Elizabeth Tellman, MESc

2013 TRI Fellow in El Salvador

Ecosystem Services for Disaster Risk Reduction: Modeling the effects of land use on freshwater flood mitigation in El Salvador

My research measures and maps the impact of urbanization and deforestation on flood risk in El Salvador.  I am comparing flooding over time in two urban watersheds to one forested watershed. For each watershed, I rely on remote sensing to map land use in 5 major storms over the past 15 years, and household surveys to understand maximum flood depth at various locations for each storm.  I surveyed river cross-sections and sediment samples to understand both the geomorphology and infrastructure of the river system in each watershed. Finally, I have precipitation and elevation information for my entire region of study. All of this data is used to create a computer flood model of each watershed, calibrated on historic floods. This allows me to understand how changing land use has affected stream flow and flood patterns historically. The model is then used to run future scenarios of land use and climate change, showing flooded area and depth for each situation. Visualizing and understanding the differences in future scenarios of flood risk will help the NGO I work with, ACUA (Association of Community Water and Agriculture) develop better watershed management plans by optimizing forest protection where it is most crucial to mitigating flood risk.


Land Use Change and Ecosystem Service Sheds: Where Does Deforestation Impact Flood Mitigation in El Salvador?

Publication Text

Land Use Change and Ecosystem Service Sheds: Where Does Deforestation Impact Flood Mitigation in El Salvador?

Beth Tellman, MESc 20141


Country-scale studies and statistics of deforestation fail to capture the impact of land use on the watershed scale. The discourse of Forest Transition Theory (Mather 1992) to explain patterns of deforestation and reforestation in the tropics inadequately addresses spatial patterns of land use vital to understanding hydrologic ecosystem services. This article revisits Forest Transition Theory in the context of El Salvador, a place undergoing rapid land use change and vulnerable to flooding. As El Salvador recently passed its first national land use and zoning law in 2011, watershed science can influence decisions at this critical time. This article focuses on quantifying land use dynamics from 1992 – 2013 with remote sensing and analysis of the impacts on local hydrology for three peri-urban watersheds. A transition from forest to urban land use in the headwaters of the study sites has reduced flood mitigation services. Understanding hydrological impacts of land use change is an important part of comprehensive strategies for local NGOs navigating local land use processes to protect ecosystem services.

Introduction: Flood Mitigation, an Ecosystem Service at the Watershed Scale

Understanding how land cover and land use, such as forests, affect biophysical processes is key to planning sustainable landscapes that can support human well-being. Forests provide ecosystem services on discrete biophysical scales defined by a “serviceshed,” where ecosystems provide benefits to specific beneficiaries over a determined area (Tallis et al. 2013). The location and scale of a forest is essential to understanding the ecosystem services it might provide. Yet global statistics on deforestation are often aggregated at the country level, obfuscating the places where loss and gain in forest cover most acutely impact ecosystem services. Country-scale studies of deforestation fail to capture the impact of land use on the watershed scale, the serviceshed of flood mitigation.

Country-scale studies are often used to support Forest Transition Theory (Mather 1992), which explains patterns of deforestation and reforestation in the tropics. This article revisits Forest Transition Theory in the context of El Salvador, a place that is undergoing rapid land use change and is vulnerable to flooding. As El Salvador recently passed its first national land use and zoning law in 2011 (Decree #644, Government of El Salvador), watershed science can influence land use decisions and planning, which may erode or protect flood mitigation ecosystem services. My research quantifies land use dynamics and explores the potential impacts on local flood hydrology to inform this planning process for CORCULL (the Watershed Rescue Committee of La Libertad), a local NGO. After reviewing the literature on Forest Transition Theory and the implications of land use change on flooding in El Salvador, this article quantifies land use change on three peri-urban watersheds and discusses implications of local flood hydrology for the land use planning process.

Forest Transition Theory in El Salvador

Mather’s Forest Transition Theory (1992) describes the drivers of reforestation in developed countries as either economic transition or forest scarcity. Economic development transition implies that increased urbanization and intensification of production of agricultural lands allows other agricultural land to lie fallow. Lower population growth reduces demand to continue to expand agricultural land. The forest scarcity transition is driven by official government policies and programs to “save the forest.” While Mather mused that this may be occurring in the tropics, political ecologists have challenged and refined this theory in the context of El Salvador (Blackman et al. 2012, Hecht and Saatchi 2007, Hecht et al. 2006).

Studies by Hecht et al. (2006) and Hecht and Saatchi (2007) support the theory of a forest transition in El Salvador, estimating a 22% increase in low density forest cover, and a 6.5% increase in dense forest cover from 1990-2000. By analyzing this change with socioeconomic statistics, the authors found areas of reforestation to be highly correlated with areas of high out-migration rates (ibid.). As Salvadoran families become more dependent on remittances for cash, it may preclude the need to farm as a source of food and income. Furthermore, globalization, trade policies, and poor agrarian reform policies have caused a severe drop in agricultural prices, providing a further incentive to abandon farms. Hecht and Saatchi (2007) assert that the forest transition in El Salvador represents an opportunity for conservation. However, this country-scale analysis does not capture spatial patterns of deforestation on a watershed scale. Other studies indicate that driving forces of land use change in El Salvador stem from the Salvadoran civil war (1980-1992) and natural disasters, which drive urbanization and reduce population density in rural areas (Halliday 2006). Likewise, recent studies identify urbanization and international agricultural trade as global drivers of deforestation (DeFries et al. 2010). These urban land transitions, especially peri-urbanization (Seto et al. 2012), can greatly impact ecosystem services, including flood mitigation.

The forest to urban transition is the land cover change that most increases flood risk, as impervious surfaces such as cement allow for little infiltration. The effects of urbanization on flood mitigation are especially detrimental on peri-urban fringes near cities, which tend to be near catchment headwaters (Hollis 1975, Shuster et al. 2005). Thus, conversion from forest to urban cover near the headwaters of a watershed poses the greatest threat to losing the ecosystem service of flood mitigation and may accelerate flood risk of downstream populations.

Land Use Change on Peri-urban Watersheds

While studies such as Hecht et al. (2006) have used remote sensing to capture reforestation and deforestation dynamics in El Salvador over the past 20 years, none focus on the effects of deforestation on hydrological function. The Salvadoran Ministry of the Environment (MARN) has become increasingly concerned about the impacts of deforestation on the fringes of San Salvador and increasing flood vulnerability. MARN estimates that a 20% increase in urbanization from 1990-1999 increases peak flow up to 70% in small floods and 10% in extreme events in San Salvador (Erazo 2010). MARN emphasized a need for further research on impacts of land use and land cover change in peri-urban watersheds, three of which I selected for further analysis (see area of interest in Figure 1). The Comasagua, Jute, and San Antonio watersheds (~40, 20, and 22 km2 in area) have a mixture of humid and dry tropical forests on mostly andosol volcanic soil type, which has high infiltration rates and capacity for flood mitigation.

This research site, the Cordillera del Balsamo, a mountain range with coastal watersheds whose headwaters are on the fringe of San Salvador, is the operational region of CORCULL. CORCULL is interested in the hydrologic function of land use, as its main mission is to develop and implement watershed management plans to protect water resources and reduce risk for the region. Hydrologic studies are especially important now, as El Salvador recently passed its first zoning law in 2011 (Government of El Salvador), providing a 5-year policy window for municipal governments to develop land use plans.


I obtained Landsat satellite images for the Cordillera del Balsamo area for 1992, 2011, and 2013 at a 30-meter resolution. I chose 9 land cover classes based on hydrologic response (Su 2000). The original land cover classes were further divided into 12 land uses based on fieldwork and community land use workshops with CORCULL2. I performed a supervised classification for these 12 land classes with representative field data using a Maximum Likelihood algorithm. The 2013 land use map (Figure 3) reflects the final classes, including dense urban, medium urban, residential, agriculture, pasture, bare ground, coffee forest, riparian forest, secondary forest, mangrove, mines, and parks. CORCULL facilitated participatory workshops with communities living in the watersheds to identify errors in the 2013 land classification. Errors were verified with additional field visits to confirm actual land use and manually correct the image.


Figure 2 represents the percentage of major land covers in each of the studied watersheds (Comasagua, Jute, and San Antonio) for 1992, 2013, and projections according to the 2030 development plan (FIDSL 2005). The map in Figure 3 shows detailed land use for 1992 and 2013 in the three watersheds. The Comasagua watershed showed the strongest trend of reforestation in the 19-year period from 1992 to 2013, while San Antonio and El Jute showed moderate reforestation in the middle and lower portions of their respective watersheds. Urban land uses (dense, urban, and residential land uses together) have increased cover from 11% to 21% of the El Jute watershed, and from 8% to 25% of the San Antonio watershed. This means total urban cover has doubled in the El Jute basin and tripled in the San Antonio basin. While forest cover slightly increased, this secondary forest growth occurred in lower parts of these basins on sparsely vegetated and agricultural lands. Urban growth has occurred at the expense of coffee forest at headwaters for these two catchments.

Hydrological Implications of Land Use Change

The land use change data for this region suggests that while El Salvador may be gaining secondary forest regrowth in rural areas, urbanization is replacing forested headwaters of the San Antonio and El Jute watersheds. The rapid transition from coffee forest to urban land use has occured in part because the land at the headwaters of the San Antonio/Jute watersheds has become attractive for development. The cool climate (16 – 20˚C), high elevation, forest cover, proximity to the capital and a major highway make this land appealing for upscale housing and shopping. In addition, as coffee prices and support for the agricultural sector have decreased since the 1960s (Tellman et al. 2011), land has become less profitable for agroforestry and has been sold to major land developers. Comasagua, on the other hand, has strict municipal codes preventing deforestation and protecting coffee forest, preventing the land use transition seen in neighboring watersheds.

Forest transitions matter most to people in peri-urban watersheds such as San Antonio and El Jute, where flood risk may be greatly increased. Land use changes in El Jute and San Antonio watersheds have implications for the flood vulnerability of the 12,000 people who live in the downstream communities located in the floodplain. I am testing the hypothesis that this land use change is decreasing infiltration rates and exacerbating flooding in the San Antonio and El Jute watersheds with hydrologic and hydraulic models. Preliminary results for the El Jute watershed indicate that discharge rates at the peak of Hurricane Ida (a larger than 100-year storm that hit El Salvador in November 2009) have increased by 4 – 15 % as result of land use changes over the past 20 years. The largest increases in discharge are immediately downstream of dense urbanization built on previously sparsely vegetated or forested andosol soils with high infiltration rates.

One emblematic example of dense urbanization is the housing development of Los Cumbres, which is inhabited by upper class Salvadorans and foreigners. Los Cumbres paved over andosol soil and destroyed the ecosystem service of flood mitigation that this vegetated area in the upper watershed previously provided to poor local communities that have lived downstream for decades. Figure 4 shows the location of this land use change in the upper Jute watershed, and an inset of the Los Cumbres housing development.

Fieldwork to collect stream data and run participatory mapping workshops corroborated model results and gave insight into how changing disaster vulnerability affects the lives of Salvadorans living in the floodplain. One deforested hill slope in the El Jute watershed, which had been recently cleared for a large housing development, collapsed onto several small community homes during the last major storm in October 2011. While the developer rebuilt the damaged homes, the community continued to work with CORCULL to push local and regional government to stop the development project. Community leaders asserted that events such as these floods were larger and more frequent every year in part due to urbanization, “porque el agua ya no se consuma cuando llueve” (“because the water is no longer absorbed when it rains”). My modeling efforts quantify these changes and add scientific analysis to the experiences local Salvadorans have been trying to communicate to their government for years.

As the San Antonio/El Jute watersheds urbanize, the need to understand the links between land use change and impacts on the watershed is imperative. The national government’s priorities for development in this region include projections for increased urbanization to 2030; they differ greatly from CORCULL’s vision for land use and conservation. CORCULL’s land use plan for the region is being constructed at the appropriate ecological scale, one watershed at a time, with hundreds of community members over dozens of participatory workshops.

While CORCULL’s plan is conceived at the correct biophysical scale to consider hydrologic ecosystem services, the watershed scale does not match the political land use planning scale. A watershed is determined ecologically, not politically. Zoning plans are written and approved by each municipality, whose zoning districts and politics are incongruent with the watershed scale. The ecological properties of a watershed imply that the decisions of municipalities upstream affect those downstream. This means the supply of flood mitigation (upper watershed) is located in a separate municipality from the demand of the flood mitigation service (lower watershed). The lack of a National Water Law currently prevents these downstream communities from taking legal action against developers whose constructions upstream affect their flood vulnerability. While my results inform CORCULL’s land use plan at the watershed scale, implementation of scientific recommendations at the appropriate political scales is another challenge.


Modeling ecosystem services at the right scale is a useful first step to manage resources for environmental services. A careful analysis of land use changes in El Salvador reveals that while the country may be increasing its forest cover by 6.5% for dense forest (according to Hecht et al. 2006), deforestation at headwaters near the capital may be changing local hydrology. Forest Transition Theory obscures changes at a local level, by focusing on secondary forest recovery on a country-wide scale. In El Salvador, increases in urban cover by nearly 100% have displaced forests in upper watersheds near the capital city, producing consequentially detrimental effects for rural people downstream. Using country-level reforestation statistics as an indicator of decreasing environmental degradation in places like El Salvador ignores the scale at which environmental change matters for essential ecosystem services like flood mitigation.

Localized studies of environmental change in El Salvador can inform policy for local land use planning at the municipal scale, to match restoration of ecosystem services at the watershed scale. As municipalities develop their first zoning plans, hydrological modeling and remote sensing can add important information for decision-makers. However, understanding the biophysical changes in an ecosystem services shed is only the first step. Well-constructed watershed policy and increased community are necessary to make promise of the science of watershed management a reality.


This study was made possible by grant from the Tropical Resources Institute at Yale, Travel Grant for Latin America and Iberian Studies, Yale Institute for Biospheric Research, and the National Science Foundation. Field work was made possible and enjoyable by Oscar Ruiz, Coordinator of Water Resources for ACUA (United Community Association for Agriculture and Water), Zulma Hernandez, community coordinator for South La Libertad for CRIPDES (Association for Development El Salvador) and field assistants Manuelito, Rene Landaverde, and the community leaders and policemen who accompanied and ensured the safety of all field visits.


ACUA (Asociacion Comunitaria Unida para el Agua y la Agricultura). 2007. Management plan for storing and cleaning water in rural areas in the south of La Libertad: Study of watersheds Conachilo, Chilama, La Danta, Jute-San Antonio, Estero San Diego, Bocana Toluca. San Salvador; Engineers without Borders. [In Spanish.]

Blackman, A., B. Avalos Sartorio, and J. Chow. 2012. Land cover change in agroforestry: Shade coffee in El Salvador. Land Economics 88, 75-101.

Defries, R, T. Rudel, M. Uriarte, and M. Hansen. 2010. Deforestation driven by urban population growth and agricultural trade in the twenty-first century. Nature Geoscience 3, 178-181.

Erazo, A.M. 2010. Impacts of Land Use on Run-off in the Arenal-Montserrat Basin in the City of San Salvador, El Salvador, 1992-2009. Report for Ministry of Environment and Natural Resources and National Service of Land Studies. San Salvador, El Salvador. [In Spanish].

FISDL (Fondo de Inversion Social para el Desarrollo Local). 2005. Regional Development Plan for La Libertad. [In Spanish]

Government of El Salvador. 2011. Decree #644, Legislative Assembly of El Salvador. Ley de Ordenamiento Territorial y Desarrollo Territorial. Available: http://www.asamblea.gob.sv/eparlamento/indice-legislativo/buscador-de-documentos-legislativos/ley-de-ordenamiento-y-desarrollo-territorial. [In Spanish]

Halliday, T. 2006. Migration, risk, and liquidity constraints in El Salvador. Economic Development and Cultural Change 54, 893-925.

Hecht, S.B., S. Kandel, I. Gomes, N. Cuellar, and H. Rosa. 2006. Globalization, forest resurgence, and environmental politics in El Salvador. World Development 34, 308-323. doi:10.1016/j.worlddev.2005.09.005

Hecht, S.B. and S.S. Saatchi. 2007. Globalization and forest resurgence: Changes in forest cover in El Salvador. BioScience 57, 663-672.

Hollis, G.E. 1975. The effect of urbanization on flooding at difference recurrence intervals. Water Resources Research 3, 431-435.

Kauth, R.J. and G.S. Thomas. 1976. The Tasseled Cap, a Graphic Description of the Spectral-Temporal Development of Agricultural Crops as Seen by LANDSAT. In Proceedings of the Symposium on Machine Processing of Remotely Sensed Data, 41-51. West Lafayette, Purdue University.

Mather, A. 1992. The Forest Transition. The Royal Geographical Society 24, 367-379.

Munoz-Villers, L.E. and J. Lopez-Blanco. 2008. Land use/cover changes using Landsat TM/ETM images in a tropical and bio diverse mountainous area of central-eastern Mexico. International Journal of Remote Sensing 29, 71-93

Seto, K., A. Reenberg, C.G. Boone, M. Fragkias, D. Haase, T. Langanke, P. Marcotullio, D.K. Munroe, B. Olah, and D. Simon. 2012. Urban land teleconnections and sustainability. Proceedings of the National Academy of Science 109, 7687-7692. http://www.pnas.org/cgi/doi/10.1073/pnas.1117622109.

Shuster, W., J. Bonta, E. Thurston, E. Warnemuende, and D.R. Smith. 2005. Impacts of impervious surface on watershed hydrology: a review. Urban Hydrology 2, 263-275.

Su, Z. 2000. Remote sensing of land use and vegetation of mesoscale hydrological studies. International Journal of Remote Sensing 21 213-233.

Tallis, H., S. Wolny, J.S. Lozano, S. Benitez, S. Saenz, and A. Ramos. 2013. ‘Servicesheds’ Enable Mitigation of Development Impacts on Ecosystem Services. Natural Capital Project. Working Paper. Available: http://www.naturalcapitalproject.org/pubs/Servicesheds_Working_Paper_2013.pdf

Tellman, B., L.C. Gray, and C.M. Bacon. 2011. Not fair enough: Historic and institutional barriers to fair trade coffee in El Salvador. Journal of Latin American Geography 10, 107-127.


Tellman, B. 2014. Land Use Change and Ecosystem Service Sheds: Where Does Deforestation Impact Flood Mitigation in El Salvador? Tropical Resources Bulletin 32-33, 63-71.


  1. Beth Tellman is currently a doctoral student at Arizona State University School of Geographical Sciences and Urban Planning as a National Science Foundation Graduate Research Fellow. Her research focuses on understanding social and ecological resilience to flooding from community to global scales. Beth Tell- man received her MS from the Yale University School of Forestry and Environmental Studies, her B.S from Santa Clara University, and was a Fulbright scholar in El Salvador in 2010.]

  2. Adding data in addition to satellite imagery for classification, such as slope and elevation, is crucial to determine which forest land cover is used for coffee farming, such as forest cover over 800 m (Munoz-Villers and Lopez-Blanco 2008). Other data, such as brightness, greenness, and wetness bands helped distinguish urban cover from fallow agricultural land (Kauth and Thomas 1976).