Seasonal Water Yield – biodiversity-armenia.am

We apologize for possible errors in the machine translation of the site

ES of seasonal water flow redistribution
(InVEST Seasonal Water Yield)

ES of regulating seasonal water flow by terrestrial ecosystem was estimated and mapped using InVEST Seasonal Water Yield module. Seasonal flow redistribution is extremely important for Armenia, which has a seasonal climate with dry summers over a significant part of the territory. The model takes into account the monthly amount of precipitation, soil permeability, and the water balance of each pixel, including moisture that comes into it from the overlying pixels. The main resulting indicators for assessing the ecosystem service are quick flow (QF), that is, the generation of streamflow with watershed residence times of hours to days; and baseflow (B), that is the generation of streamflow with watershed residence times of months to years. Values of cumulative baseflow (Bsum) show the flow through a pixel, contributed by all upslope pixels. Baseflow ensures runoff maintaining during the dry season and possible droughts.

The InVEST Seasonal Water Yield model diagram (from Hamel et al., 2020)

The main results of ES modeling and mapping

GIS modeling and analysis, presentation of results – Eduard Kazakov (NextGIS OÜ, Estonia);
Analysis and presentation of results – Elena Bukvareva (BCC-Armenia);
Assessment of the alignment of results with real watersheds in Armenia – Vardan Asatryan (NAS RA)
Assessment of the relevance of modeling results for Armenia – external expert Alexander Arakelyan (NAS RA)

The presented results were obtained on the basis of the following open data sources
FAA – Forest atlas of Armenia (digital maps);
ARMSTAT – regional statistics of Armenia;
ESRI land cover – data for 2017 and 2023
WorldClim – World Maps, graphs, tables, and data of the global climate;
Weather in Armenia;
Global-AI-PET_v3 – Global Aridity Index and Potential Evapotranspiration Climate Database v3
Hydrosheds – boundaries of watersheds

InVEST model outputs are proxy variables that should be interpreted in relative terms, rather than physical quantities. Nevertheless, the identified values are useful for analyzing the spatial distribution of services across the country’s territory and their balance with indicators of service utilization by the population and the economy.
This section presents preliminary results of testing models for assessing and mapping ecosystem services. In the future, if a decision is made to use these models, they should be calibrated using hydrological measurements made in Armenia.
The methodology and results will be described in detail in a forthcoming publication.

The analysis was made for parts of 6th-level watersheds (Hydrosheds), which are located on the territory of Armenia. These parts of the watersheds are further named after their largest rivers:
– Aghstev (involves Getik and Voskepar tributaries)
– Akhuryan
– Arpa (involves the Arpa River, the Azat River and the Vedi River)
– Debed (involves Pambak and Dzoraget tributaries)
– Hrazdan (involves two parts – Lake Sevan drainage basin and its outlet River Hrazdan)
– Metsamor (involves Kasagh tributary)
– Vorotan (involves Vorotan River, the Voghji River, and the Meghri River)

Figure 1. Watersheds and points of cumulative baseflow values in the lower reaches of rivers

1. ES evaluation and mapping in physical indicators

The baseflow values (B) for the actual land cover are much higher, while the quick flow (QF) values, on the contrary, are lower compared to hypothetical cases where all natural ecosystems are replaced with bare soil or croplands. The difference between runoff values for the actual land cover and bare ground can be interpreted as the ecosystem service provided by terrestrial vegetation: B = 47.8 mm in average, and QF = -22.2 mm in average.

Table 1. Mean ES indicators for Armenia

Scenario	Baseflow, mm (B)	Quick flow, mm (QF)	Total runoff, mm (B+QF)	Share of B in total runoff, %
Land cover 2017	51.97	97.01	148.97	34.88
Land cover 2023	51.28	98.04	149.32	34.34
Bare ground scenario	3.43	120.22	123.65	2.78
Cropland scenario	3.58	124.96	128.54	2.78

Actual landcover 2023, B, mm (mean B=51.3) Actual landcover 2023, QF, mm (mean QF= 98.0)

Bare ground, B, mm (mean B= 3.4) Bare ground, QF, mm (mean QF= 120.2)
Croplands, B, mm Croplands, QF, mm

Figure 2. Maps of ES indicators for different scenarios. For detailed maps see section “Seasonal Water Yield” here

Terrestrial ecosystems provide more than 90% of the baseflow within watersheds and cumulative baseflow in the lower reaches of rivers

With the current landcover, the baseflow is on average 35% of the total rflow (from 28 to 40% in different watersheds). With the bare soil scenario, the baseflow is only 3% (from 2 to 4%).

Figure 3. Water flow in watersheds with current land cover and under bare ground scenario

Figure 4. Baseflow and quick flow in provinces with current land cover

We estimated ES volume provided by terrestrial vegetation as difference between ES indicator values for the current land cover in 2023 and the bare ground scenario (grasslands and trees were replaced by bare ground). Across watersheds, ecosystems provide 92%–95% of baseflow and reduce quick runoff by 13%–36%.

Figure 5. Baseflow and quick flow with current land cover (blue) and changes in both provided by terrestrial ecosystems (orange)

Figure 6. The share of baseflow in total flow with current land cover and under bare ground scenario

Table 2. Average baseflow and quick flow in watersheds with land cover 2023 and for bare ground scenario (all natural ecosystems are replaced by bare ground)

	Baseflow, mm			Share of B provided by terrestrial ecosystems, %	Quick flow			Total rflow (B+QF), mm		Share of B in total flow, %
	Land cover 2023	Bare ground scenario	Difference B₂₀₂₃-B_b_are	Share of B provided by terrestrial ecosystems, %	Land cover 2023	Bare ground scenario	Difference QF₂₀₂₃-QF_b_are	Land cover 2023	Bare ground scenario	Land cover 2023	Bare ground scenario
Aghstev	42	4	39	91.5	88	117	-30	130	121	33	3
Akhuryan	73	5	68	93.2	121	138	-17	194	143	38	3
Arpa	37	2	35	94.9	59	81	-21	97	82	39	2
Debed	78	7	72	91.7	116	148	-33	194	155	40	4
Hrazdan	53	3	50	94.2	132	150	-18	185	153	29	2
Metsamor	50	3	48	94.5	78	91	-13	128	94	39	3
Vorotan	31	2	29	92.7	79	105	-25	110	107	28	2
Armenia	51	3	48	93.3	98	120	-22	149	124	34	3

Table 3. Cumulative baseflow (Bsum) in the lower river reaches for actual land cover 2023 and for scenario when all natural ecosystems are replaced by bare ground

	Land cover, 2023	Bare ground scenario	Difference B_sum2023-B_{sum bare}	Share of B_sumprovided by terrestrial ecosystems, %
Aghstev	686063872	53702228	632361644	92.2
Akhuryan	1562785152	82415360	1480369792	94.7
Argichi	75929520	3704933	72224587	95.1
Arpa	313119360	11837469	301281891	96.2
Azat	349204672	11007498	338197174	96.8
Debed	2672169472	179894192	2492275280	93.3
Gavaraget	422896480	13807621	409088859	96.7
Hrazdan	3543649536	154972240	3388677296	95.6
Masrik	43279976	1879218.75	41400757.25	95.7
Metsamor	1676233984	72298784	1603935200	95.7
Vedi	134872176	6922562	127949614	94.9
Voghji	75920344	4065560.75	71854783.25	94.6
Vorotan	300472096	15817257	284654839	94.7

Table 4. Average baseflow and quick flow in provinces with land cover 2023

	Baseflow, mm	Quick flow, mm
Aragatsotn	61	91
Ararat	28	50
Armavir	22	47
Gegharkunik	55	151
Kotayk	67	90
Lori	74	116
Shirak	85	126
Syunik	31	79
Tavush	37	86
Vayots Dzor	39	66

.
Figure 7. The share of baseflow (%) provided by terrestrial ecosystems in watersheds and in the lower river reaches

A comparison of ES indicators for different vegetation types shows that this ES is performed most effectively in alpine and subalpine grassland zones, and in the desert, while its effectiveness is minimal in forest zone and marsh areas. This counterintuitive result is explained by the fact that this ES is determined not only by the type of vegetation, but also by other factors — namely, soil type, slope, precipitation, and temperature which affects evapotranspiration. The relatively low ES effectiveness in the forest zone can be explained by the fact that forests are mostly located in gorges, where the average slope (20°) is steeper than that in zones of herbaceous vegetation (16° in alpine, 17° in subalpine, 10–11° in steppe). Moreover, in this model, all differences between types of herbaceous vegetation are determined not by the vegetation itself, but by the conditions in which it grows. Unexpectedly, the highest effectiveness in baseflow generation was found in the desert zone, as the desert is represented by a single unique site entirely located on light, permeable soils.

__________________________a__________________b________________ c________________d________________e

Figure 8, a, b, c. Mean ES values of baseflow and quick flow (red dots) and value ranges (colored bars) of indicators for different vegetation types

2. ES changes from 2017 to 2023

All changes identified are determined only by changes in the landcover. Weather and climate changes are not taken into account. The maps show that the changes are sporadic and oppositely directed. Nevertheless, overall, they can be characterized as negative. In cases where ES indicators changed significantly, baseflow (B) decreased while quick flow (QF) increased. This means that the ability of terrestrial ecosystems to sustain baseflow during dry periods is declining. The only exception is the Arpa basin, where B has increased. The most significant negative changes occurred in the Shirak province and the Akhuryan watershed corresponding to land cover changes in the Shirak province. The reason is the expansion of the croplands at the expense of the grasslands in the Shirak province (see here).

Changes in B, mm Changes in QF, mm

For detailed maps see sections “Seasonal Water Yield – Dynamics” here

Changes in baseflow and quick flow from 2017 to 2023

		Baseflow, mean				Quick flow, mean
		B 2017, mm	B 2023, mm	Change in B, mm	Change in B, % relative to 2017	QF 2017, mm	QF 2023, mm	Change in QF, mm	Change in QF, % relative to 2017
Watersheds	Aghstev	42.1	42.4	0.2	0.6	87.3	87.6	0.3	0.3
	Akhuryan	79.0	73.2	-5.9	-7.4	116.6	120.7	4.1	3.5
	Arpa	37.0	37.3	0.2	0.7	59.3	59.3	0.0	0.0
	Debed	78.6	78.3	-0.3	-0.4	114.3	115.6	1.3	1.1
	Hrazdan	53.5	53.4	0.0	-0.1	131.4	132.0	0.6	0.5
	Metsamor	51.7	50.4	-1.2	-2.4	75.8	77.5	1.7	2.3
	Vorotan	31.0	31.0	0.0	0.1	78.6	79.2	0.5	0.7
Provinces	Aragatsotn	62.5	61.0	-1.5	-2.5	88.5	90.5	2.0	2.2
	Ararat	27.6	28.1	0.5	1.7	49.5	49.6	0.1	0.3
	Armavir	22.2	21.9	-0.3	-1.3	46.2	46.8	0.6	1.2
	Gegharkunik	54.6	54.5	-0.1	-0.1	150.5	151.1	0.6	0.4
	Kotayk	66.4	66.7	0.3	0.4	89.8	90.3	0.5	0.6
	Lori	74.1	74.0	-0.2	-0.2	114.5	115.8	1.3	1.2
	Shirak	91.3	85.1	-6.2	-6.8	121.3	125.7	4.4	3.6
	Syunik	30.9	30.9	0.0	0.1	78.6	79.1	0.5	0.7
	Tavush	36.7	36.9	0.1	0.3	86.2	86.3	0.1	0.1
	Vayots Dzor	39.3	39.4	0.1	0.3	66.0	66.0	0.0	0.0
Armenia		52.0	51.3	-0.7	-1.3	97.0	98.0	1.0	1.1

Cumulative baseflow (Bsum) in the lower reaches has decreased most in the Akhuryan River (-9.25%), and the Metsamor River (-3.03%), while in Voghji River it has visibly grown (+5.28%)

Changes in cumulative baseflow in the lower river reaches from 2017 to 2023

Rivers	Bsum 2023, mm	Bsum 2017, mm	Changes, mm	Changes, % relative to 2017
Aghstev	686,063,872	685,638,016	425,856	0.06
Akhuryan	1,562,785,152	1,722,085,760	-159,300,608	-9.25
Argichi	75,929,520	75,653,040	276,480	0.37
Arpa	313,119,360	309,875,328	3,244,032	1.05
Azat	349,204,672	345,485,152	3,719,520	1.08
Debed	2,672,169,472	2,684,654,848	-12,485,376	-0.47
Gavaraget	422,896,480	425,799,968	-2,903,488	-0.68
Hrazdan	3,543,649,536	3,543,354,368	295,168	0.01
Masrik	43,279,976	42,957,568	322,408	0.75
Metsamor	1,676,233,984	1,728,691,072	-52,457,088	-3.03
Vedi	134,872,176	132,632,480	2,239,696	1.69
Voghji	75,920,344	72,110,264	3,810,080	5.28
Vorotan	300,472,096	300,990,784	-518,688	-0.17

3. Supply – use balance

This section is based on a comparison of runoff estimates by the SWY InVEST model and data on water consumption from ArmStat regional statistics.

Water use in marzes in 2023, mln m3

Marzes	Drinking, domestic	Industry, communal, building	Agriculture, fish breeding, forestry	Total use
Aragatsotn	3.7	1.3	226.3	231.3
Ararat	6.7	6	799.3	812
Armavir	10.2	13.5	514.1	537.8
Gegharkunik	24.5	0.8	29	54.3
Kotayk	8.6	8.6	147.1	164.3
Lori	9.3	5.7	3	18
Shirak	7.9	2.2	53	63.1
Syunik	16.3	57.1	6	79.4
Tavush	8.1	1.7	11.1	20.9
Vayots Dzor	2.4	0.5	21	23.9
Yerevan city	59.8	40.6	156.5	256.9

Ararat, Armavir and Yerevan city are water-deficient marzes. The discharge from their territory is significantly less than the water consumption. The total flow from the territories of other marzes provides the total water consumption.
Baseflow is most important for agricultural water use because it provides flow during the hot summer months when water demand is greatest. Baseflow residence time spans months to years and the baseflow is distributed throughout this period, not just in the summer months. However, the comparison of baseflow and agricultural water use can be useful for assessing the water deficit in marzes. In these terms, the Ararat and Armavir marzes become even more water-deficient. Baseflow is less than agricultural water use also in Yerevan city and marzes Aragatsotn and Kotayk.

In the Aragatsotn, Ararat, and Armavir and marzes baseflow provides 5-7% of agricultural water consumption, in Kotayk 95% (it should be remembered that this figure does not include the flow from Sevan), in other marzes the baseflow is several times higher than water consumption.

	Baseflow, mln m3	Quick flow, mln m3	Total flow, mln m3	Water use in agriculture, fish breeding, forestry, mln m3	Total water consum-ption, mln m3	Difference between total runoff and total water consumption in marzes, 2023	Difference between baseflow and agricultural water consumption in marzes, 2023	Share of baseflow in agricultural consumption, %
Vayots Dzor	91	152	243	21	24	220	70	434
Tavush	100	233	333	11	21	312	89	898
Syunik	139	357	496	6	79	417	133	2323
Shirak	228	337	565	53	63	502	175	431
Lori	281	440	721	3	18	703	278	9369
Kotayk	139	189	328	147	164	164	-8	95
Gegharkunik	222	615	838	29	54	783	193	766
Armavir	27	58	85	514	538	-453	-487	5
Ararat	59	104	162	799	812	-650	-741	7
Aragatsotn	168	249	418	226	231	186	-58	7

The boundaries of marzes largely coincide with the boundaries of large basins

Basins	Marzes
Aghstev	Tavush
Akhuryan	Shirak
Arpa	Ararat and Vayots Dzor
Debed	Lori
Hrazdan	Gegharkunik, Kotaik and Erevan city
Metsamor	Armavir and Aragatsotn
Vorotan	Syunik

If errors caused by discrepancies between watershed and marz boundaries are disregarded, it is possible to roughly estimate the ratio of provided to utilized services in the basins. According to this rough estimate, water consumption exceeds flow in Arpa and Metsamor watersheds, which means that the population and economy in these basins consume underground waters or surface water from other watersheds.

4. A small by-product for education

The Seasonal Water Yield module requires data on the distribution of monthly precipitation across the territory of Armenia. However, the module does not account for the snow period when snow does not melt, and, consequently, there is no liquid precipitation. To address this, we have estimated the amount of liquid precipitation in Armenia by months. In addition to being used in the model, these results may be useful in explaining the climate characteristics of Armenia to a wider audience.

References
Hamel, P., Valencia, J., Schmitt, R., Shrestha, M., Piman, T., Sharp, R. P., Francesconi, W., & Guswa, A. J. (2020). Modeling seasonal water yield for landscape management: Applications in Peru and Myanmar. Journal of Environmental Management, 270, 110792. https://doi.org/10.1016/j.jenvman.2020.110792