. For this reason, several numerical simulation codes have been developed, employing techniques such as the finite difference and finite element approaches [13]. The most used groundwater modeling code is MODFLOW, which is based on the finite difference method.

According to the Tourism Authority of Thailand as of December 2024 (TAT), Chiang Mai, the largest province by area, was ranked as the third most-visited destination in Thailand, attracting a total of 6.38 million domestic and international visitors, compared to 5.6 million in 2019. The province is experiencing growth, increasing its demand for groundwater. Therefore, protecting the areas that contribute most to groundwater recharge is essential, which can be further utilized for artificial replenishment. This research will help identify the most vulnerable areas for groundwater recharge, contributing to sustainable groundwater management.

This study presents research work that combines field survey hydrogeological data (e.g., hydraulic head measurements, river stage measurements, and groundwater sample for stable isotopes analysis) and a comprehensive groundwater flow model to evaluate groundwater recharge and delineate important recharge zones of the Chiang Mai–Lamphun basin. Hydraulic head data from 48 piezometers acquired between April and August 2024 were used for calibration targets. PEST program [9,14] was used in conjunction with MODFLOW to calibrate the flow model. The pilot point model calibration approach helped to delineate important areas or zones where groundwater recharge was crucial and assess yearly recharge variability.

2. Background Information

2.1. Geology

The Mesozoic (221–210 Ma) and late Cenozoic (21–16 Ma) events, including subduction, accretion, collision, and basin formation, generated a complex geological history that included the Chiang Mai–Lamphun basin, which is a portion of the greatest Cenozoic basin in northern Thailand. The region is situated between Doi Inthanon and Doi Suthep to the west and has experienced geological changes, such as the Doi Khun Tan Batholith’s existence to the east and the truncation of the Chiang Mai Low-Angle Normal Fault (CMLANF) during the Eocene–Oligocene period [15,16,17].

There are many kinds of rocks and complex geological structures in the basin because of this diverse geological past [15,18,19,20]. The rocks of the Chiang Mai–Lamphun basin are diverse and include Permian limestone, Permo-Triassic volcanic rocks, Carboniferous sedimentary rocks, Silurian–Devonian metamorphic and sedimentary rocks, Ordovician limestone and other sedimentary rocks, and Triassic granitic rocks and migmatites (Figure 2).

Precambrian metamorphic rocks, such as pegmatites, mica schists, calc-silicates, and orthogneisses, are essential for determining the amount and direction of geological extension. These gneisses have been intruded by substantial quantities of S-type granitic rocks, many of which date back to the Triassic period, according to earlier research [21,22,23]. Paleozoic sedimentary and low-grade metasedimentary rocks, primarily phyllitic shales and limestones from the continental shelf, lie on top of the crystalline complex. During the Permo-Triassic collision between the Indosinian Craton and the Shan Thai terrane, these rocks most likely underwent metamorphism, deformation, and intrusion by granitoid [24].

2.2. Hydrogeology

Hydrogeological units are categorized into three primary types: unconsolidated, semi-consolidated, and consolidated rocks. Numerous geological units, from the Precambrian to recent periods, are seen in Figure 3. Much of the Chiang Mai–Lamphun basin is covered by older rock formations on which unconsolidated Quaternary alluvium deposits rest in an unconformable manner [25,26,27]. With clayey strata scattered throughout the sequence, the core portion of the basin is primarily made up of silts, sands, and gravels carried by the major rivers. The average depth of groundwater wells in this region is between 20 and 70 m below the surface, making them comparatively shallow. Groundwater extraction is most prevalent on the central alluvial channel. According to pumping measurements, these aquifers’ hydraulic conductivities varied from 0.01 to roughly 20 m per day [8]. The total dissolved solids (TDSs) in groundwater are below 250 mg/L, demonstrating typically high quality. Because of higher oxygen levels and faster flow rates, fluoride and iron concentrations are often low, particularly in the upper portion of the aquifer [3,4,5].

Complex interactions between geological formations, aquifer characteristics, and groundwater recharge define the hydrogeology of the Chiang Mai–Lamphun basin in northern Thailand. Estimates of groundwater recharge vary from 597.2 to 896.2 mm/y, depending on land use, slope, and rainfall [28]. Significant drops in groundwater levels, especially in metropolitan areas, indicate that the basin’s unconsolidated aquifers have poor potential [29]. Groundwater vulnerability has been evaluated using the DRASTIC approach, which highlights locations that need cautious management [30]. Furthermore, a history of tectonic activity that influences hydrology and groundwater flow patterns is shown by the geological structure, which includes the Mae Rim Formation [31].

Figure 3. Hydrogeological units of the Chiang Mai–Lamphun basin (modified from [29]).

Figure 3. Hydrogeological units of the Chiang Mai–Lamphun basin (modified from [29]).

2.3. Climate and Meteorological Data

Rainfall distribution in the Chiang Mai–Lamphun basin shows significant seasonal variability, which is typical of tropical monsoon climates. Figure 4 indicates that the highest precipitation occurs during the monsoon season (May–October), peaking in September (220.9 mm) and August (209.6 mm). In contrast, the dry season (November–April) experiences minimal rainfall, especially in February (17.7 mm) and January (26.3 mm). This variability has direct implications for groundwater recharge. Intense monsoon rains likely lead to short-term recharge events, while extended dry periods increase reliance on stored groundwater, heightening stress on aquifers.

The temperature data from the Chiang Mai–Lamphun groundwater basin over the past three decades (1995–2024) reveal a clear warming trend, highlighting the broader implications of climate change at the regional level. Annual temperatures increased from about 26.7 °C in 1995 to 28.7 °C in 2024, representing a rise of approximately 1.2 °C over 30 years (increasing rate of 0.04 °C/y) (see Figure 5). This trend is consistent with global observations of increasing temperatures but also underscores localized vulnerabilities in this essential groundwater-dependent region.

2.4. Previous Modeling Studies

The groundwater modeling of the Chiang Mai basin has been the subject of several studies. According to a one-layer groundwater modeling study by Tatong [33], a 20% increase in groundwater extraction could reduce groundwater levels in the colluvial aquifer from 7 to 12 m. Based on the three-dimensional groundwater flow model established in DGR’s study [8], the basin’s safe yield and yearly water budget were 125 and 255 Mm³/y, respectively. Saenton [34], however, employed stochastic groundwater flow modeling approaches and found that the annual water budget was approximately 241 Mm³/y, with a variance of 12 Mm³/y. The results of the transient groundwater flow modeling study by Taweelarp et al. [29], which was performed using MODFLOW, showed that the Chiang Mai–Lamphun basin has a limited safe yield of 51.2 Mm³/y, which is significantly less than current extraction rates and calls for sustainable groundwater management.

2.5. Previous Recharge Assessment Studies

Intrasuta [35] employed flow net analysis to estimate the overall recharge rate of the Chiang Mai basin. This analysis is based on Darcy’s law, which states that three variables influence the flow rate through porous material: transmissivity, hydraulic gradient, and the distance perpendicular to the flow stream. The recharge rate for the entire region can be determined using Darcy’s equation for both high and low transmissivity values. The minimum and maximum recharge rates were found to be 17 mm/y and 143 mm/y, respectively.

Uppasit [36] calculated the groundwater recharge of the Chiang Mai basin using the water table fluctuation (WTF) method. The WTF approach is based on the concept that recharge water or effective rainfall reaching the water table causes increases in groundwater levels. The specific yield (S_y) is multiplied by the overall rise in the groundwater table over a specified period to calculate groundwater recharge. The average estimated groundwater recharge in this study was 349.5 Mm³/y or 126.1 mm/y, accounting for approximately 11% of the typical annual precipitation.

2.6. Application of Stable Isotopes

As surface water and groundwater are components of a single, combined water resource system, their interactions must be identified for effective water management, especially in basins that face water challenges. To understand the relationship between surface water (SW) and groundwater (GW), environmental tracers, such as stable isotopes of hydrogen and oxygen, radon, and various chemical compounds, have been employed [37,38]. Specifically, deuterium (δ²H) and oxygen-18 (δ¹⁸O) isotopes are valuable for pinpointing water sources and revealing hydrological processes and isotopic fractionation [39,40]. The Chiang Mai–Lamphun basin lacks sufficient stable isotopic studies on surface water–groundwater interaction, despite its susceptibility to water problems and hydrological changes. For this reason, this research incorporates stable isotope analysis into groundwater recharge assessment to understand SW-GW interaction, as well as the source of recharge, for a better understanding of the basin’s hydrogeology.

3. Materials and Methods

3.1. Conceptual Model

The conceptual model was a digital 3D hydrogeological representation of the physical groundwater system organized in a computerized format. Lithologic logs, borehole investigations, and comprehensive field surveys were conducted to compile hydrogeological data for the conceptual model. Three hydrostratigraphic units, namely Qfd, Qyt, and Qot, were identified in the Chiang Mai–Lamphun basin, with deposits ranging from semi-consolidated to unconsolidated. The total thickness of floodplain (Qfd) deposits is 0–50 m, with the upper unconfined portion having a thickness of 30 m and the lower portion designated as semi-confined to confined. The low terrace deposit (Qyt) aquifer has a variable thickness (0–150 m), which is classified as an unconfined aquifer, with a thickness of around 30 m located in the east and west parts of the floodplain deposits, and the deeper aquifer located beneath the Qfd is confined. The high terrace deposit (Qot) aquifer includes Qcl, which has an average 120 m thickness. Approximately 30 m is unconfined, and the aquifer beneath the low terrace deposits is semi-confined to confined.

Groundwater generally flows to the center of the basin, as can be seen in Figure 6. To balance groundwater in- and outflow, the boundaries in the north and south were replaced with the general head boundary (GHB) in MODFLOW [41]. Taweelarp et al. [29] conducted more than 200 pumping tests and analyzed datasets for hydraulic conductivities listed in

Table 1. Initial parameter values for the conceptual model.

Parameter Type	Parameter Name		Initial Values	Note
Hydrostratigraphic Units	Qfd	HK_1	0.15–18 m/d	Pilot points
	Qyt	HK_2	0.035–30 m/d
	Qot + Qcl	HK_3	0.043–63 m/d
	Lower aquifers	HK_4	0.12–13.5 m/d
Vertical anisotropy	-	VANI	0.1	-
Recharge rate	-	RCH	100 mm/y	Pilot points
Max evapotranspiration (ET) rate	-	EVT	83.5 mm/y
Rive-bed conductance factor	Ping River	RIV_1	0.25 m2/m/d	-
	Kuang River	RIV_2	0.25 m2/m/d
North and south GHB conductance factors	-	GHB	0.25 m2/m/d	-

, which were used in this research.

3.2. Data Collection Method

Hydraulic head data were collected from 48 observation wells, distributed across the Chiang Mai–Lamphun basin, in April and August 2024 (minimum and maximum hydraulic heads), representing measurements in dry and wet seasons, respectively. The measurements were taken using a water level meter, which records the groundwater depth from the top of the well casing. The elevation of each well casing was determined using a GPS device to reference the average mean sea level (amsl.). The hydraulic head was then calculated to show the difference between the well casing elevation and the depth of water. This dataset provides critical information on groundwater levels and flow patterns, enabling the calibration and validation of the MODFLOW groundwater model. The seasonal variation in hydraulic head measurements also offers insights into the dynamic behavior of the aquifer system in response to recharge and extraction. Note that there are more than 2500 pumping wells (Figure 7) in the Chiang Mai–Lamphun basin, which are being used for various purposes such as industrial, agricultural, and residential. The number of wells and their extraction rate vary annually. In 2019, their combined extraction rate was 29.8 Mm³/y [29] and was expected to increase to approximately 40–50 Mm³/y in 2023–2024.

3.3. Groundwater Flow Model Setup

The conceptual model was converted into a 3D finite difference numerical model (MODFLOW), which had 71 columns, 80 rows, and 4 layers of variable thickness, with a grid size of 1 × 1 km². The model simulated the steady-state hydraulic head (h) using Equation (1). The materials in the model included Qfd: floodplain unconsolidated sediment aquifer unit; Qyt: young terrace unconsolidated sediment aquifer unit; and Qot: high terrace unconsolidated and semi-consolidated and lumped lower aquifer unit. Modeling packages that were used in this simulation included layer–property flow (LPF), recharge (RCH), evapotranspiration (EVT), river (RIV), multi-node well (MNW), and general head boundary (GHB).

$${\displaystyle \frac{\partial }{\partial x}}\left(K_x{\displaystyle \frac{\partial h}{\partial x}}\right)+{\displaystyle \frac{\partial }{\partial y}}\left(K_y{\displaystyle \frac{\partial h}{\partial y}}\right)+{\displaystyle \frac{\partial }{\partial z}}\left(K_z{\displaystyle \frac{\partial h}{\partial z}}\right)\pm W=0,$$

(1)

where K is the hydraulic conductivity (m/d), and W refers to the internal source/sink (1/d).

3.4. Model Calibration

Aquifer’s hydraulic conductivity, recharge, and evapotranspiration are examples of spatially variable qualities that can be estimated using the pilot point technique [41]. Pilot points are used to estimate areas with high and low differences in hydraulic properties, which are inferred during the calibration process, rather than designating zones with homogeneous attributes [42]. Using a pilot point technique to reflect aquifer heterogeneity and regional differences in recharge and evapotranspiration, PEST [9,14] was utilized to help calibrate the highly parameterized model and estimate the aquifer parameter values at each of the pilot points that were added to the model domain (Figure 8). Aquifer heterogeneity was then simulated by employing an algorithm to interpolate these values spatially across all active cells in the model.

During model calibration, PEST iteratively updates parameter values to minimize the sum of weighted squared residuals ($\Phi$):

$$\Phi ={\displaystyle \sum _{i=1}^N{\omega }_i{\left({\tilde{h}}_i-h_i\right)}^2},$$

(2)

where $h_i$, ${\tilde{h}}_i$, ${\omega }_i$ and N represent the measured data (such as hydraulic heads), simulated equivalents, weights of observation $i$, and the number of measurements, respectively. The extent to which parameter values are adjusted at each iteration depends on the sensitivity of the parameter. Quantitatively, the sensitivity of a parameter $P_j$ or $\partial \Phi /\partial P_j$, which indicates the parameter’s relative influence on model output, can be estimated using the forward difference perturbation technique as follows:

$${\displaystyle \frac{\partial \Phi }{\partial P_j}}\approx {\displaystyle \frac{\Phi \left(P_j+\Delta P\right)-\Phi \left(P_j\right)}{\Delta P}}$$

(3)

Following each iteration and after the calibration procedure, PEST reports the parameter sensitivity. To evaluate the goodness of fit, the RMSE (root mean squared error) and NRMSE (normalized root mean squared error) values, determined using Equation (4), are usually used as indicators. According to Anderson et al. [43], if the RMSE is less than 10% of the intended heads, the calibrated model is deemed satisfactory.

$$RMSE=\sqrt{{\displaystyle \frac{1}{N}}{\displaystyle \sum _{i=1}^N{\left({\tilde{h}}_i-h_i\right)}^2}}, \mathrm{and} NRMSE(\%)=100\times \sqrt{{\displaystyle \frac{1}{N}}{\displaystyle \sum _{i=1}^N{\left(\frac{{\tilde{h}}_i-h_i}{h_i}\right)}^2}}$$

(4)

3.5. Water Sample Collection for Stable Isotope Analysis

Groundwater, rainwater, and surface water samples were collected throughout the wet season (late July to early August 2024) to study the isotopic composition of δ¹⁸O and δ²H. A total of 10 groundwater samples were collected from wells with depths varying from 16 to 28 m below ground surface (bgs), and 4 surface water samples were collected along the Ping River (Figure 9). Groundwater samples were extracted using a submersible pump, filtered through a 0.45 μm nylon filter, and stored in 50 mL HDPE bottles. Rainwater samples were collected at three locations during the mid-rainy season using a rainfall collector device designed to minimize evaporation loss, following the methodology by Deshpande et al. [44]. The stable isotope composition of water (δ¹⁸O and δ²H) was determined using a liquid water isotope analyzer (PICARRO L2130-i model) (Picarro, Inc., Santa Clara, CA, USA). Measurement errors (precision) in estimation were within the limits of ±0.1 ‰ for δ¹⁸O and ±1‰ for δ²H.

4. Results

4.1. Groundwater Level and Flow Direction

Figure 10 displays a contour plot of the hydraulic head measured in April and August 2024. The flow direction is toward the center of the basin. A few cones of depression can be seen in the south of the basin, especially in the Doi Lo, Pa Sang, Wiang Nong Long, and Muaeng Lamphun districts, depicting a low water table. On the other hand, the basin’s northeastern part, which includes the Mae Rim, San Sai, and Muaeng Chiang Mai districts, has a high water table.

4.2. Groundwater Modeling and Calibration Results

4.2.1. Simulated Head

The groundwater flow model was calibrated with the pilot point technique. Figure 11 compares the simulated and measured hydraulic heads, indicating that most values fell within the 95% confidence interval. The model’s reliability was further supported by an RMSE of 3.01 and NRMSE of 3.86%, implying a good match between observed and simulated data.

Figure 12 illustrates the simulated hydraulic head distribution of all model layers. The color gradient, ranging from blue (low head) to red (high head), shows the groundwater flow direction—from the northern and northwestern regions with high head toward the central and southern areas with low head. This pattern is typical of intermontane basins, where groundwater discharges into rivers and tributaries in the central lowlands.

Residual analysis revealed the difference between measured and simulated heads and confirmed that most residuals fell within the acceptable range of ±5 m. Nevertheless, a few outliers, such as wells G265 (–16.5 m) and Q49 (+9.9 m), indicated localized discrepancies. Despite these outliers, the overall model performance was strong, with RMSE and NRMSE values confirming its accuracy in simulating the basin’s groundwater system.

4.2.2. Groundwater Budget and Recharge Assessment

Table 2. Groundwater budget after steady-state simulation.

Components	Inflow (Mm3/y)	Outflow (Mm3/y)
Recharge	258.9	-
Evapotranspiration	-	226.5
Pumping wells	-	44.4
General Head Boundary	13.5	11.7
Rivers Leakage	11.4	1.5
Total (Mm3/y)	283.8	284.0

shows the results of groundwater budget calculation from the calibrated groundwater flow model. The water budget analysis pointed to a near-equilibrium relationship between the total inflow of 283.8 Mm³/y and the outflow of 284.0 Mm³/y, with a slight difference of 0.2 Mm³/y, indicating a well-calibrated model. Recharge was the dominant inflow component, contributing 258.9 Mm³/y (91.2%), equivalent to 104.0 mm/y. Additional inflows included 13.5 Mm³/y from the general head boundary (GHB), representing external groundwater sources, and 11.4 Mm³/y from river leakage, suggesting that the river primarily acts as a losing stream. In outflows, evapotranspiration (ET) dominated, which accounted for 226.5 Mm³/y (80% of total outflow), reflecting the combined effects of soil evaporation and plant transpiration. The actual ET, interpolated from seven ET measurements across the basin, was approximately 207 Mm³/y [32]. This discrepancy could be attributed to the groundwater model approximation or interpolation scheme of the measurements. Groundwater abstraction for irrigation, industrial, and domestic use totaled 44.4 Mm³/y (15.6% of outflow), while GHB and river leakage contributed 11.7 Mm³/y and 1.5 Mm³/y, respectively, to outflow.

Recharge rates varied across hydrostratigraphic units, with floodplain deposits (Qfd) exhibiting the highest rate of 104.41 mm/y due to their proximity to rivers and high infiltration capacity. Younger terrain deposits (Qyt), covering the largest area of 1314 km², contributed the most to the total recharge volume despite a slightly lower rate of 99.8 mm/y. Older terrain deposits (Qot) showed a moderately high recharge rate of 103.2 mm/y but contributed less overall due to their smaller area. The basin-wide average recharge rate was 101.6 mm/y, reflecting the combined influence of geology, permeability, and spatial distribution. These findings underscore the importance of delineating and protecting high-recharge zones, particularly floodplain deposits, for sustainable groundwater management.

4.3. Important Recharge Zonation

Recharge zones with annual recharge exceeding 104 mm/y (i.e., an average recharge) were mapped to locate areas of high infiltration efficiency, based on the basin’s average recharge rate of 104.0 mm/y (

Table 3. The significant recharge zones in the Chiang Mai–Lamphun groundwater basin.

Zones	Total RCH (Mm3/y)	Area (km2)	Average (mm/y)
Zone 1	13.0	117	111.4
Zone 2	1.0	9	112.0
Zone 3	26.9	240	112.1
Zone 4	42.9	330	130.2
Zone 5	1.5	14	109.1
Zone 6	9.0	67	134.6
Zone 7	22.3	97	230.3
Total	258.9	2490	104.0

). Spatially, recharge values increased toward the center of each polygonal zone, reflecting localized variations in soil permeability, proximity to rivers, and rainfall patterns (Figure 13). Zone 4, covering parts of Mueang Lamphun, Ban Thi, and Saraphi districts, had the largest area of 330 km² and a high recharge rate of 130.2 mm/y. Zone 6, encompassing areas of Wiang Nong Long, Bai Hong, and Pa Sang districts, exhibited the highest recharge rate of 134.6 mm/y but covered a smaller area of 67 km². Zone 7 showed an exceptionally high recharge rate of 230.3 mm/y, likely due to unique hydrogeological conditions. The remaining zones (1, 2, 3, and 5) had recharge rates ranging from 109.1 mm/y to 112.1 mm/y, with smaller areas contributing less to the total recharge.

4.4. Implication of Stable Isotopes on Groundwater Recharge

Stable isotopes of δ¹⁸O and δ²H were utilized to assess the origin or source of recharged groundwater in the Chiang Mai–Lamphun basin. Precipitation samples from the Regional Irrigation Office 1 (located in the Chiang Mai city center) rainfall station were analyzed, and their isotopic compositions were plotted along the Local Meteoric Water Line (LMWL) (Equation (5)), derived from data collected between 2001 and 2023. The LMWL closely aligns with the Global Meteoric Water Line (GMWL) (Equation (6)), although small differences in the intercept indicate localized effects of isotopic fractionation due to evaporation and moisture recycling under higher temperatures and humidity. Rainwater samples collected during the mid-rainy season (July–August 2024) confirmed this trend, with δ¹⁸O values ranging from 2.5 to 5.6 ‰ and δ²H values from 12.7 to 38.8 ‰ (Figure 14).

$$\mathrm{LMWL}: \left({\delta }^2\mathrm{H}\right)=7.95\left({\delta }^{18}\mathrm{O}\right)+7.54,$$

(5)

$$\mathrm{GMWL}: \left({\delta }^2\mathrm{H}\right)=8\left({\delta }^{18}\mathrm{O}\right)+10,$$

(6)

Shallow groundwater samples displayed a broad range of isotopic compositions, with δ¹⁸O values averaging –3.8 ± 0.6 ‰ and δ²H values averaging –22.3 ± 4.2 ‰. The strong alignment of groundwater isotopes with the LMWL indicates direct recharge from recent precipitation, with minimal evaporation or isotopic exchange, suggesting that the groundwater is relatively young. Similarly, the Ping River’s isotopic values during the rainy season aligned closely with the LMWL, although slight abnormalities indicated minor evaporation effects.

5. Discussion

5.1. Flow Direction and Measured Hydraulic Head

Hydrogeological heterogeneity, structural controls, and climatic–seasonal drivers govern the groundwater flow direction in the Chiang Mai–Lamphun basin. The topographic gradient directs flow from recharge zones in the northern highlands, underlain by fractured pre-Cambrian to Quaternary consolidated rocks and permeable overlying sediments, to discharge areas in the central–southern lowlands dominated by unconsolidated alluvium (silt, sand, gravel). The Central Alluvial Channel is composed of sand and gravel deposited by the Ping River and its tributaries. These coarse, poorly sorted sediments create highly interconnected pore spaces, enabling high groundwater flow (up to 20 m/d). So, the combination of high hydraulic conductivity and dense well networks make the zone prone to cones of depression, as seen in the Mueang Lamphun district (Lamphun city center).

Seasonally, the dry season has a low hydraulic head due to limited precipitation, amplified evapotranspiration losses of 226.5 Mm³/y (80% of outflow), and over-extraction via pumping, creating cones of depression in districts like Pa Sang. Conversely, low temperatures between 24 and 27 °C and wet-season monsoon precipitation (4–12 mm/month) partially replenish aquifers, so the head is high compared to the dry season.

5.2. Groundwater Modeling and Calibration

The simulated hydraulic head distribution was derived from the calibrated MODFLOW model. It represents a mathematical approximation of a complex natural system. Such predictive models cannot perfectly replicate reality due to a few reasons, such as natural heterogeneity in aquifer properties and uncertain boundary conditions; however, the model reliably replicates the basin flow direction, with high heads found in the northern and northwestern highlands (e.g., Mae Rim and San Sai districts) and gradual discharge into the central–southern lowlands (e.g., Doi Lo and Pa Sang districts). This pattern aligns with the basin’s intermontane structure, where elevated recharge zones feed permeable alluvial aquifers in the lowlands. The strong statistical performance, with an RMSE of 3.0 m and an NRMSE of 3.9%, validates the model’s capability to capture basin-scale trends, as 95% of residuals fall within ±5 m. However, some discrepancies in observation wells, such as CM72 and G147, are possibly due to unmapped geological heterogeneity, like localized clay lenses or fractures near these wells. Overall, these discrepancies do not invalidate the model but rather indicate further opportunities for improvement.

The transition of the Ping River from a gaining stream in 2021 based on a study by Taweelarp et al. [29] to a losing stream in the current study provides compelling evidence of lesser precipitation in subsequent years, as well as aquifer stress in the Chiang Mai–Lamphun basin. This hydrological reversal, confirmed through both field observations and modeling results, demonstrates that groundwater extraction has now surpassed natural recharge rates, a consequence of intensified pumping for agriculture, industry, and domestic use. The declining water tables, exacerbated by increasing evapotranspiration, have dropped below river-bed levels, thus altering river–aquifer interaction. The loss of baseflow to the Ping River threatens aquatic ecosystems, reduces surface water availability for irrigation, and jeopardizes the sustainability of water supplies for growing urban demands. Particularly alarming is how this shift reflects the basin’s transition from a historically healthy, groundwater-sustained system to one that now clearly indicates overexploitation.

5.3. Recharge Zonation

Zones 3 and 4, covering larger areas of 240 km² and 330 km², respectively, play a crucial role in sustaining the basin’s groundwater resources. Zone 3, located in Mueang Chiang Mai district, receives 112 mm/y of recharge and supports a high population density, rapid urbanization, and a thriving tourism industry. Similarly, Zone 4, encompassing Mueang Lamphun district, has the largest recharge area and a recharge rate of 130.2 mm/y, making it an integral part of groundwater replenishment. The high number of groundwater wells in these zones reflects their dependence on groundwater, highlighting the need for targeted management strategies to protect these recharge areas. Prioritizing Zones 3 and 4 for conservation and sustainable development is essential to ensure long-term water security for the basin’s growing population and economy.

The delineated recharge zones encompass a total area of 874 km², with Zones 3 and 4 comprising 570 km² (65% of the overall recharge area). This emphasizes the unequal importance of these zones in preserving the basin’s groundwater resources. Additionally, the groundwater flow direction from Zone 3 to Zone 4 indicates a natural hydrological link between these areas, which means Zone 3 encompasses the Mueang Chiang Mai district, which is very important since it will indirectly affect Zone 4. Zone 3, situated in the Mueang Chiang Mai district, and Zone 4, which includes the Mueang Lamphun district, are the largest recharge zones as well as the most densely populated and rapidly urbanizing areas in the basin. In recent decades, these districts have witnessed significant growth due to population increases, tourism, and economic development, resulting in heightened groundwater demand. The combination of high recharge rates, expansive coverage, and increasing water demand makes Zones 3 and 4 the most critical areas identified in this study. Safeguarding these zones from overexploitation and land-use changes is essential for ensuring the long-term sustainability of groundwater resources in the Chiang Mai–Lamphun basin.

In the model, the basin had a total area of 2490 km², with 874 km² delineated as significant recharge zones. The remaining 1616 km² comprised 65% of the basin and had a low recharge rate due to several possible reasons. Geologically, orthogenesis, migmatites, and S-type granites, particularly the Doi Khun Tan Batholith, exhibit extremely low primary porosity. Additionally, part of this area consists of agricultural land, where these activities compact the soil and allow runoff to pass, thereby reducing percolation. The thicker vadose zone serves as a storage capacity for evapotranspiration, further reducing recharge. Ultimately, the effect of low recharge places stress on high-recharge zones, and therefore the water table begins to decline, resulting in reduced groundwater availability and increased pumping costs. Reduced flow rates in low-recharge areas can lead to the concentration of dissolved solids and potential accumulation of contaminants, affecting groundwater quality.

6. Conclusions

The steady-state groundwater flow model, modified from [29], using MODFLOW, effectively simulated the groundwater dynamics of the Chiang Mai–Lamphun basin. The model, comprising four layers with an 80-row and 71-column grid (1000 × 1000 m² resolution), was calibrated using the PEST pilot point technique, yielding an RMSE of 3.0 m and an NRMSE of 3.9%, indicating high reliability. The water budget highlights the basin’s strong dependence on natural recharge, which accounted for 91.2% of the total inflow (258.9 Mm³/y). The groundwater flow direction was predominantly toward the basin’s center, with higher hydraulic heads in the northeast and lower heads in the south and southwest.

Recharge rates changed significantly across hydrostratigraphic units, with floodplain deposits (Qfd) showing the highest rate (104.0 mm/y), followed by younger terrain deposits (Qyt, 99.8 mm/y) and older terrain deposits (Qot, 103.2 mm/y). The delineation of recharge zones identified seven important zones with recharge exceeding 104 mm/y, with Zones 4 and 6 showing the highest rates of 130.2 mm/y and 134.6 mm/y, respectively. These zones, covering parts of Mueang Lamphun, Ban Thi, Saraphi, Wiang Nong Long, Bai Hong, and Pa Sang districts, are critical for sustainable groundwater management. The isotopic composition of δ¹⁸O and δ²H in rainwater and shallow groundwater further supports these findings, with values aligning closely with the Local Meteoric Water Line (LMWL), which confirms that groundwater recharge primarily results from direct precipitation, with minimal evaporation or isotopic exchange. These findings provide critical insights into groundwater recharge origin, and these seven important recharge zones must be protected by relevant authorities for sustainable groundwater management of the basin.