Research presentation

Drought Index Monitoring based on

Multi-Source Remote Sensing Data

Introduction

• Drought is a major environmental hazard, especially in arid and semi-arid regions like parts of Nairobi County; impacting agriculture, water availability and livelihoods.

​

• Traditional drought monitoring relies on ground-based observations, which are often limited and delayed.

​

• This study leverages multi-source remote sensing data (NDVI, LST, TVDI) to enhance drought monitoring accuracy and early warning capabilities.

​

Background

Why this research matters?

​

• Kenya experiences frequent droughts, affecting millions.
• Existing drought monitoring systems lack high spatial and temporal resolution.
• Remote sensing-based indices like NDVI, LST, and TVDI provide spatially

continuous drought assessment.

• Key Concept: Temperature Vegetation Dryness Index (TVDI)
• Combines NDVI (vegetation health) and LST (land surface temperature)

to assess soil moisture deficit.

• Higher TVDI = Drier conditions, Lower TVDI = Moist conditions.

​

TVDI Calculation

• TVDI = ((LST - LSTmin) / (NDVImax – NDVImin)) × (NDVI - NDVImin)

​

TVDI Formulae Comparison

Methods & Data

Study Area

• Nairobi County – selected due to its urban-rural landscape, allowing analysis of drought trends in different land-use types.

Data Sources

• MODIS (NDVI & LST) & Landsat (NDVI & LST) from Google Earth Engine (GEE) for 2000-2020.
• Nairobi County shapefile for spatial analysis.

Methodology Workflow (KNIME Analytics Platform)

Data Collection:

• Downloaded NDVI & LST for Nairobi County from GEE.

Preprocessing:

• Corrected scanline errors (Landsat 7 SLC-off issue).
• Clipped raster data to Nairobi County boundary.

Merging & TVDI Calculation:

• Combined NDVI & LST into merged rasters.

Visualization & Analysis:

• Created TVDI maps & histograms (2000-2020).
• Performed comparative analysis of drought trends over time.

​

​

​

​

KNIME Workflow

Results

TVDI Trends (2000-2020)

• 2000-2012:
• TVDI values mostly moderate (~0.3-0.6 range).
• More vegetative cover & lower land surface temperatures.
• 2013-2020:
• Significant increase in TVDI values (~0.6-1).
• Higher drought intensity, possibly due to climate variability & urbanization effects.

​

Comparison of TVDI Maps

• 2000 vs 2010: Moderate changes in dryness.
• 2010 vs 2020: Significant drying trend observed, especially in urban/peri-urban areas.

​

Histograms:

• Shift towards higher TVDI values post-2012, indicating increasing dryness.

​

Key Observations

2000 vs. 2010:

• Expansion of Dry Areas: The 2010 map shows a noticeable increase in the spatial extent of areas with higher TVDI values (drier conditions) compared to 2000. This is most prominent in the eastern part of the region.
• Moderate Increase in Dryness: While there's an expansion of dry areas, the overall intensity of dryness (as indicated by the color scale) doesn't show a dramatic shift. The colors are generally similar in tone, suggesting a moderate increase in water stress.

​

2010 vs. 2020:

• Significant Intensification and Expansion of Dryness: The 2020 map exhibits a substantial increase in both the intensity and spatial coverage of high TVDI values. Nearly the entire region is covered in darker shades of red, indicating widespread and severe potential water stress.
• Dramatic Shift: The change from 2010 to 2020 is much more pronounced than the change from 2000 to 2010. It suggests a significant environmental shift towards much drier conditions.

​

Overall:

• The two comparisons together paint a picture of a gradual increase in potential water stress from 2000 to 2010, followed by a sharp escalation in drought conditions by 2020. The maps suggest a concerning trend towards increased dryness in the region over this 20-year period.

Comparing the 2000 and 2010 TVDI histograms reveals a subtle shift towards drier conditions in 2010. While both show a positive skew favoring wetter conditions, the 2010 histogram exhibits a higher frequency of mid-range TVDI values (0.02-0.06) and a slightly higher maximum TVDI (0.09 vs. 0.06), suggesting increased moderate dryness and slightly more intense localized dry spots. This indicates a potential, though not dramatic, increase in water stress between the two years.

Overview

This significant increase in TVDI values from 2000 and 2010 to 2020 suggests a marked escalation in drought severity in Nairobi County. Such a shift could stem from a variety of contributing factors:

• Climate Change Impacts: Rising temperatures and altered rainfall patterns could be exacerbating drought conditions.
• Land Use and Urbanization: Increased land conversion for urban development reduces vegetation cover, which can elevate surface temperatures, reduce soil moisture, and contribute to higher TVDI values. Reduced vegetation means less evapotranspiration, leading to increased surface dryness.
• Population Growth and Water Demand: Increased population growth puts pressure on water resources. Higher water demand for domestic, industrial, and agricultural use can lead to groundwater depletion and reduced availability of water for vegetation, exacerbating drought conditions.
• Environmental Degradation: Deforestation and land degradation, due to human activities like farming, logging, and construction, can lower soil moisture retention and reduce natural vegetation, increasing drought vulnerability. This degradation could lead to more extreme TVDI values as soils become drier.
• Data Collection and Processing Differences: It’s also important to consider potential differences in the data sources, sensors, or processing methods between the years. Improvements in satellite technology, resolution, or data accuracy may result in more sensitive measurements in 2020, which might highlight drought effects more clearly than in earlier years.

Discussion

How I Benefited from This Internship:

• Expanded and gained hands-on experience with Google Earth Engine (GEE), Python, and KNIME for geospatial data processing.
• Improved my skills in geospatial analysis, raster processing, and visualization.
• Learned scientific research methodologies for climate and drought monitoring.

​

Challenges Faced & Solutions:

• Scaling issues in TVDI calculations (2013-2020) – Resolved through rescaling and validation.
• Scanline errors in Landsat 7 data – Corrected using row interpolation techniques.
• Mismatch in NDVI & LST raster dimensions – Ensured spatial alignment before processing.

​

Future Plans:

• Improve TVDI calibration using ground-truth data.
• Investigate the impact of urbanization on drought intensity.
• Expand analysis to other Kenyan ASAL regions.

​

Conclusion

• Remote sensing-based TVDI monitoring effectively tracks drought patterns.

​

• Nairobi County shows increasing dryness post-2012, highlighting climate change & land-use changes.

​

• Findings can support early warning systems & climate adaptation strategies in Kenya.