Sediment accumulation expectations for growing desert cities: a realistic desired outcome to be used in constructing appropriately sized sediment storage of flood control structures

Many rapidly urbanizing desert cities (RUDC) around the globe experience an acute risk of flooding. To reduce this risk, properly engineered flood control structures (FCS) must account for sediment accumulation as well as flood waters. While the Phoenix area, USA, uses regional data from non-urban, non-desert watersheds to generate sediment yield rates, the proposed desired outcome for RUDCs is to base FCS on data related to urbanization. Wolman (1967 Geogr. Ann. A 49 385–95) recognized that sediment yields spike during a relatively short period of bare-ground exposure associated with urban growth, followed by surface sealing resulting in a great reduction in sediment yield. This research presents a new analysis of empirical data where two regression models provide estimates of a more realistic sediment accumulation for arid regions and also urbanization of a desert cities: (i) linear regression between drainage area and sediment yield based on a compilation of more than 150 global sediment yield data for warm desert (BWh Köppen‐Geiger) climate; and (ii) linear regression relating percent urban growth with sediment yield using available data on urbanization-generated sediment associated with growth of a desert city. The new model can be used to predict the realistic sediment accumulation for helping provide data where few data exists in urbanizing parts of arid Africa, southwest Asia, and India.


Introduction
Rapidly urbanizing desert cities (RUDC) often experience loss of life and property from flooding (Warner 2004). In particular, urban sprawl on alluvial surfaces (Jeong et al 2018) can place inhabitants at high risk of flooding (Middletone and Sternberg 2013). UNISDR (2012) reported more than half of the Jordan's disaster related mortalities during 1981-2010 could be explained by flash flooding, and the 2008 floods alone in Yemen led to increased national poverty (UNSIDR 2010). Flash floods in the Indian state of Rajasthan caused by once-in-a-200 year event in the Thar Desert killed more than 150 people (Telegraph 2006).
Resilient urban planning efforts recognize the need for flood control structures (FCS) (Muller 2007, Djordjević et al 2011, Liao 2012. Over-estimation of sediment volume generates additional costs, while under-estimation reduces life expectancy, or worse yet failure. Sweasy Dam on the Mad River of northern California exemplifies a circumstance where the FCS had completely filled and no longer functioned, requiring its removal (Mount 1995).
With sufficient funding, empirical data acquisition enables statistical estimation of sediment yield in developed settings such as the southwestern USA. However, lesser developed RUDC settings in Africa, Middle East, India, and Asia typically lack such data. Based on a global data compilation of sediment yield in warm deserts (the Köppen-Geiger BWh climate setting) by Jeong and Dorn (2019) and Vanmaercke et al (2014), only eleven published data points exist for urban settings outside of the USA (supplemental table 1 is available online at stacks.iop.org/ERL/14/ 125005/mmedia); all other measurements are from agricultural fields or grazing lands. Thus, the focus of this paper rests in establishing a new desired outcome for engineering FCS in growing desert cities that largely lack empirical data on sediment yields.
The Phoenix metropolitan region (PMR) is often used as a case study for the analysis urban climate changes (Georgescu et al 2009), the impact of ecosystem services and disservices (Zhuo et al 2012), urban biodiversity (Bateman et al 2015) and issues of urban sprawl and water quality (Hale et al 2015). The PMR, however, is an atypical desert city from the perspective of urban flood planning. Whereas many growing desert cities lack data on sediment yield (Dedkov and Mozzherin 1984, Gallaire 1986, Laronne 1990, Terfous et al 2003, Nyssen et al 2004, Liénou 2007, FAO 2008, Amogu 2009, Billi and el Badri Ali 2010), decisions on the size of FCS and sediment delivery infrastructure in PMR benefited from an abundance of data.
The desired outcome here rests in estimating sediment yields informed by the 'Wolman model' using published and unpublished data from warm desert watersheds. Wolman (1967) recognized that land-use land cover (LULC) changes increase sediment yields that spike during bare-ground exposure from urbanization, followed by sealing of the urban surface that then dramatically reduces sediment yield. This paper presents a new analysis of empirical data for the PMR, showing how FCS are experiencing sedimentation (FCDMC 2015a), and then uses this new analysis to proposes a strategy whereby RUDC can estimate sedimentation rates.

Study area
Phoenix, Arizona, is the fifth largest USA city, increasing an order-of-magnitude since the advent of air conditioning (Grimm et al 2013). Since the 1950s, the PMR sprawled (figure 1) from migrants seeking employment and low-cost housing. Phoenix rests entirely in the Sonoran Desert-classified as a warm desert, Köppen-Geiger BWh climate setting with precipitation averaging 208 mm split evenly between summer and winter maxima (Climate Office of Arizona, https://azclimate.asu.edu/climate/climateof-phoenix-summary/).
The flood control district of Maricopa County (FCDMC) operates and maintains 22 flood control dams (figure 1 and table 1), where 15 were built from 1954 to 1991 by the Soil Conservation Service (USDA-SCS) for 100 year flood protection. Five dams were built from 1974 to 1985 by Army Corps of Engineers (USACE) for 200-500 year events. One dam was designed by the FCDMC (2015b), where the total storage capacity was calculated from a gage heightvolume relation curve.
The latest hydraulics design manual for FCS in the PMR focuses on empirically derived equations to estimate sediment yields. Unfortunately, model building data come exclusively from 21 non-urbanizing basins (FCDMC 2015a) monitored from 1942 to 1974 (figure 2 and supplemental table 1). Furthermore, 10 of these basins are in non-desert settings of California and New Mexico (FCDMC 2015a), and no data pertain to urbanization (figure 2). In contrast, sediment yield data from 1989 to 2013 (figure 2) reported by showing the locations of main dams and their watersheds. Urban sprawl of PMR over the period from 1912 to 2010 shows the dam watersheds have experienced a wide range of land-use changes. Sediment yield data associated with the urban sprawl was reported by Jeong and Dorn (2019). The data used to estimate sediment accumulation behind 22 main dams in PMR using extrapolation. See method section for details. The numbers refer to dams identified in this paper's data tables. Urban boundaries from 1912 to 2010 are extracted from land cover classification by Central Arizona-Phoenix long-term ecological research. Data available at https://sustainability.asu.edu/caplter/data/view/knb-lter-cap.1.11 and https://sustainability. asu.edu/caplter/data/view/knb-lter-cap.650.1/. Because of the limited extent of the land use/land cover maps, some dam catchments on the margin of central Phoenix did not included in the extent of urban sprawl. Jeong and Dorn (2019) focus on changes in sedimentation associated with urbanization (figure 1).

Methods
This paper starts by presenting a commonly used model to estimate sediment yield using the relation between sediment yield and drainage area under a 'condition without urban sprawl'. Then, the proposed model developed to meet the desired outcome uses the relation between sediment yield and annual urban growth (AUG) under 'a condition with urban sprawl. Both models are applicable only in Koeppen's BWh warm desert settings. Sediment yield estimates enable the calculation of an annual rate of storage loss (SL) and expected life of FCS when the storage capacity of FCS is known (table 1). The predicted sediment yield for FCS in PMR will be compared to the designed sediment yield (table 2) for FCS built by USACE that were designed for 100 years of sediment accumulation (USACE 1982(USACE , 1984b, except for the Saddleback FRS and Powerline FRS that were designed for 50 years (USDA-SCS 1977, EMNRCD andFCDMC 2013).
Unfortunately, published sediment yield data from soil erosion are presented in different ways, sometimes to different audiences whether they are scientists or policy makers (Leopold 1968, UNCCD 2008, Li et al 2015, Zhang and Huang 2015. To avoid confusion, this paper defines SSY as an annual area-specific sediment yield measured in units of weights (Mg km −2 yr −1 or tons acre −1 yr −1 ) and area-specific bulked sedimentation volume (SSV) as the annual area-specific sediment yield measured in units of volume (m 3 km −2 yr −1 or acre-feet mile −2 yr −1 ).

Model 1: sediment yield prediction under a condition without urban sprawl
An inverse relation exists between SSY and drainage area (Dendy and Bolton 1976, Milliman and Syvitski 1992, Lahlou 1996, Einsele and Hinderer 1997, and this relationship allows prediction of SSY in ungauged basins (De Vente et al 2007). The methods supplement provides detailed information about this inverse relationship, where in model 1, the SSV of FCS is best Figure 2. Period of reservoir sedimentation surveys in the Sonoran Desert where PMR situated in. Note the sparseness of data for the latter part of the 20th century that used as representative area-specific sedimentation volume (SSV) when major dams of PMR built (FCDMC 2015a), which is inconsistent with rapid urban growth of PMR. The reservoir identification numbers held on supplemental table 1. PMR urban area was calculated by LULCC map provided by Central Arizona-Phoenix long-term ecological research. Circles marked the published year because measurement period is not reported, indicating the SSV measurement probably earlier than the published year. Table 2. 100 year-capacity loss and expected life of FCS based on the total capacity of FCS and sediment yields predicted by two models. The reported error term derives from the 95% confidential interval. Abbreviation: n.a., not available.
Model 1: without urban sprawl condition Model 2: with urban sprawl condition 100 year capacity loss (%) Expected life (years)  (table 2). Then, the annual rate of storage loss (ASL) calculates: where ASL is measured as a percent of storage volume lost through sedimentation (Graf et al 2010), TSC indicates the total storage capacity of FCS (m 3 ). FCS in PMR was designed to provide a total cumulative sediment storage capacity for 50 years or 100 years of sediment storage (USDA-SCS 1977, USACE 1981, 1982, 1984a, 1984b, EMNRCD and FCDMC 2013 3.2.1. Measuring urbanization impact on the SSV from desert watersheds Figure 4 illustrates the conceptual model used here to quantify LULCC-related parameters using GIS procedures. Urbanization processes entail exposure of bare ground due to construction, that then accelerates soil erosion by enhancing rain splash and overland flow (Jeong and Dorn 2019). To estimate the magnitude of bare ground exposure or urban growth during the time period from T1 to T2 (BG1 in figure 4), the area of intersection of natural land cover in T1 and urban land cover in T2 was calculated using GIS. The imperviousness for the time period from T1 to T2, followed by sealing of the urban surface, resulted in a great reduction in sediment yield in T2 (U2). Then, the length of the time period (L1) allows calculation of AUG and imperviousness (AI).

Sediment delivery ratio (SDR)
As described in the development of Model 1, a negative relation exists between SSV and drainage area (A D ). The model 2, however, does not consider the negative relation between SSV and A D . Therefore, the concept of SDR related to A D is adopted for model 2 and an equation provided by the latest FCDMC Hydraulics manual (FCDMC 2015a) is used to calculate SDR. The methods supplement provides detailed information.

Prediction of urban acceleration on sediment yield
Based on Wolman's (1967) model, I hypothesized that the two LULCC parameters AUG and imperviousness may be related to eroded volume (EV) (figure 5; see methods supplement). The p-value for AUG (0.017) is lower than the alpha level of 0.05 and shows statistical significance ( figure 5(a)), but the p-value for annual imperviousness (0.293) is greater than 0.05, indicating that it is not statistically significant ( figure 5(b)). Unlike Wolman's (1967) model prediction, there is no observed inverse relation between EV and annual imperiousness, which may likely because of the increased runoff in urbanized watersheds and high sediment transport efficiency of urban drainage network (Russell et al 2017). Therefore, only the regression between EV and AUG was used to model the prediction of SSV for FCS ( figure 5(a)). The relation between EV (m 3 km −2 yr −1 ) and AUG (km 2 ) for the 18 catchments experiencing urbanization in the Sonoran Desert and PMR is ( figure 5(a) The storage capacity loss and life expectancies of FCS were estimated using equations (1) and (2) (table 2). Table 3 presents predicted SSV for each time period (SSV(Tx)).

Results
4.1. Sediment yield prediction from the two models Model 2 was designed to meet the desired outcome of modeling sediment erosion in rapidly urbanizing desert cities with minimal costs associated with acquiring empirical data. Accordingly, model 2 shows a significant increase in SSV where sites experienced urbanization ( figure 1 and table 2). Urbanization . Conceptual model to quantify land use-related parameters using GIS. The dashed line is the imaginary line that does not exist on the land use/land cover classification map, which is to show the exposed bare ground. The exposed bare ground during the time period from T1 to T2 (BG1) can be calculated by the intersection of natural land cover (natural vegetation and soil/desert) in T1 and urban land cover (asphalt/road, concrete/buildings, urban mixture and residential) in T2. The imperviousness for the time period from T1 to T2 can be quantified by urban land cover in T2 (U2). The annual urban growth and imperviousness was calculated from dividing by the length of time period. See method section for details.   (1955-1975), T2 (1975-1985), T3 (1975-1995), T4 (1985-1990), T5 (1990-1995), T6 (1995-2000), T7 (2000-2005), T8 (2005-2010), n.a.: not available. n.g.: no growth. The unit of annual urban growth (AUG) is km 2 yr −1 and SSV is m 3 km −2 yr −1 . * The PMR is a NSF site for long-term ecological research in an arid urban setting; two LULCC datasets with a 30 m resolution from CAP LTER quantified urban growth and imperviousness: (i) LULC in 1912LULC in , 1934LULC in , 1955LULC in , 1975LULC in and 1995 (https://sustainability.asu.edu/caplter/data/view/knb-lter-cap.1.11); and (ii) LULC in 1985,1990,1995,2000,2005 and 2010 (https://sustainability.asu.edu/caplter/data/view/knb-lter-cap.650.1/). These data were supplemented by analysis of historic aerial photography from the Maricopa County Tax Assessor website (https://gis.maricopa.gov/GIO/HistoricalAerial/index.html) and Google Earth. Supplemental table 2 shows data from 18 stock pond catchments (Jeong and Dorn 2019) used to make the regression equation.

**
Because the coverage of the two LULC datasets are different, LULCC for some time periods was not fully spatially covered. FCS ID 1, 2, 20, 21, and 22 in table 2 was excluded for quantification of urban growth and imperviousness, because the LULC does not fully cover the whole basin area. For FCS ID 3, 4, and 5, the former LULC dataset spatially covers whole drainage basin, but the latter dataset does not cover it; thus, only T1 and T3 could be used. The FCS ID 18 and 19 drainage basins were only covered by the latter dataset; thus, T4, T5, T6, T7 and T8 were used. For the remaining FCS where the two datasets fully cover the whole drainage area, T1, T2, T4, T5, T6, T7 and T8 were used unless there was no urban growth for each time periods. impact on sediment yields is best illustrated with Phoenix-area FCS ID 6, 9, 11-14 and 17 (figure 1). In contrast, the controls of 7,8,15,16,18 and 19 did not experience urbanization (figure 1), and there were only minor differences in predictions by model 1 and model 2 ( figure 6).
Based on model 2 predictions, Cave Buttes Dam (FCS ID14) has the highest SSV, followed by McMicken Dam (FCS ID11) that experienced considerable urban growth ( figure 1 and table 3). In contrast, the model 1 prediction shows the highest SSV from Sunset FRS (FCS ID4) followed by Casandro Dam (FCIS ID3) that both have small drainage areas (table 2).
Eleven FCS had larger predicted SSV from model 2 than model 1 by up to 11.3×(mean: 3.8×), but six FCS had larger predicted SSV from model 1 than model 2 by up to 1.2×(mean: 1.1×) (table 2). These findings reveals that urbanization's impact on sediment yield is far more dominant than drainage area.
The sustainability of FCS depend on SL over a 100 year period (100 year-SL) and ELF using equations (1) and (2). Figure 2 and table 2 present these data. The highest 100 year-SL is predicted by model 2 for McMicken Dam, followed by Cave Buttes Dam  (figure 7). The ELF of the two dams are 161±114 and 398±289 years, respectively. Model 1 predicted the highest 100 year-SL from White Tanks FRS#4 (FCS ID9), followed by Casandro Dam and the ELF of the two dams are 960±17 and 1013±20 years, respectively ( figure 7). Projects for sustainable sediment managements were planned and accomplished where sedimentation issues had been identified based on accumulated data. For example, for the Rittenhouse Basin, the first phase of excavation was planned to increase the capacity of the East Maricopa Floodway  . Comparison between the designed SSV when FCS built and predicted SSV based on the two regression models with the 95% confidential interval. Table 2 provides data used to construct this figure. along with a regular maintenance plan to mitigate the effects of erosion and sedimentation. This resulted in the McMicken Dam Fissure Risk Zone Remediation Project (FCDMC 2006).
The predicted SSV from the two models identified considerable spatial variation in the sustainability of FCS in the PMR (figure 8). The most vulnerable FCS were once far from the urban boundary (figures 8(a), (c)), but urban growth has resulted greater vulnerability of FCS in drainage areas experiencing this growth (figures 8(b), (d)). In contrast, Dreamy Draw Dam (FCS ID15) and the Guadalupe FCS (FCS ID16) are slightly overbuilt ( figure 6). This is because of a combination of two factors: they are in locations where urban growth was decades ago; and they are in locations where much of the watersheds are now natural preserves that are free from the urban acceleration of erosion.   Table 2 provides data used to construct this figure.

Comparison with designed sediment yield
The SSV used when the FCS were originally designed ('designed SSV') is higher than predicted by model 1 for Saddleback FRS (FCS ID2), but lower for powerline FRS (FCS ID20). The predicted SSV of Dreamy Dram Dam from both model 1 and model 2 is lower than designed SSV. The designed SSV for FCS ID 12-14 was higher than predicted SSV from model 1, but lower than predicted SSV from model 2 ( figure 6).
The designed SSV underestimated SSV up to 4.2× for three FCS when using model 2 and overestimated SSV up to 3.0× for five FCS when using model 1 (table 2). The general tendency of underestimated designed SSV compared to model 2 prediction and overestimated designed SSV compared to model 1 prediction may indicate that storage capacity might have been appropriate for the time of FCS completion. However, urban sprawl condition has been an ongoing process for the planned life of the structures (50-100 years; supplemental table 3).
The designed storage capacity for sedimentation under conditions of a 'future with urban sprawl' maybe too small; this requires efforts to reduce sediment inflow to the reservoir or remove sedimentation from behind FCS such as by dredging. Consider the designed SSV (143 m 3 km −2 yr −1 ) of Cave Buttes Dam estimated from USACE (1982) when the dam was built in 1980 (table 1). The original design SSV is only a bit lower than the model 2 SSV (180 m 3 km −2 yr −1 ) during the T1 (1955-1975) period of an 'existing condition' (table 3). Based on model 1 SSV (53 m 3 km −2 yr −1 ) its storage capacity for sedimentation was over-built in a 'future without urban sprawl condition', but its SSV increased significantly after the same was constructed (SSV in T2, T4-T8 in 'future with urban sprawl condition' table 3, figures 1 and 2). The urban acceleration of erosion led to a 4.2× higher SSV than designed SSV (table 2), and this reduces the expected design life for sediment storage from 100 years to 23 years.
Other examples come from an analysis of the storage capacity for sedimentation behind New River Dam, Adobe Dam (FCS ID13) and Dreamy Draw Dam. These structures were over-built at the time of FCS completion based on the model 2 (table 3), but the SSV of New River Dam and Adobe Dam increased significantly due to urbanization starting in 1995 (table 3), leading to the insufficient storage for sedimentation (table 2). In contrast, Dreamy Draw Dam had sufficient storage for sedimentation, even after urbanization (tables 2 and 3).
Very little published research exists on the importance of rapid urbanization for sediment projection in a desert city. However, the identification of earth fissures and land subsidence around three Phoenix-area FCSs led to the need to re-estimate sedimentation rates during the period of rapid urbanization. Powerline, were classified as high hazard potential structures by Arizona Department of Water Resources due to the proximity of earth fissures and land subsidence (EMNRCD and FCDMC 2013). Thus, SSV was re-estimated by consulting engineers in and FCDMC in 2010and 2016(Maricopa County 2010. The new SSV was estimated as 286 m 3 km −2 yr −1 , 291 m 3 km −2 yr −1 , and 284 m 3 km −2 yr −1 for the Powerline, Vineyard and Rittenhouse structures, respectively (JE Fuller 2008). Then, a 2010 study (FCDMC 2010a(FCDMC , 2010b(FCDMC , 2010c estimated SSV for these three respective structures at 113, 130, and 72 m 3 km −2 yr −1 . Both JE Fuller (2008) and the FCDMC (2010) estimate higher SSV than designed. These independent studies provide broad confirmation of the importance of urbanization on increasing sediment yields in a rapidly urbanizing desert city.

Discussion of the desired outcome of reliable sediment yield forecasting in RUDCs
Forecasting reliable future sediment yield is an essential part of designing storage capacity for sedimentation behind FCS (Vanmaercke et al 2011(Vanmaercke et al , 2015. Rather than measuring SSV of reservoirs by bathymetry surveying (e.g. Rakhmatullaev et al 2011), or backestimated sediment curves from reservoir sedimentation survey data (Tebbi et al 2012), many researchers use models to predict the reliable sediment yield in RUDCs due to the lack of existing sediment yield data (Vanmaercke et al 2011, Fathizad et al 2014, Abdullah et al 2017. A variety of models estimate soil erosion that fills in FCS. The thesis of this paragraph is that none of the existing models truly or appropriately consider the effects of urbanization. The drainage area-sediment yield relationship is commonly employed (e.g. Wasson 1994, Griffiths et al 2006, Vanmaercke et al 2011 and does not include impact of land use changes. The Revised universal Soil Loss equation (RUSLE) (Renard et al 1997) is also often used to predict soil erosion (e.g. Abdullah et al 2017) and sediment yield (e.g. Fathizad et al 2014) using the SDR. C factor in the RUSLE accounts for land use-land cover and management impacts on soil erosion, but the C-factor was based on empirical data from North America so that might be impractical in other regions where land cover conditions and cropping practices differ (Zhao et al 2017). Models are also are designed for the plot scale or, at most, small catchments (Boardman 2006, Syvitski andMilliman 2007) that do not have the capability for sediment delivery through channels to FCS (Nearing et al 2005). A semi-quantitative model developed by the Pacific Southwest Inter-Agency Committee (PSIAC 1968) uses a rating technique through field visits, complemented with data on surface geology, soils, climate, runoff, topography, ground cover, upland erosion, channel erosion and sediment transport and land use (e.g. Fathizad et al 2014, Abdullah et al 2017; however, acceleration of soil erosion and sediment yield was not included in the land use factor. A global empirical model, BQART developed by Syvitski and Milliman (2007), was also used to calculate long-term suspended sediment loads where the anthropogenic factor (Eh) is part of factor B in the BQART model. However, the BQART model was trained by 294 global river basins with 10 2 -10 7 km 2 drainage area. Importantly, the BQART model estimates human and urbanization impacts on sediment yield using an indirect method (population density and GNP per capita) rather than direct method such as urban growth rate proposed in this paper ( figure 4).
In summary, previous methods to predict sediment yield have limitations that include: (1) considerable costs and time to measure local sediment yield data (e.g. bathymetric survey in reservoir, installation of gauging station); (2) drainage area scale (e.g. RUSLE, BQART); and (3) urban acceleration of erosion either not considered or based on indirect variables such as population density and GNP per capita (BQART). Prior models used to predict sediment yield in RUDC do not directly quantify urban growth and its impact on sediment yield, this despite evidence that urban growth is important factor leading to one to two orders of magnitude increases in sediment yield (Russell et al 2017).
This research reveals that predicting soil erosion without considering urban sprawl led to a significant underestimation of sediment yield (figure 6), resulting in reducing available storage for flood control (figures 7 and 8). These findings emphasize that estimating sediment yield requires assessing urban growth's impact. Yet, most urbanizing desert cities around the globe do not have local financial and technical resources to carry out the detailed empirical research presented in this study (The World Bank 2014). A cost-efficient and simple method to predict reliable future sediment yield is an important priority.
The suggested solution for forecasting reliable sediment yield in RUDCs would be a two-step process First, urban growth would be estimated (e.g. Al-Ahmadi et al 2009, Tayyebi et al 2014, Yagoub and Al Bizreh 2014, Alqurashi et al 2016. Then, planners would use model 2 suggested in this paper to provide an appropriate output for designing sustainable FCSs in RUDCs.

Conclusion
RUDC must reduce flood risks to become economically, socially, and environmentally more resilient. Within cities, poorer inhabitants often experience greater risks in self-built informal housing (O'Hare and Rivas 2005). Thus, the growing need for flood control in RUDC has led urban planners to build FCSs (Muller 2007, Djordjević et al 2011, Liao 2012. However, sediment accumulation behind these FCS often reduce usable capacity (Morris and Fan 1998). The sedimentation problem is closely linked to LULCC from bare ground exposure during urban urbanization (Wolman 1967) that increases soil erosion one to two orders of magnitude (Russell et al 2017). Therefore, forecasting reliable sediment yields from urban sprawl is a desired outcome for RUDC sustainability.
The basic finding of this research is that the desired outcome of a new model to predict sediment yield under conditions of urban sprawl is more accurate than the prior approach using empirical data from non-urban settings. The proposed model is based on the relation between sediment yield and AUG rate; hence, planners in RUDC can now predict sediment yield to improve sustainable FCS construction. The proposed desired outcome would reduce RUDC vulnerability to flood risks, particularly where the cities do not have substantial financial or technical resources (The World Bank 2014).
After implementation of the desired outcome, an eventual goal for RUDCs would be sustainable sediment management (SSM) (Kondolf et al 2014). The city of Taiz in Yemen exemplifies one approach to SSM when it built 21 sedimentation traps upstream (The World Bank 2014), which was an initial step towards resilience. Based on the fiscal reality of many RUDCs, another feasible management approach would be a sediment yield reduction enforced by regulating construction permits to comply with the Arizona Pollutant Discharge Elimination System (FCDMC 2015c).