Subscriber access provided by LUNDS UNIV
Characterization of Natural and Affected Environments
Spatiotemporal Dimensions of Water Stress Accounting: Incorporating Groundwater-Surface Water Interactions and Ecological Thresholds Sara Alian, Alex S. Mayer, Ann Maclean, David Watkins, and Ali Mirchi Environ. Sci. Technol., Just Accepted Manuscript • DOI: 10.1021/acs.est.8b04804 • Publication Date (Web): 12 Feb 2019 Downloaded from http://pubs.acs.org on February 13, 2019
Just Accepted “Just Accepted” manuscripts have been peer-reviewed and accepted for publication. They are posted online prior to technical editing, formatting for publication and author proofing. The American Chemical Society provides “Just Accepted” as a service to the research community to expedite the dissemination of scientific material as soon as possible after acceptance. “Just Accepted” manuscripts appear in full in PDF format accompanied by an HTML abstract. “Just Accepted” manuscripts have been fully peer reviewed, but should not be considered the official version of record. They are citable by the Digital Object Identifier (DOI®). “Just Accepted” is an optional service offered to authors. Therefore, the “Just Accepted” Web site may not include all articles that will be published in the journal. After a manuscript is technically edited and formatted, it will be removed from the “Just Accepted” Web site and published as an ASAP article. Note that technical editing may introduce minor changes to the manuscript text and/or graphics which could affect content, and all legal disclaimers and ethical guidelines that apply to the journal pertain. ACS cannot be held responsible for errors or consequences arising from the use of information contained in these “Just Accepted” manuscripts.
is published by the American Chemical Society. 1155 Sixteenth Street N.W., Washington, DC 20036 Published by American Chemical Society. Copyright © American Chemical Society. However, no copyright claim is made to original U.S. Government works, or works produced by employees of any Commonwealth realm Crown government in the course of their duties.
Page 1 of 31
Environmental Science & Technology
1
Spatiotemporal Dimensions of Water Stress Accounting: Incorporating
2
Groundwater-Surface Water Interactions and Ecological Thresholds
3
Sara Alian1, 3*, Alex Mayer2, Ann Maclean1, David Watkins2, Ali Mirchi3
4
1 School
5
University, 1400 Townsend Drive, Houghton, Michigan 49931.
6
2 Department
7
University, 1400 Townsend Drive, Houghton, Michigan 49931.
8
3 Department
9
111 Agricultural Hall, Stillwater, Oklahoma 74078.
of Forest Resources and Environmental Science, Michigan Technological
of Civil and Environmental Engineering, Michigan Technological
of Biosystems and Agricultural Engineering, Oklahoma State University,
10 11
*Corresponding Author: Sara Alian, Department of Biosystems and Agricultural
12
Engineering, Oklahoma State University, 111 Agricultural Hall, Stillwater, Oklahoma
13
74078. Email:
[email protected] 14
15
Abstract
16
Coarse temporal (i.e., annual) and
17
spatial
18
camouflage water stress associated
19
with withdrawals from surface water
20
and groundwater sources. To address
(i.e.,
TOC Art
watershed)
scales
1
ACS Paragon Plus Environment
Environmental Science & Technology
21
this “curse of scale”, we developed a framework to characterize water stress at different
22
time scales and at fine spatial scales that have not been explored before. Our framework
23
incorporates surface water-groundwater interactions by accounting for spatially cumulative
24
consumptive and non-consumptive use impacts and associated changes in flow due to
25
depletion and return flow along stream networks. We apply the framework using a rich
26
data set of water withdrawals from more than 6,800 principal facilities (i.e., withdrawal
27
capacity > 380,000 liters/day) across the U.S. Great Lakes Basin. Results underscore the
28
importance of spatiotemporal scale and return flows when characterizing water stress.
29
Although the majority of catchments in this water-rich region do not experience large
30
stress, a number of small headwater catchments with sensitive streams are vulnerable to
31
flow depletion caused by surface water and shallow groundwater withdrawals, especially
32
in a high-withdrawal, low-flow month (e.g., August). The return flow from deep
33
groundwater withdrawals compensates for the stream flow depletion to the extent that
34
excess flow is likely in many catchments. The improved ability to pinpoint the imbalance
35
between natural water supply and withdrawals based on stream-specific ecological water
36
stress thresholds facilitates protecting fragile aquatic ecosystems in vulnerable catchments.
37
Keywords: Water stress, spatiotemporal scale, geospatial analysis, water withdrawal,
38
flow depletion, return flow
39
1. Introduction
40
Water stress is a complex concept for which there is no universally accepted definition.1-2
41
It has been conceptualized from the perspectives of human needs and activities,3-7as well 2
ACS Paragon Plus Environment
Page 2 of 31
Page 3 of 31
Environmental Science & Technology
42
as ecological impacts.8-11 Two widely used water stress indices in terms of human water
43
needs are the national-scale water availability per capita per year3 and the ratio between
44
total annual withdrawals and annual water supply. 2,4,12 Seckler et al.13 developed physical
45
and economic scarcity indicators that account for evapotranspiration and return flow on a
46
national scale, as well as the impact of improved water management policies on society’s
47
adaptive capacity to cope with water stress. From an ecological perspective, water stress
48
caused naturally by droughts or artificially by water withdrawals and flow alterations
49
affects aquatic habitats directly through changing flow regimes and connectivity and
50
indirect food chain dynamics.9-11,14,15 The adverse ecological impacts have motivated
51
minimum environmental flows16 and more holistic guidelines and indicators to maintain
52
streamflow variability (e.g., refs 8 and 17).
53
The spatiotemporal dimensions of water stress and accounting for interactions
54
between components of the hydrologic cycle, e.g. surface and groundwater, are important
55
for water and environmental management at local to regional scales to safeguard vulnerable
56
aquatic habitats. Global assessments of annual water stress have been carried out at spatial
57
scales ranging from country-level to major watersheds and 1ᵒ×1ᵒ and 0.5ᵒ×0.5ᵒ grid cells.18-
58
24
59
distribution of water scarcity, particularly in arid and semi-arid regions. However, the
60
“curse of scale” in the high-level water stress analyses hinders detecting localized
61
imbalances between natural water supply and human activities (withdrawals and
62
diversions) that could threaten fragile aquatic ecosystems. Seasonal and monthly water
63
stress analyses on 0.5ᵒ×0.5ᵒ global grid cells,25 user-delineated subbasins,26 and 8-digit and
These water stress analyses have provided a high-level understanding of the global
3
ACS Paragon Plus Environment
Environmental Science & Technology
64
12-digit Hydrologic Unit Code watersheds 27-30 confirm that water shortages occur only as
65
occasional regional or local deficits, at certain times of the year, as previously speculated
66
by Meigh et al.31 Furthermore, differentiation of water supply sources and their dynamic
67
interactions has been recognized as an important research gap in water stress analysis
68
frameworks.26 By considering surface and groundwater separately, previous efforts in
69
water stress accounting have often ignored stream depletions caused by pumping from
70
shallow aquifers, as well as the return flow to streams from non-consumptive groundwater
71
use (e.g., ref 29).
72 73
Using a rich data set of water withdrawals from more than 6,800 principal facilities (i.e.,
74
facilities with a withdrawal capacity >380,000 liters/day) across the U.S. Great Lakes Basin
75
(U.S. GLB; Figure 1), we characterize water stress at fine spatial scales which have not
76
been explored before. The five Laurentian Great Lakes (GL) cover an area of
77
approximately 243,460 km2 (slightly larger than the United Kingdom) and hold about 18%
78
of the global freshwater,32 creating a water-abundant region that has little to no water stress
79
based on the Falkenmark water scarcity indicator.33 The basin is networked with thousands
80
of streams ordered from 1 to 7 based on the Strahler stream order classification,34 with
81
average annual flows ranging from 0.01 cms to more than 5,000 cms. The GLB supports a
82
$5 trillion regional economy and is home to about 34 million people, with more than 80%
83
living in the U.S. portion of the basin.35 Mapping water stress in the U.S. GLB is especially
84
relevant from an ecological standpoint28,36 to develop sound water management policies
85
that consider the connection between water use, withdrawals, and biodiversity.37 4
ACS Paragon Plus Environment
Page 4 of 31
Page 5 of 31
Environmental Science & Technology
86
This paper demonstrates the implications of spatiotemporal scales and surface water-
87
groundwater interactions in pinpointing vulnerable catchments with sensitive aquatic
88
habitats in order to mitigate adverse impacts of excessive localized water withdrawal. We
89
describe the water stress characterization framework and present annual and warm season
90
results from across the U.S. GLB to illustrate the effects of temporal resolution and
91
catchment size on water stress. We emphasize the critical importance of accounting for
92
return flows to streams, especially from pumped groundwater, when characterizing
93
localized water stress. Further, we highlight ecological water stress in Michigan, a state
94
where ecological-based guidelines have been adopted for water withdrawal permitting. Our
95
framework can potentially improve the permitting process and guide monitoring
96
campaigns with the aim of mitigating ecological water stress.
97 98
Figure 1. The U.S. portion of the Great Lakes drainage basin. 5
ACS Paragon Plus Environment
Environmental Science & Technology
99
2. Methods
100
We apply an integrative water accounting framework to quantify catchment-scale water
101
stress at annual and monthly time intervals. The approach resolves several limitations in
102
geospatially explicit water stress mapping. First, surface-groundwater interactions are
103
incorporated in the analysis by estimating stream depletion from shallow groundwater
104
withdrawals and accounting for return flows from non-consumptive water use. Second,
105
cumulative impacts of water withdrawals from different sources and return flow are
106
accounted for by aggregating consumptive and non-consumptive use impacts from
107
upstream to downstream along stream networks. Third, spatially variable water stress
108
thresholds are considered based on ecological characteristics of streams. The framework is
109
based on comparing surface water stress index (SWSI), defined here as the ratio of flow
110
disturbance to adjusted streamflow--similar to the Modified Water Exploitation Index--12
111
with specified stress thresholds. It uses a comprehensive, logically structured geodatabase
112
of major water withdrawals for different use categories in the U.S. GL states, including
113
domestic and public, commercial and industrial, agricultural and golf course irrigation,
114
livestock, mining, and thermoelectric power generation. Figure 2 illustrates the analysis
115
framework.
116 117 118 119 120 6
ACS Paragon Plus Environment
Page 6 of 31
Page 7 of 31
Environmental Science & Technology
121 122 123 124 125 126 127 128 129 130 131 132 133
Figure 2. Water stress analysis framework.
134
Required data include georeferenced surface water and groundwater withdrawals (Table
135
S1 in Supporting Information lists water withdrawal data sources) based on water use
136
categories (Figure S1 in Supporting Information), consumptive use coefficients by water
137
use category (Table S2), and hydrologic data inputs (e.g., stream network, catchment areas,
138
and flows). Geospatial data, including monthly water withdrawal magnitudes, water use
139
categories and catchment boundaries and associated flows, were compiled in an ArcMap
140
file geodatabase (GDB), providing a systematic data repository. The water withdrawal data
141
set is composed of 6,805 georeferenced water withdrawals for 2010 from facilities with
142
capacities greater than or equal to 380,000 liters/day, scattered over 106,000 catchments.38
143
The consumptive use coefficients are USGS-published values based on synthesis of 7
ACS Paragon Plus Environment
Environmental Science & Technology
144
consumptive water use data in Indiana and Ohio, with additional data from Wisconsin.39
145
We converted annual water withdrawal data into monthly values using the arithmetic
146
average of the median monthly fraction (%) of water withdrawals for different use
147
categories in Indiana and Ohio (Table S3 in Supporting Information).39
148 149 150
The catchments defined in the National Hydrography Dataset Plus Version 2 (NHDPlus
151
V2) are generally very small (less than 2 km2), although a few hundred catchments are 20
152
km2 or larger (Figure S2). The NHDPlus V2 database provides hydrographic, hydrologic,
153
and spatial attributes at the catchment scale and at different temporal scales for the
154
conterminous U.S.40 The hydrologic data include mean annual and mean monthly
155
unimpaired flows derived from the Enhanced Runoff Method.41 EROM uses temperature
156
and precipitation data, simulated runoff, and gauge flow measurements within a grid-based
157
(900m×900m) water balance framework to estimate the hydrologic response at ungauged
158
sites for the period 1971 to 2000.
159 160
We only consider withdrawals within the tributary watersheds of the Great Lakes, rather
161
than the Great Lakes themselves, since SWSIs are unlikely to be substantial at the scale of
162
the Great Lakes volume (23,000 km3).42 Illinois is excluded from the analysis as it occupies
163
a small portion of the U.S. GLB, with Lake Michigan being the dominant water source.
164
Water withdrawal sources include tributary surface water, shallow groundwater, and deep
165
groundwater. Direct surface water withdrawals and groundwater extractions from shallow 8
ACS Paragon Plus Environment
Page 8 of 31
Page 9 of 31
Environmental Science & Technology
166
aquifers that are hydraulically connected to the streams both potentially deplete
167
streamflow. Thus, the magnitude of streamflow depletion (𝑄𝑑) due to withdrawal depends
168
on the withdrawal source:
169
𝑄𝑑𝑖,𝑡,𝑠 = 𝑆𝐷𝐹𝑖,𝑡,𝑠𝑊𝑖,𝑡,𝑠
170
where 𝑄𝑑 = stream depletion (volume/time); 𝑆𝐷𝐹 = streamflow depletion fraction
171
(unitless); W = withdrawal (volume/time); i = location index (catchment); t = time index
172
(annual or monthly); and s = source index.
(Eq. 1)
173 174
A streamflow depletion fraction 𝑆𝐷𝐹𝑖,𝑡,𝑠 = 1 is assigned to surface water withdrawals prior
175
to accounting for return flows because these withdrawals are directly extracted from the
176
streams. Deep wells (≥30 m) or wells open to consolidated materials are assumed not in
177
hydraulic connection with streams and are assigned 𝑆𝐷𝐹𝑖,𝑡,𝑠 = 0, meaning the withdrawals
178
from these sources have only a recharge effect on the streamflow. Shallow wells (≤30 m)
179
in coarse-grained drift material (e.g., sand and gravel) are considered hydraulically
180
connected to the nearby stream, as is typically the case in the U.S. GLB.43 We obtained
181
quantitative and qualitative information about well depth and aquifer material by linking
182
the georeferenced water withdrawal data set to the Wellogic GIS layers.44 Wellogic is a
183
database of water well records that provides information about well depth and
184
hydrogeologic characteristics of the aquifers.44 Where georeferenced coordinates did not
185
match between the Wellogic system and the withdrawal locations, the properties of the
186
nearest well in the Wellogic system within a 1000-m radius of the withdrawal point were
187
utilized. Wells outside the coverage of the Wellogic database were considered connected 9
ACS Paragon Plus Environment
Environmental Science & Technology
Page 10 of 31
188
to streams if they were located in surficial glacial and alluvial deposits based on a map of
189
surficial materials for the eastern and central U.S.45 Values of SDFi,t,s for shallow
190
groundwater range from 0 to 1 and are estimated using the Hunt analytical solution46 as a
191
function of withdrawal and hydrogeological properties (Eq. 1), as described in Watson et
192
al.: 47 𝑆𝐷𝐹𝑖,𝑡,𝑠𝑔𝑤 =
193
Qdi,t,sgw QGWi,t,sgw
[ ( ) (
= erfc
λ2tpw λd Sd2 ― exp + erfc 4ST 2T 4Ttpw
) (
Eq.2)
λ2tpw
Sd2 + 4Ttpw 4ST
)]
( 𝑖,𝑡,𝑠𝑔𝑤
194
where QGW = pumping rate for shallow groundwater (sgw) (volume/time); tpw = average
195
length of pumping in a given year (time); i = location index (catchment); and t = time index
196
(annual or monthly). The hydrogeologic properties in each catchment include S = storage
197
coefficient (dimensionless) of the shallow aquifer underlying catchment i; d = distance
198
from each pumping well to the stream segment in catchment i (length); T = transmissivity
199
of the shallow aquifer underlying catchment i (length/time); and λ = streambed
200
conductance for the stream segment in catchment i. The spatial dependence of equation (2)
201
corresponds to the variable d, which is the distance from the point locations of each
202
pumping well location to the nearest stream segments, estimated using Arc GIS tools. For
203
the temporal dependence in equation (2), we assume that the depletion, Qdi,t,sgw, reaches
204
steady state instantaneously as the pumping rate changes each time step, t, avoiding the
205
necessity to assign arbitrary values for the variable tpw. Following the approach of Watson
206
et al.47 we use tpw = 100 days to assure that equation (2) has reached steady state for all
207
pumping locations. 10
ACS Paragon Plus Environment
Page 11 of 31
Environmental Science & Technology
208 209
Aggregated streamflow disturbance (ASD) is determined as the collective effect of return
210
flow and flow depletion (i.e., sum of gains and losses) at the catchment scale, which can
211
be positive or negative. Return flows are the converse of the consumptive proportion of
212
water withdrawal, i.e., the amount of water removed locally due to evapotranspiration,
213
incorporation in exported products, and consumption by humans and livestock.48
214
ArcHydro49 is used to propagate depletions and return flows from upstream to downstream
215
by accumulating the ASDs from one catchment to the next along the stream network.
216
Streamflows are adjusted (𝑄𝑎𝑖,𝑡) in each catchment by adding the cumulative aggregated
217
streamflow disturbance (𝐶𝐴𝑆𝐷𝑖,𝑡) to the NHDPlus V2 unimpaired streamflow (Eqs. 3 and
218
4), i.e., flow (𝑄𝑖,𝑡) that would occur at a given catchment if unaffected by human activities
219
(NHD, 2016). The SWSI is calculated as the ratio of the cumulative aggregated streamflow
220
disturbance and adjusted streamflow (Eq. 5).
221
𝑄𝑎𝑖,𝑡 = 𝐶𝐴𝑆𝐷𝑖,𝑡 + 𝑄𝑖,𝑡 = [(∑𝑖∑𝑡((1 ― CUi,t, wu )𝑊𝑖,𝑡,𝑤𝑢 ― Qdi,t)] + 𝑄𝑖,𝑡
222
𝑘 𝐶𝐴𝑆𝐷𝑖,𝑡 = ∑𝑖 = 𝐴𝑆𝐷𝑖,𝑡 1
(Eq. 4)
223
𝑆𝑊𝑆𝐼𝑖,𝑡 = (CASDi,t/Qai,t) × 100
(Eq. 5)
224
where 𝑄𝑎 = adjusted (index) streamflow (volume/time); 𝐶𝐴𝑆𝐷 = cumulative aggregated
225
streamflow disturbance (volume/time); 𝑄 = unimpaired streamflow (from NHD Plus V2);
226
𝑆𝑊𝑆𝐼 = surface water stress index (%); CU = consumptive use coefficient (fraction);39 i =
227
catchment index; nk = number of stream segments in a given aggregated set of catchments,
228
k; t = time index (annual or monthly); wu = water use category index; and 𝑄𝑑 = streamflow
229
depletion (volume/time).
𝑛
11
ACS Paragon Plus Environment
(Eq. 3)
Environmental Science & Technology
230
Note that, on the one hand, in equation 5, if streamflow depletions exceed return flows in
231
a given catchment, the SWSI has a negative value, implying depleted flow; and, on the
232
other hand, the SWSI has a positive value if the return flow exceeds the stream depletion,
233
implying excess flow, as in the water use regime analyses of Weiskel et al.50 Accounting
234
for net increased streamflow due to return flows is important in terms of impacts on the
235
streams’ thermal regime, and water quality, especially in the context of non-point source
236
nutrient loading. It is assumed that return flows occur within the same catchment as the
237
withdrawal. August is designated as the warm low-flow month, since 93% of the lowest
238
monthly stress (stream depletion) occurs in this month, due to a combination of lower
239
streamflows and higher water withdrawals and consumptive use coefficients.
240 241
In the U.S. GLB, Michigan is currently the only state that has implemented geographically-
242
specific ecological flow thresholds into policy. The Michigan thresholds are based on field
243
studies of relationships between target fish populations and streamflow and are classified
244
by stream thermal regime and size.51 The Michigan policy also classifies depletion
245
thresholds as low, intermediate or high risk, according to the level of stream depletion
246
(Table 1); and the policy uses average monthly flows in the low-flow month (e.g., August)
247
flows, rather than average annual flows, to assess stream depletions. We use the Michigan
248
flow criteria, which provide specific stream depletion thresholds for each stream segment
249
in the network, to assess the potential ecological impacts of stream depletions across the
250
state. These flow criteria are temporally and spatially commensurate with our SWSI
251
analyses. 12
ACS Paragon Plus Environment
Page 12 of 31
Page 13 of 31
Environmental Science & Technology
252
3. Results and Discussion
253
Accounting for return flows from groundwater pumping substantially reduces and even
254
reverses estimated stream depletions at annual time scales. Water in the U.S. GLB is
255
mostly withdrawn from surface waters, with the exception of Pennsylvania, which relies
256
predominantly on shallow groundwater (Figure S3 in Supporting Information). The overall
257
fractions of withdrawals from tributary surface water, shallow groundwater, and deep
258
groundwater are 81%, 10%, and 9% respectively. Thus, in the majority of catchments,
259
streamflow depletion occurs due to direct surface water extractions (93%), rather than
260
stream depletions associated with shallow groundwater pumping (7%). Figure 3a shows
261
the distribution of annual SWSI for the case where groundwater return flows are ignored.
262
The majority of stressed catchments (86%) have SWSIs between 0% and -5%. Of the
263
catchments that have lower annual SWSI, 60% occur within cultivated lands, 16% in
264
developed areas, and 24% in non-developed non-agricultural land use types. In these
265
catchments, approximately 40% of total water withdrawal is supplied from shallow
266
groundwater sources, while surface water provides 60%. Clusters of low SWSIs are
267
detected in areas of more intensive irrigated agriculture (e.g. central Wisconsin, southern
268
Michigan, northeastern Indiana, northwestern Ohio, and northern New York) (Figure 3a).
269
Lower SWSIs from agricultural withdrawals are also due to high consumptive use
270
coeffients for this use category (~80%, see Table S2). For the SWSIs less than -10%, almost
271
all (98%) of the withdrawals are for irrigation, while a small number of withdrawals are
272
attributed to the industrial and mining sectors (1% each).
13
ACS Paragon Plus Environment
Environmental Science & Technology
273
The neglect of groundwater return flow generally leads to an overestimation of water stress
274
levels (Figure 4). When groundwater return flow is accounted for (Figure 3b and Figure
275
4), the results are substantially different. Thirteen percent of the stressed catchments now
276
have SWSI values greater than 5%, and 2% have SWSIs less than -5%. The clusters of
277
higher SWSIs are associated with larger cities that use deep groundwater for water
278
supplies, which are scattered around the study area but tend to be at least 10s of km from
279
the Great Lakes shoreline. For annual SWSIs greater than 10%, more than 50% of
280
withdrawals belong to domestic and public use categories. The remaining withdrawals are
281
split between industrial (30%), mining (10%) and agricultural (9%) purposes. Higher SWSI
282
for public water supplies is explained by the relatively low consumptive use coeffients for
283
this use category (~10%, see Table S2).
284 285
Warm season withdrawals and streamflows produce much greater stresses that are more
286
sensitive to groundwater return flows. For the case where groundwater return flows are
287
ignored, Figure 3c shows that a substantially larger number of catchments have low SWSIs
288
in the warm month (August), when water withdrawals and consumptive use coefficients
289
are larger and streamflows are smaller, compared to annual averages (Figure 3a). In this
290
case, SWSIs less than -10% are detected in 23% of the stressed catchments (Figure 4), and
291
16% of the warm month SWSIs are less than -20%. This is a very different picture than
292
the annual scale analysis which detects water stress levels less than -10% in only 8% of the
293
catchments that experience water scarcity (Figure 4). Approximately 60% of the total water
294
withdrawal in catchments with water stress levels less than -10% in August is from shallow 14
ACS Paragon Plus Environment
Page 14 of 31
Page 15 of 31
Environmental Science & Technology
295
groundwater. When accounting for groundwater return flow, the proportion of stressed
296
catchments with SWSIs greater than +5% reaches 13% in the warm month (Figure 4). In
297
this case, 10% of the catchments have SWSIs less than -5%. Further, 8% and 5% of the
298
catchments have SWSIs below -10% and -20%, respectively, in the warm month, which
299
are significantly greater than the corresponding SWSI distributions at the annual time scale.
300
The results indicate that offsets of flow depletions from adding groundwater return flow
301
have an almost negligible effect in many catchments during the warm month. Highly
302
stressed catchments due to streamflow depletion (i.e., SWSI less than -10%) are generally
303
more sensitive to warm season withdrawals.
15
ACS Paragon Plus Environment
Environmental Science & Technology
304 305 306 307 308 309 310 311 312 313 314 315
Figure 3. Spatial distribution of annual and warm month (August) water stress in the U.S. portion of the Great Lakes Basin: (a) annual
316
SWSI without GW return flow, (b) annual SWSI with GW return flow, (c) warm month SWSI without GW return flow, and (d) warm
317
month SWSI with GW return flow. Groundwater return flows compensate for surface water depletions resulting in positive SWSI
318
values (excess flow) in many catchments. Flow depletion is generally more pronounced and widespread in the high-withdrawal, low-
319
flow month of August, creating clusters of catchments with low SWSI even after considering the effect of groundwater return flows.
320 16
ACS Paragon Plus Environment
Page 16 of 31
Page 17 of 31
Environmental Science & Technology
321 322
Figure 4. Summary of stressed catchments for the four cases: annual and warm month water
323
stress with and without groundwater return flows (GW). Ignoring groundwater return flow
324
causes an overestimation of streamflow depletion, especially in the warm month.
325 326
Largest stresses occur primarily in catchments with sizes less than 25 km2, with stream
327
orders of 2 or less. Catchment-level geospatial analysis provides an opportunity to
328
systematically investigate SWSI at much finer spatial scales than previous watershed-scale
329
analyses (e.g., refs 28 and 29), as shown in Figure 5a and b. In line with the outcomes of
330
similar attempts to quantify water stress at coarser scales (e.g., refs 26-29) our findings
331
demonstrate that the ability to detect vulnerable catchments is sensitive to the
332
spatiotemporal scale of analysis. As expected, larger (absolute) SWSIs are estimated for
333
smaller catchments (e.g., smaller than 10 km2), and water stress index values generally
334
recede towards zero as catchments become larger. This effect is observed in both annual
335
and warm month water stress analyses (Figure 5a and b). A significant number of larger
336
catchments (e.g., 10-20 km2) that were not identified as highly stressed (in excess of +/-
337
10%) in the annual scale analysis do exceed the 10% threshold in the warm month. In 17
ACS Paragon Plus Environment
Environmental Science & Technology
338
general, catchment sizes greater than 25 km2 rarely exhibit stresses less than - 10% or
339
greater than +10%.
340 341
Clusters of catchments with negative SWSI values in the warm month tend to be found in
342
headwater catchments (i.e., Strahler stream order 1) in southwestern and southeastern
343
Michigan, northwestern and northeastern Ohio, and western New York (Figure S4 in
344
Supporting Information). Seventy eight percent of catchments with streamflow depletion
345
in the warm month are found to have stream order type 1. Stream order 2 is the second
346
most vulnerable, where 16% of the negative warm month SWSIs occur, while only 7% of
347
catchments with stream orders 3 or higher are found to have negative SWSIs in August.
348
Furthermore, small versus large catchments results imply that water stress assessments of
349
small catchments are particularly sensitive to the neglect of groundwater-surface water
350
interactions and higher temporal resolution.
351
18
ACS Paragon Plus Environment
Page 18 of 31
Page 19 of 31
Environmental Science & Technology
352 353
Figure 5. Distribution of water stress relative to catchment size: (a) annual and (b) warm
354
month. Small catchments are more vulnerable to localized water stress, especially in the
355
warm month. Excluding groundwater return flows will lead to overestimating stream flow
356
depletion (i.e., negative water stress) and failing to pinpoint catchments that may
357
experience excess flow (i.e., positive stress) due to large groundwater return flows,
358
especially in small catchments.
359 360
Incorporation of ecological thresholds facilitates detecting vulnerable stream segments.
361
The results of the spatially-explicit ecological surface water stress characterization in
362
Michigan are shown in Table 1. The ecological thresholds were applied in a post-process
363
SWSI analysis phase to evaluate the stressed stream segments in Michigan in terms of
364
ecological impact. The results reveal that while most catchments are not currently
365
threatened, approximately 650 km of streams fall under the low, intermediate, or high-risk
366
category, when accounting for groundwater return flow. More than 20% of stream
367
segments with negative SWSI values in Michigan are prone to intermediate to high
368
ecological risk due to flow depletion. In the high-risk category, most of the streams 19
ACS Paragon Plus Environment
Environmental Science & Technology
Page 20 of 31
369
exceeding depletion thresholds are small warm or small cold-transitional, generally due to
370
high withdrawals in small warm streams and stringent thresholds in the case of the cold-
371
transitional streams. Excluding groundwater return flows results in significant
372
overestimation of vulnerable streams (i.e., 2918 km).
373 374 375
Table 1- Ecological vulnerability of Michigan streams to streamflow depletion in the warm month.
Stream type
Stress range (%)
Cold stream
Low risk* Stream length in range (km)
Intermediate risk* Stream length in range (km) Stress range With GW Without (%) return GW return flow flow
With GW return flow
Without GW return flow
24
25
77
17
1
1
22
--
22
--
511
2242
--
76
506
--
57
170
376
*Risk
377 378 379
**Streamflow
380
The framework developed herein facilitates identification of locales where site-specific
381
evaluations should be performed to ensure existing or new water withdrawals will not put
382
aquatic habitats at risk (Figure S5). Identifying the areas of concern is critical for
intervals are based on absolute streamflow depletion thresholds in
MDNR.51
depletion range threshold not available
20
ACS Paragon Plus Environment
Page 21 of 31
Environmental Science & Technology
383
developing campaigns to monitor adverse ecological impacts related to excessive water
384
withdrawals, and for devising strategies to maintain ecological integrity. Ecological
385
guidelines should be developed for streams in all GL states to facilitate the spatially-
386
explicit analysis of the potential adverse impacts of anthropocentric water use on aquatic
387
habitats. Furthermore, ecological impacts of return flows in terms of discharge location,
388
magnitude, thermal regime, and water quality should be better understood.
389 390
Sources of uncertainty in the input data and modeling framework. Interpreting the results
391
to derive management insights requires recognition of sources of uncertainty and caveats
392
of the presented SWSI analysis framework. Our water-accounting framework uses
393
withdrawal data collected from seven Great Lakes states, all of which use different
394
protocols for collecting data on water withdrawal magnitude and location for individual
395
users. As such, verifying the magnitude and location of the water withdrawal data over
396
such a large geographic extent is a challenge. On an aggregate level, water withdrawals
397
across sectors from our database compare well with aggregated withdrawals reported by
398
the Great Lakes Commission.52 Overall, the Great Lakes Commission’s estimate of total
399
withdrawals in the study area is about 10% larger than our independent estimate, which,
400
although significant in terms of volume (i.e., 60 million m3/day), indicates reasonable
401
agreement between the two databases. The difference is most likely due to our refinement
402
of the raw data to assemble a consistent database, excluding some water withdrawal data
403
due to concerns about accurate location and reported withdrawal source. For example,
404
withdrawal points whose coordinates fell on inland lakes were excluded in order to 21
ACS Paragon Plus Environment
Environmental Science & Technology
405
consider only surface water withdrawals from streams, and wells that are very close (i.e.,
406
less than 1 km) to the Great Lakes were assumed to have minimal impact on inland
407
streamflow disturbance due to hydraulic connection to the lakes. Accurate spatial locations
408
for the withdrawal points are also important to estimate the fraction of streamflow depletion
409
in areas where shallow aquifers hydraulically connected to nearby streams are the main
410
water source.47
411 412
The presented surface water stress characterization framework and reported findings about
413
spatiotemporal dimensions and limitations of water stress accounting are generally
414
applicable for catchment level waters stress analysis in other settings. This study highlights
415
a need for an extensive, unified water withdrawal, transfer, and discharge reporting
416
protocol. The process of cross-walking the water withdrawal data from different
417
management jurisdictions (e.g., U.S. Great Lakes States) to a common classification
418
scheme is time-consuming and potentially involves subjective interpretation of water use
419
classes. While water withdrawal locations with capacities less than 380,000 liters/day could
420
affect SWSI calculations in small catchments, obtaining the required data to extend the
421
SWSI calculations to these locations is difficult. Further, we assume the return flows
422
remain in the same catchment as the withdrawals, although in reality there may be cases in
423
which water utilities transfer water to users across catchment boundaries. Cross-catchment
424
water transfers can also occur due to irrigation of large agricultural areas. Finally, a better
425
understanding of the spatial variation of consumptive water use for different use categories
426
and seasons across the GL states will improve SWSI calculations. Monthly consumptive
427
use values for Indiana and Ohio were used for all the GL states based on climatic 22
ACS Paragon Plus Environment
Page 22 of 31
Page 23 of 31
Environmental Science & Technology
428
similarities.39,53 Better estimates of consumptive use coefficients are especially critical for
429
headwater catchments with large agricultural and livestock farm operations, which are
430
found to be most sensitive to streamflow disturbances.
431 432
Acknowledgements
433
We acknowledge funding from the U.S. National Science Foundation (NSF award No.
434
CBET 0725636) and the Great Lakes Protection Fund (award No. 946). We also appreciate
435
technical and data support from Katelyn Watson, Meredith Ballard LaBeau, and Rabi
436
Gyawali, as well as helpful comments from three anonymous reviewers. Any opinions,
437
findings, and conclusions or recommendations expressed in this material are those of the
438
authors.
439 440
Supporting Information. Input data, additional results, and maps of water stress indices.
441 442 443
References (1) Molle, F.; Mollinga, P. Water poverty indicators: conceptual problems and policy
444
issues. Water Policy. 2003, 5.5-6, 529-544, DOI
445
https://doi.org/10.2166/wp.2003.0034.
446
(2) Pedro-Monzonís, M.; Solera, A.; Ferrer, J.; Estrela, T.; Paredes-Arquiola, J. A review
447
of water scarcity and drought indexes in water resources planning and management. J.
448
Hydrol. 2015, 527, 482–493; DOI http://dx.doi.org/10.1016/j. jhydrol.2015.05.003.
23
ACS Paragon Plus Environment
Environmental Science & Technology
449 450
(3) Falkenmark, M. The massive water scarcity now threatening Africa: why isn't it being addressed? Ambio. 1989, 112-118.
451
(4) Raskin, P.; Gleick, P.; Kirshen, P.; Pontius, G.; Strzepek, K. Water futures: Assessment
452
of long-range patterns and problems. Comprehensive assessment of the freshwater
453
resources of the world, SEI. Stockholm, 1997; https://mediamanager.sei.org/
454
documents/Publications/SEI-Report-WaterFutures-AssessmentOfLongRange
455
PatternsAndProblems-1997.pdf.
456
(5) Ohlsson, L. Water conflicts and social resource scarcity. Phys.Chem. Earth, Part B:
457
Hydrology, Oceans and Atmosphere. 2000, 25, 213-220; DOI https://doi.org/
458
10.1016/S1464-1909(00)00006-X.
459
(6) Sullivan, C. A.; Meigh, J. R.; Giacomello, A. M. The water poverty index:
460
development and application at the community scale. Nat. Resour. Forum. 2003, 189-
461
199; DOI https://doi.org/10.1111/1477-8947.00054.
462 463
(7) Rijsberman, F. R. Water scarcity: fact or fiction? Agric. Water Manag. 2006, 80, 5-22; DOI https://doi.org/10.1016/j.agwat.2005.07.001.
464
(8) Richter, B. D.; Davis, M. M.; Apse, C.; Konrad, C. A presumptive standard for
465
environmental flow protection. River Res. Appl. 2012, 28(8), 1312–1321; DOI
466
https://doi.org/10.1002/rra.1511.
467
(9) Lake, P.S. Ecological effects of perturbation by drought in flowing waters. Freshwater
468
Biol.. 2003, 48(7), pp.1161-1172. DOI https://doi.org/10.1046/j.1365-
469
2427.2003.01086.x.
24
ACS Paragon Plus Environment
Page 24 of 31
Page 25 of 31
Environmental Science & Technology
470
(10) Dewson, Z.S.; James, A.B.; Death, R.G. A review of the consequences of decreased
471
flow for instream habitat and macroinvertebrates. J. N. Am. Benthol. Soc. 2007, 26(3),
472
pp.401-415.
473
(11) Rolls, R.J.; Leigh, C.; Sheldon, F. Mechanistic effects of low-flow hydrology on
474
riverine ecosystems: ecological principles and consequences of alteration. Freshw. Sci.
475
2012, 31(4), pp.1163-1186; DOI https://doi.org/10.1899/12-002.1.
476
(12) Results and lessons from implementing the Water Assets Accounts in the EEA area.
477
From concept to production; EEA Technical report No 7/2013; European
478
Environment Agency (EEA): Copenhagen, Denmark, 2013;
479
https://www.eea.europa.eu/publications/water-assets-accounts-report
480
(13) World water demand and supply, 1990 to 2025: Scenarios and issues; IWMI
481
Research Report 19, International Water Management Institute (IWMI): Colombo, Sri
482
Lanka, 1998; DOI 10.3910/2009.019.
483
(14) Mirchi, A.; Watkins, D.W.; Huckins, C.J.; Madani, K.; Hjorth, P. Water resources
484
management in a homogenizing world: Averting the Growth and Underinvestment
485
trajectory. Water Resour. Res. 2014, 50(9), pp.7515-7526. DOI
486
https://doi.org/10.1002/2013WR015128.
487
(15) Rosenfeld, J.S. Developing flow–ecology relationships: implications of nonlinear
488
biological responses for water management. Freshwater Biol. 2017, 62(8), pp.1305-
489
1324; DOI https://doi.org/10.1111/fwb.12948.
490 491
(16) Tennant, D.L. Instream flow regimens for fish, wildlife, recreation and related environmental resources. Fisheries. 1976, 1(4), pp.6-10. 25
ACS Paragon Plus Environment
Environmental Science & Technology
492
(17) Poff, N. L.; Allan, J. D.; Bain, M. B.; Karr, J. R.; Prestegaard, K. L.; Richter, B. D.;
493
Sparks, R. E.; Stromberg, J. C. The natural flow regime. BioScience. 1997, 47, 769-
494
784; DOI 10.2307/1313099.
495 496
(18) Arnell, N. W. Climate change and global water resources. Global. Environ. Chang. 1999, 9, S31–S49, DOI10.1016/S0959-3780(99)00017-5.
497
(19) Vörösmarty, C. J.; Green, P.; Salisbury, J.; Lammers, R. B. Global water resources:
498
Vulnerability from climate change and population growth. Science. 2000, 289, 284–
499
288; DOI 10.1126/science.289.5477.284.
500
(20) Alcamo, J.; T. Henrichs; T. Rösch. World water in 2025—Global modeling and
501
scenario analysis for the World Commission on Water for the 21st Century. Kassel
502
World Water Ser. Rep. 2, 2000, Cent. for Environ. Syst. Res., Univ. of Kassel, Kassel,
503
Germany.
504
(21) Oki, T.; Agata, Y.; Kanae, S.; Saruhashi, T.; Yang, D.; Musiake K. Global
505
assessment of current water resources using total runoff‐integrating pathways.
506
Hydrolog. Sci. J. 2001, 46, 983–995; DOI 10.1080/02626660109492890.
507
(22) Arnell, N. W. Climate change and global water resources: SRES emissions and
508
socio‐economic scenarios. Global. Environ. Chang. 2004, 14, 31–52, DOI
509
10.1016/j.gloenvcha.2003.10.006.
510
(23) Islam, S.; Oki, T.; Kanae, S.; Hanasaki, N.; Agata, Y.; Yoshimura, K. A grid‐based
511
assessment of global water scarcity including virtual water trading. Water Resour.
512
Manag. 2007, 21, 19–33, DOI 10.1007/s11269-006-9038-y.
26
ACS Paragon Plus Environment
Page 26 of 31
Page 27 of 31
Environmental Science & Technology
513
(24) Hanasaki, N.; Kanae, S.; Oki, T.; Masuda, K.; Motoya, K.; Shirakawa, N.; Shen, Y.;
514
Tanaka, K. An integrated model for the assessment of global water resources—Part 2:
515
Applications and assessments. Hydrol. Earth Syst. Sci. 2008, 12, 1027–1037, DOI
516
10.5194/hess-12-1027-2008.
517
(25) Wada, Y.; van Beek, L. P. H.; Viviroli, D.; Dürr, H. H.; Weingartner, R.; Bierkens,
518
M. F. P. Global monthly water stress: 2. Water demand and severity of water stress,
519
Water Resour Res. 2011, 47, W07518; DOI 10.1029/2010WR009792.
520
(26) Scherer, L.; Venkatesh, A.; Karuppiah, R.; Pfister, S. Large-scale hydrological
521
modeling for calculating water stress indices: implications of improved spatiotemporal
522
resolution, surface-groundwater differentiation, and uncertainty characterization.
523
Environ. Ssc. Technol. 2015, 49(8), pp.4971-4979; DOI 10.1021/acs.est.5b00429.
524
(27) Averyt, K.; Meldrum, J.; Caldwell, P.; Sun, G.; McNulty, S.; Huber-Lee, A.;
525
Madden, N. Sectoral contributions to surface water stress in the coterminous United
526
States. Environ. Res. Lett. 2013, 8(3), p.035046, DOI https://doi.org/10.1088/1748-
527
9326/8/3/035046.
528
(28) Mubako, S. T.; Ruddell, B. L.; Mayer, A. S. Relationship between water withdrawals
529
and freshwater ecosystem water scarcity quantified at multiple scales for a great lakes
530
watershed. J. Water Res Plan Man. 2013, 139, 671-681; DOI https://doi.org/
531
10.1061/(ASCE)WR.1943-5452.0000374.
532
(29) Eldardiry, H.; Habib, E.; Borrok, D. M. Small-scale catchment analysis of water
533
stress in wet regions of the US: an example from Louisiana. Environ. Res. Lett. 2016,
534
11 (12), 124031, DOI https://doi.org/10.1088/1748-9326/aa51dc. 27
ACS Paragon Plus Environment
Environmental Science & Technology
535
(30) Wang, R.; Zimmerman, J.B.; Wang, C.; Font Vivanco, D.; Hertwich, E.G. Freshwater
536
vulnerability beyond local water stress: Heterogeneous effects of water-electricity
537
nexus across the continental United States. Environ. Sci. Technol. 2017, 51(17),
538
pp.9899-9910; DOI: 10.1021/acs.est.7b01942.
539
(31) Meigh, J. R.; McKenzie, A. A.; Sene, K. J. A grid‐based approach to water scarcity
540
estimates for eastern and southern Africa. Water Resour. Manag. 1999. 13, 85–115,
541
DOI 10.1023/A:1008025703712.
542
(32) Fuller, K., Shear, H., Wittig, J., Eds. The Great Lakes: an Environmental Atlas and
543
Resource Book, 3rd, ed.; Great Lakes National Program Office, US Environmental
544
Protection Agency: Chicago, IL, 1995.
545
(33) Rockström, J.; Falkenmark, M.; Karlberg, L.; Hoff, H.; Rost, S.; Gerten, D. Future
546
water availability for global food production: the potential of green water for
547
increasing resilience to global change. Water Resour. Res. 2009, 45; DOI
548
https://doi.org/10.1029/2007WR006767.
549
(34) Strahler, A. N. Quantitative analysis of watershed geomorphology. Eos, Trans. Amer.
550
Geophys. Union. 1957, 38(6), pp.913-920; DOI https://doi.org/10.1029/TR038
551
i006p00913.
552
(35) Annual Report of the Great Lakes Regional Water Use Database: Representing 2014
553
Water Use Data; Great Lakes Annual Water Use Report; Great Lakes Commission
554
(GLC): 2014; https://waterusedata.glc.org/pdf/wateruserpt2014.pdf.
28
ACS Paragon Plus Environment
Page 28 of 31
Page 29 of 31
Environmental Science & Technology
555
(36) Barlow, P. M.; Alley, W. M.; Myers, D. N. Hydrologic aspects of water sustainability
556
and their relation to a national assessment of water availability and use. J. Contemp.
557
Water. Res. Educ. 2010, 127, 10, pp. 76-86.
558
(37) Ridge, T.; Engler, J.; Mccallum, S.; O'bannon, F.; Pataki, G. E.; Ryan, G. H.; Taft,
559
B.; Ventura, J.; Harris, M.; Landry, B. The Great Lakes Charter Annex: A
560
Supplementary Agreement to The Great Lakes Charter. Documents on Water Law.
561
2001, 2; http://digitalcommons.unl.edu/lawwater/2.
562
(38) Alian, S. Characterization of ecological water stress in the U.S. Great Lakes Region:
563
A geospatial modeling approach. Ph.D. Dissertation, Michigan Technological
564
University, Houghton, MI, 2017.
565
(39) Variations in Withdrawal, Return Flow, and Consumptive Use of Water in Ohio and
566
Indiana, with Selected Data from Wisconsin, 1999-2004; Scientific Investigations
567
Report 2009–5096; US Geological Survey (USGS): Reston, VA, 2009;
568
https://pubs.usgs.gov/sir/2009/5096/pdf/sir20095096.pdf.
569
(40) Moore, R. B.; Dewald, T. G. The Road to NHDPlus—Advancements in Digital
570
Stream Networks and Associated Catchments. J. Am. Water Resour. As. (JAWRA).
571
2016, 52(4), pp.890-900; https://doi.org/10.1111/1752-1688.12389.
572
(41) NHDPlus Version 2: User Guide; Office of Water; US Environmental Protection
573
Agency: Washington, DC, 2018; https://www.epa.gov/waterdata/nhdplus-national-
574
hydrography-dataset-plus.
29
ACS Paragon Plus Environment
Environmental Science & Technology
575
(42) Great Lakes: Basic Information: Physical Facts; United States Environmental
576
Protection Agency (USEPA): Washington, DC, 2017;
577
https://www.epa.gov/greatlakes/physical-features-great-lakes.
578
(43) Estimation of shallow ground-water recharge in the Great Lakes Basin; Scientific
579
Investigations Report 2005–5284, US Geological Survey (USGS): Reston, VA, 2006;
580
https://pubs.usgs.gov/sir/2005/5284/.
581
(44) Wellogic User Manual for Local Health Department and State of Michigan Staff;
582
Michigan Department of Environmental Quality (MDEQ), 2011;
583
http://www.michigan.gov/documents/deq/dnre-ermd-dwehs-wellogic-
584
somlhdusermanual_334605_7.pdf.
585
(45) Map of surficial deposits and materials in the eastern and central United States (East
586
of 102◦ West Longitude); Geologic Investigation Series I-2789; U.S. Geological
587
Survey (USGS): Reston, VA, 2003; https://pubs.usgs.gov/imap/i-2789/.
588 589
(46) Hunt, B. Unsteady stream depletion from ground water pumping. Groundwater. 1999, 37, pp.98-102, DOI https://doi.org/10.1111/j.1745-6584.1999.tb00962.x.
590
(47) Watson, K. A.; Mayer, A. S; Reeves, H. W. Groundwater availability as constrained
591
by hydrogeology and environmental flows. Groundwater. 2014, 52, 225-238; DOI
592
https://doi.org/10.1111/gwat.12050.
593
(48) Molden, D.; Sakthivadivel, R. Water accounting to assess use and productivity of
594
water. Int. J. Water Resour. D. 1999, 15(1-2), pp.55-71; DOI https://doi.org/10.1080/
595
07900629948934.
30
ACS Paragon Plus Environment
Page 30 of 31
Page 31 of 31
Environmental Science & Technology
596 597
(49) Maidment, D. R., Morehouse, S. Arc Hydro: GIS for water resources, ESRI Press: Redlands, CA. 2002.
598
(50) Weiskel, P.K.; Vogel, R.M.; Steeves, P.A.; Zarriello, P.J.; DeSimone, L.A.; Ries,
599
K.G. Water use regimes: Characterizing direct human interaction with hydrologic
600
systems. Water Resour. Res. 2007, 43(4); DOI
601
https://doi.org/10.1029/2006WR005062.
602
(51) Michigan’s water withdrawal assessment process and internet screening tool;
603
Fisheries Division Special Report 55; State of Michigan Department of Natural
604
Resources (MDNR), 2011.
605
http://www.miwwat.org/wateruse/documents/Hamilton%20
606
and%20Seelbach%20SR55.pdf.
607
(52) Annual Report of the Great Lakes Regional Water Use Database: Representing 2010
608
Water Use Data; Great Lakes Annual Water Use Report; Great Lakes Commission
609
(GLC): 2012; https://waterusedata.glc.org/pdf/wateruserpt2010-liters.pdf.
610
(53) Consumptive Water, Use Coefficients for the Great Lakes Basin and Climatically
611
Similar Areas; Scientific Investigations Report 2007–5197; US Geological Survey
612
(USGS): Reston, VA, 2007; https://pubs.usgs.gov/sir/2007/5197/pdf/SIR2007-
613
5197_low-res_all.pdf.
614
31
ACS Paragon Plus Environment