13.3 GIS and remote sensing applications in hydrology

GIS and Remote Sensing Data Integration

GIS and remote sensing provide the spatial backbone for modern hydrologic modeling. Instead of relying on sparse point measurements, these tools let you capture how terrain, land cover, precipitation, and soil moisture vary continuously across a watershed. That spatial detail is what makes distributed and semi-distributed models possible.

Beyond model inputs, remote sensing data serves as an independent check on model outputs. Satellite-derived estimates of precipitation, soil moisture, and evapotranspiration can validate whether your model is producing realistic results. GIS techniques then tie everything together through terrain preprocessing, watershed delineation, and visualization of results.

Integration of GIS in Hydrologic Modeling

Digital Elevation Models (DEMs) are the starting point for almost any watershed modeling effort. A DEM is a gridded representation of land surface elevation, and from it you can derive the terrain attributes that drive water movement:

• Slope and aspect control how fast water moves and how much solar radiation a surface receives (affecting ET)
• Flow direction is computed using algorithms like D8 (assigns flow to one of eight neighboring cells) or D-infinity (allows flow to split between cells for more realistic divergent flow)
• Watershed delineation and stream network extraction use flow direction grids to trace drainage paths; tools like ArcHydro and TauDEM automate this process

Land Use/Land Cover (LULC) Maps represent the spatial distribution of surface types such as urban areas, cropland, and forest. These matter because land cover directly controls key hydrologic processes:

• Infiltration is high under forest canopy with organic soils but low on impervious urban surfaces
• Evapotranspiration varies with vegetation type, density, and growing season
• Surface runoff increases where infiltration is limited

In practice, you assign model parameters (like curve numbers in SCS-CN or Manning's roughness coefficients) based on LULC classes using lookup tables or empirical relationships.

Satellite Imagery from platforms like Landsat (30 m resolution), Sentinel-2 (10 m), and MODIS (250 m to 1 km) provides the raw data for deriving LULC maps and other products:

• Vegetation indices such as NDVI (Normalized Difference Vegetation Index) and EVI (Enhanced Vegetation Index) quantify vegetation health and density from multispectral bands
• Soil moisture can be estimated from microwave and optical data
• Land surface temperature and ET are derived from thermal infrared bands (e.g., the MODIS LST product), which feed into energy balance calculations

The tradeoff across these platforms is spatial resolution vs. temporal frequency. Landsat gives you detailed snapshots every 16 days, while MODIS covers the globe daily at coarser resolution.

Remote Sensing Data Analysis and Validation

Remote Sensing for Model Validation

One of the most valuable uses of remote sensing is checking whether your hydrologic model produces outputs that match independently observed conditions.

Precipitation Radar provides spatially distributed rainfall estimates that ground-based rain gauges simply cannot match:

• Missions like GPM (Global Precipitation Measurement) and its predecessor TRMM deliver near-global coverage with sub-daily temporal resolution
• These products capture the spatial variability of rainfall, which is especially critical in data-scarce regions where gauge networks are thin
• You can use radar-derived precipitation both as model input and as a validation dataset for comparing against gauge-corrected estimates

Soil Moisture Estimates from microwave remote sensing fill a major observational gap:

• The SMAP (Soil Moisture Active Passive) and SMOS (Soil Moisture and Ocean Salinity) missions measure near-surface soil moisture at roughly 9–40 km resolution
• These observations can validate soil moisture simulations or be incorporated directly into models through data assimilation, where observed values are blended with model predictions to improve accuracy
• Even at coarse resolution, satellite soil moisture helps constrain infiltration and runoff partitioning in models

Evapotranspiration Estimates derived from remote sensing provide a check on one of the hardest fluxes to measure directly:

• Energy balance models like SEBAL and METRIC use thermal imagery and meteorological data to estimate ET at the field scale
• Vegetation index-based approaches use NDVI or EVI as proxies for transpiration rates
• Comparing modeled ET against these satellite-derived estimates helps identify whether your model is over- or under-predicting water losses

GIS Techniques for Hydrologic Preprocessing

Before you run a hydrologic model, GIS preprocessing converts raw spatial data into the inputs the model actually needs.

Terrain Analysis extracts hydrologically relevant attributes from DEMs:

• Slope, aspect, curvature, and flow direction characterize how topography controls water movement
• Tools like ArcGIS Spatial Analyst and QGIS GRASS automate these calculations
• Derived products help identify areas of high runoff potential, erosion susceptibility, and likely groundwater recharge zones

Watershed Delineation defines the spatial boundaries of your modeling domain:

1. Fill sinks (artificial depressions) in the DEM so flow paths are continuous
2. Calculate flow direction for every cell
3. Compute flow accumulation to identify where drainage concentrates
4. Define a pour point (the watershed outlet)
5. Trace all cells that drain to that pour point to define the watershed boundary

This process also yields watershed characteristics like drainage area, shape indices, and drainage density, all of which influence hydrologic response.

Flow Routing determines how water moves through the delineated watershed:

• Routing algorithms like Muskingum-Cunge (a simplified hydraulic approach) and kinematic wave (based on shallow water equations) simulate the timing and attenuation of flow through channels
• These methods use channel geometry and slope derived from the DEM, combined with roughness parameters from LULC data
• The result is a prediction of when and how much flow arrives at different points in the stream network

Visualization of Model Results with GIS

Producing results is only half the job. Communicating those results clearly is just as important.

Spatial Visualization turns model outputs into maps that reveal patterns you'd miss in a table of numbers:

• Flood inundation maps, runoff depth grids, and recharge potential surfaces all communicate spatial variability at a glance
• Tools range from desktop GIS (ArcGIS, QGIS) to custom web mapping applications
• These maps help stakeholders quickly identify hotspots like flood-prone areas or drought-affected regions

Time Series Visualization shows how hydrologic variables change over time:

• Streamflow hydrographs, soil moisture trends, and precipitation time series are standard outputs
• Plotting observed vs. simulated values side by side is the most direct way to evaluate model performance
• Tools like matplotlib, ggplot2, or Tableau handle this well, and you can layer in statistical metrics (NSE, RMSE, bias) alongside the plots

Interactive Visualization takes communication a step further by letting users explore results on their own:

• Web-based dashboards built with libraries like Leaflet, D3.js, or Plotly Dash allow querying, filtering, and scenario comparison
• Users can zoom into specific sub-basins, toggle between variables, or adjust model scenarios
• This approach promotes stakeholder engagement and supports collaborative decision-making, especially when multiple management alternatives are being evaluated