4. DISEÑO DE LA PLANTA DE TRATAMIENTO DE AGUA POTABLE
4.6. ESQUEMA ELÉCTRICO DE LA PLANTA DE TRATAMIENTO DE
Modeled products are preferred due to the challenges that come with remotely sensed products such as algorithm changes when using multiple satellites and hence multiple sensors and algorithms (Liu et al. 2017). Additionally, short temporal coverages and gaps due to flight paths of the satellites prove acquisition of a continuous dataset for uninterrupted monitoring a task.
128
A Land surface model for food security monitoring should have a rainfall product, a long history of data and both drought and hydrological communities should utilize the datasets to estimate land surface states. GLDAS is a global terrestrial modeling system that produces optimal fields of land surface states by incorporating satellite and ground-based observations (Rodell et al. 2004). It drives NOAH, Mosaic, Community Model and Variable Infiltration Capacity (VIC). However, GLDAS utilizes a global rainfall dataset not optimized for data scarce food insecure regions (McNally et al. 2017) and hence the preference in utilizing the FLDAS products.
FEWSNET was created in 1985 by USAID as a provider of early warning and analysis of food insecurity issues. It acts as a support system as it provides data to help governments and agencies plan and respond to humanitarian crises. FLDAS is a customized LIS by NASA as a decision support tool in the functions of FEWSNET by producing models and estimates of hydroclimatic data over food insecure areas of Africa (McNally et al. 2017). FLDAS relies on the FEWSNET rainfall and is thus customized for the needs of data scarce food insecure regions in Africa. FLDAS constitutes of meteorological input, a land surface model input, and a post-processing evaluation component. NOAH and VIC Land Surface Models drive FLDAS.
FLDAS_NOAH is a 4-soil layer water and energy balance Land Surface Model with a spatial resolution of 0.1°. It available from 1982 to present, uses CHIRPS and MERRA2 and 1 KM NCEP modified IGBP land cover map from MODIS. NOAH soil moisture is available at daily, 5-day, 10-day, monthly temporal resolution. Evaporation in NOAH land surface models includes the weighted sum of canopy intercepted evaporation, transpiration from vegetation canopies and evaporation from bare soils based on the respective land surface coverage fractions (McNally et al. 2017).
The assumption in the calculation of potential evapotranspiration is a saturated ground surface with no canopy resistance. Canopy intercepted evapotranspiration scales PET by intercepted canopy water content and maximum canopy capacity. Transpiration scales PET with solar radiation, vapor pressure deficit, air temperature and soil moisture. Bare soil evaporation scales PET by a factor of current soil moisture availability. The surface runoff is computed with a two-layer conceptual approach from the Simple Water Balance Model. The upper layer produces surface runoff from excess precipitation when there is no soil moisture deficit. Subsurface runoff is produced as gravitational drainage from the bottom soil layer. Other parameters include monthly greenness fraction and albedo, soil texture datasets, vegetation and soil parameters.
In FLDAS_VIC, FLDAS uses VIC in water and energy balance to better represent surface fluxes (sensible heat, latent heat, ground heat, outgoing longwave radiation) where evapotranspiration is a greater proportion of rainfall than runoff. Run at a sub-daily time step, it closes the energy balance by iteratively adjusting the surface temperature and surface energy fluxes to balance incoming radiation. It has a spatial resolution of 0.25°. It is available from 2001 to present, uses RFE and GDAS, 1 KM UMD land cover classification from AVHRR. RFE
129
is from NOAA CPC derived from infrared and microwave observations blended with WMO GTS data, available from 2000 at 0.1° with 1-day latency (McNally et al. 2017).
Evaporation in VIC land surface models includes the weighted sum of canopy intercepted evaporation, transpiration from vegetation canopies and evaporation from bare soils based on the respective land surface coverage fractions. Canopy intercepted evapotranspiration is calculated by scaling Reference evapotranspiration (ReET) with canopy architectural resistance for humidity and aerodynamic resistance for heat and water. Canopy transpiration is calculated by scaling ReET with canopy resistance. Bare soil evaporation is calculated by scaling ReET by the current soil moisture conditions, wilting point soil moisture, and field capacity. Subtracting runoff and infiltration capacity at 0-10cm soil layer determines the surface runoff. Other parameters include soil texture and bulk density, soil characteristic curve parameters, monthly LAI and UMD land cover classification. ERA-Interim is a reanalysis product describes atmosphere, land surface and/or ocean conditions using forecast models and assimilation systems. The models extrapolate archived and current observations to generate estimates of the atmosphere or land surface parameters such as air temperature soil moisture (Dee et al. 2011). ERA-Interim is a global atmospheric reanalysis product from 1979. ERA-Interim is produced through sequential assimilation in 6-hour cycles (Dorigo et al. 2012). Estimation of the evolving state of the atmosphere and its underlying surface in each cycle encompasses a combination of available observations and prior information from a forecast model. ERA-Interim uses the ECMWFs Integrated Forecast System to constrain the state evolution within each analysis window and update bias correction estimate parameters. IFS comprise of three coupled components for atmosphere, land surface and ocean waves. The ERA-Interim forecast is provided by TESSEL land surface scheme. It has a resolution of 79km (Berrisford et al. 2011). Optimal assimilation technique produces estimates of temperature and relative humidity by combining 2m relative humidity and temperature over land with estimates from the recent analysis (Douville et al. 2000, Decker et al. 2012). The model produces gridded products of volumetric soil moisture content at 0.75°.
6.3 Datasets
NASA Land Data Assimilation System consists of land surface models forced with observations. The FLDAS_NOAH and VIC daily dataset covering a duration of 2010-2016 was downloaded from the NASA Land Data Assimilation System website. The resolution of NOAH is 0.1° where the resolution of VIC is 0.25°. The NOAH and VIC netCDF files were extracted in MATLAB software to acquire daily means of soil moisture within the catchment. The NOAH 0.1° data was resampled to 0.25° to be spatially consistent with the VIC data. ERA-Interim is a reanalysis data set based on the ECMWF Integrated Forecast System. The downloaded dataset had a spatial resolution of 0.75°, at a depth of 0-7cm (Albergel, de Rosnay, Balsamo, et al. 2012). The dataset was selected as it has a high global spatial consistency (Scipal et al. 2008, Dorigo et al. 2010). The daily readings
130
downloaded were at 06:00 UTC. The data assessed covered the period January 2010 to December 2016. The data were resampled to a resolution of 0.25°.
SMOS reprocessed level 2 soil moisture version 5.51 (SM_REPR_MIR_SMUDP2) Microwave Imaging Radiometer (MIR) NetCDF dataset is distributed by ESA. The ascending mode was considered as it is least affected by surface heating. Night time are preferred because of equilibrium conditions of surface soil, canopy and near-surface air (Owe et al. 2008). SMOS was selected for the study as it is one of the sensors dedicated for remotely sensed soil moisture retrievals operating at L band and has a relatively long acquisition period (from 2010) as compared to SMAP which was launched in 2015. At low frequencies, L band is sensitive to water content of soils and can penetrate through vegetation, which is an important factor in selection of the dataset given that the Kilombero catchment comprises of varying land covers (Entekhabi et al. 2010, Kerr et al. 2012). The modeled soil moisture products (FLDAS_NOAH, FLDAS_VIC, and ERA-Interim) are produced daily. The satellite-derived moisture from SMOS however, suffers from data gaps explained by the flight swaths of the satellite and hence weekly composites were generated from the SMOS dataset. Inevitably, this translated to computing the weekly averages for the modeled datasets in preparation for the data merging (Section 6.4.2) 16-day composites MODIS NDVI on Terra satellite was obtained on the LP DAAC from the year 2010 to 2016. 16-day MODIS NDVI composites were adopted for this study because they had the least variation in phenometric dates when compared to the 8 day composite MODIS in a study to assesses the influence of different phenometric extraction methods (Wessels et al. 2009). There are therefore 23 images available per year. These were preferred to reduce the atmospheric effects and clouds. Dummy data was generated for 2009 and 2017 as the phenology is not calculated in the first and last year of assessment as seasonality is only possible for n-1 center most seasons (Eklundh and Jönsson 2017). The data are in the Sinusoidal projection. The study area is covered by tile h21v09. On downloading the data sets, they were batch projected with the Modis Reprojection Tool to geographic coordinates.
TRMM is a multi-satellite precipitation analysis product. An assembled precipitation estimate is generated from measurements of passive microwave measurements on multiple satellites. Infrared measurements are used to estimate precipitation. These are then calibrated against the assembled precipitation. Infrared estimates fill the gaps of the assembled precipitation. Precipitation from ground-based radar, rain gauge, and disdrometer are used in the calibration of the multi-satellite product. TRMMs, spatial (0.25°) and temporal (3h) resolutions favor its use over other precipitation products (Koutsouris et al. 2016). The daily timescale was selected for the analysis. Daily precipitation is generated from the 3-h TRMM Multi-Satellite Precipitation Analysis TMPA (3B42). NASA Goddard Earth Sciences Data and Information Services Center (GES DISC) produce the data. Datasets used covered the period 2010 to 2016.
131
6.4 Methodology
The flow diagram (Figure 6.1) describing the research process is shown below;
Figure 6.1: The data flow diagram in soil moisture assessment in the Kilombero catchment