Applications

We applied the model to the Upper Segura River catchment (2,589 km2) under present and future projected climate conditions. The Upper Segura catchment is located in the headwaters of the Segura River in southeastern Spain. The climate in the catchment is classified as temperate (80% of the catchment) and semi-arid (20%). The catchment-average annual precipitation is 570 mm (1981-2000) and the mean annual temperature is 13.2 ºC (1981-2000). The main landuse types are forest (45%), shrubland (40%), cereal fields (7%) and almond orchards (4%). The main soil classes are Leptosols (38%), Luvisols (27%), Cambisols (16%) and Calcisols (11%). There are 5 reservoirs located in the catchment with a total capacity of 663 Hm3, which are mainly used to store water for irrigation purposes.

All input data were prepared at a 200 m grid size. Daily precipitation data were obtained from the SPREAD dataset (Serrano-Notivoli et al., 2017) and temperature data were obtained from the SPAIN02 dataset (Herrera et al., 2016). Soil textural fractions (sand, clay and silt) and soil organic matter content were obtained from the global SoilGrids dataset (Hengl et al., 2017). A Digital Elevation Model was obtained from the SRTM dataset (Farr et al., 2007). The spatially distributed rock fraction map was obtained by applying the empirical formulations from (Poesen et al., 1998), which determine rock fraction based on slope gradient.

The soil erosion model requires landuse-specific input for plant height (PH), stem density (NV), stem diameter (D), ground cover fraction (GC) and, optionally, the Manning's roughness coefficient for vegetation (nvegetation). The user needs to specify whether the landuse class is non-erodible (e.g. pavement and water), tilled or non-vegetated (e.g. bare soil or tilled orchards). We obtained values for each of these parameters through observations from aerial photographs, expert judgement and as part of the calibration procedure (Table 1). The tillage parameter RFR was set to 6, which corresponds to Cultivator tillage (Morgan and Duzant, 2008). The input parameters change when a crop is harvested, therefore, we varied the input parameters according to the sowing-harvest cycle representing the cropping cycle for horticulture and cereals.

Table 32: Input parameters for the soil erosion model (1 T = tillage, NE = no erosion, NV = no vegetation, 2 Day of the Year, 3 Obtained from (Chow, 1959))

landuse class	PH	NV	D	GC	manning	sowing	harvest	other1
	(m)	(stems m-2)	(m)	(fraction)	(s m-1/3)	(doy)2	(doy)2
cereal	0.75	500	0.025	0.31	n.a.	288	166	T
(harvested)	0	0	0	0	n.a.			T
huerta	0.5	500	0.01	0.5	n.a.	n.a.	n.a.	T
horticulture	0.3	6.25	0.25	0.39	n.a.	288	166	T
(harvested)	0	0	0	0	n.a.			T
tree crops	2	n.a.	n.a.	< 0.01	n.a.	n.a.	n.a.	T,NV
vineyard	1	n.a.	n.a.	0.02	n.a.	n.a.	n.a.	T,NV
forest	10	n.a.	n.a.	0.53	0.23	n.a.	n.a.
shrubland	0.5	n.a.	n.a.	0.45	0.13	n.a.	n.a.
water/urban	0	0	0	0	n.a.	n.a.	n.a.	NE

Here we present a selection of model results to illustrate the main capabilities of soil erosion and sediment transport modules. Soil erosion shows an important intra-annual variability due to seasonal changes in climate forcing and vegetation cover (Figure 1). Soil erosion follows the precipitation sum for crops with little to no ground cover (i.e. tree crops and vineyard), with high values in the winter, spring and autumn months and low values in the summer months. Some crops show a distinct peak in the vegetation development in the spring (April-May), e.g. huerta and horticulture. While this period has a relatively high precipitation sum, soil erosion decreases as a consequence of the increased vegetation cover indicated by the NDVI in this period. The temporal variation of the vegetation development of cereals and horticulture shows a slightly distinct pattern from the other landuse classes. Both crops show an increase in the spring months (March-May), which indicates the rapid growth of these crops in these months. However, during the summer months (June-August) the NDVI decreases, which coincides with the period when the crops are harvested, followed by the post-harvest period. In the latter period, we assume bare soil conditions for these crops. For both crops this ultimately results in the highest annual erosion rates in the post-harvest period (October).

Figure 44: Monthly precipitation sum (mm), NDVI (-) and soil erosion (Mg km-2 yr-1) per landuse class for the period 1981-2000. The gray area indicates the period when cereals and horticulture are harvested and model parameters are changed to bare soil conditions

We simulated also the impacts of a projected climate change scenario, by comparing predicted soil erosion rates and sediment yield under the reference scenario (1981-2000) with a future scenario (2081-2100). We used a future emission scenario from the Representative Concentration Pathways (van Vuuren et al., 2011). For this exercise we used projected climate data for RCP8.5 obtained from one Regional Climate Model (CLMcom MPI-ESM-LR) from the EURO-CORDEX initiative (Jacob et al., 2014). The climate forcing (precipitation and temperature) was bias-corrected using quantile mapping (Themeßl et al., 2012). The climate change scenario projects a decrease of the annual precipitation sum decreases, however, extreme precipitation is projected to increase.

In the reference scenario the highest hillslope erosion (SSY) is projected in the river network (Figure 2), where accumulated runoff causes an increase of soil erosion rates. In the future climate scenario, the catchment-median hillslope erosion increases from 43.3 to 55.2 Mg km-2 yr-1, an increase of 27.7%. This shows that the increase in extreme precipitation has a more pronounced impact on soil erosion than the decrease of annual precipitation sum. Reservoir sediment yield (SY) decreases in all five reservoirs between 42.4-59.0% in the future climate scenario. While it is likely that a decrease of hillslope erosion in the western part of the catchment causes a decrease of reservoir SY, it is less obvious why in the eastern part of the catchment an increase in hillslope erosion is not reflected in an increase in reservoir SY. The explanation for this lies in the fact that a decrease in precipitation sum causes a decrease of accumulated runoff and, subsequently, a decrease of sediment transport capacity, increased sediment deposition and decreased reservoir SY.

Figure 45: Specific sediment yield (Mg km-2 yr-1) and reservoir sediment yield (Gg yr-1) for the reference (1981-2000) scenario and the change (%) for the future (2081-2100) scenario.