49 / 2023-12-25 12:38:36

Modeling Physicochemical Parameters of Pao Cachinche Water Reservoir, Venezuela, Using Surface Reflectance from Landsat Satellite Images

physicochemical parameters,modeling,water reservoir,satellite images,surface reflectance

This research work addresses a topic of relative novelty in the area of the environment, referring to the applications of correlating environmental variables measured in the field with those estimated by satellite remote sensing. The relationship was carried out using a vector of surface reflectances estimated from seven spectral bands, including visible and infrared regions of the electromagnetic spectrum, from the spectral radiances (ER) detected by the sensors of Landsat satellite group: 1) Landsat 5 – Thematic Mapper (L5 TM), 2) Landsat 7 - Enhanced Thematic Mapper (L7-ETM) and 3) Landsat 8 - Operational Land Imager (L8-OLI), as independent variables. The dependent variables were represented by the annual field sampling carried out at four monitoring stations (MS) within the Pao Cachinche reservoir (PC-WR), from which eight physicochemical and biological parameters (PCBP) were determined, being determined analytically by the laboratory of Central Hydrological Company. The purpose was to evaluate quality of prediction of multivariable linear mathematical models, of each of the eight water quality parameters from PC-WR, two of these determined in 20 years, while than six PCBPs in 10 years, with an SR vector from 7 spectral bands of each satellite image. The RS was estimated from the ER emitted from the surface of the PC-WR and detected in the visible and infrared regions, from the 3 Landsat satellite platforms remotely, for each MS-PCBP, with a temporal resolution of 16 days. The method involved: 1) Data collection. 6 images of L5-TM (1996-2011), 11 images of L7-ETM + (2002-2013) and 3 images of L8-OLI (2014-2016) were acquired from USGS website. 2) Data processing. Absolute radiometric, topographic, and atmospheric corrections were applied to each image and processed in ENVI V-4.7 software. PCBP monitoring stations at PC-WR were georeferenced in ArcGIS version V-10.0 package, from where RS was extracted from 7 spectral bands of each satellite image. The calibration of multivariate linear model was carried out using Statgraphic software, obtaining coefficients associated with reflectances from seven spectral bands for each satellite image and parameters: 1) total phosphorus, 2) total nitrogen, 3) plankton, 4) Biochemical Oxygen Demand, 5) Chemical Oxygen Demand, 6) electrical conductivity, 7) pH and 8) total coliforms. The modeling of physicochemical parameters from PC-WR using surface reflectances recorded in spectral bands of Landsat satellite images as independent variables was satisfactory. The determination coefficient indicated that the adjusted models explained between 70.18 and 75.18% of the variability in physicochemical and biological parameters.