Land-Cover Effects on Urban Energy Partitioning

Name:Khayrun Nahar Mitu Affiliation:Arizona State University Email:kmitu@asu.edu

Abstract

This proposal will study how different urban land-cover types are related to surface-energy partitioning. The main variables are net radiation (Rn), latent heat flux (LE), sensible heat flux (H), and ground heat flux (G). The workflow will use NCA-LDAS model outputs directly, together with a land-cover map for the study area. NCA-LDAS provides modeled Qle, Qh, Qg, Swnet, and Lwnet. Qle, Qh, and Qg will be used as LE, H, and G, respectively. Net radiation will be calculated as Swnet + Lwnet. The results will be grouped by land-cover class and compared using the energy-partitioning fractions LE/Rn, H/Rn, and G/Rn. No new energy-balance model will be developed for this proposal.

Keywords

CyberWater; urban energy balance; land cover; energy partitioning; NCA-LDAS; urban climate

1. Research Question or Modeling Problem

How does modeled energy partitioning differ among different urban land-cover types?

Urban land-cover types such as vegetation, bare soil, and impervious surfaces have different physical properties. These differences can be related to differences in surface-energy partitioning.

The proposed application will use an urban area in Phoenix, Arizona, as a hypothetical study area. The user will select a study area and simulation period and compare NCA-LDAS energy fluxes among different land-cover classes.

The main variables are:

• Net radiation (Rn)
• Latent heat flux (LE)
• Sensible heat flux (H)
• Ground heat flux (G)

The main energy-partitioning measures are:

• LE/Rn
• H/Rn
• G/Rn

The energy-balance residual will also be checked:

Rn - (LE + H + G)

The goal is to evaluate how modeled energy partitioning differs among land-cover classes.

2. Desired Workflow Outcome

The workflow will allow a CyberWater user to:

1. Select a study area.
2. Select a simulation period.
3. Retrieve NCA-LDAS model outputs.
4. Provide a land-cover map.
5. Match the land-cover map with the NCA-LDAS grid.
6. Calculate Rn from the NCA-LDAS radiation outputs.
7. Use Qle, Qh, and Qg as LE, H, and G.
8. Calculate LE/Rn, H/Rn, and G/Rn.
9. Group the results by land-cover class.
10. Compare energy partitioning among the land-cover classes.

The workflow will produce:

• Time series of Rn, LE, H, and G.
• Diurnal comparisons, if a sub-daily data source is used.
• Energy-partitioning fractions.
• Land-cover maps.
• Land-cover comparison plots.
• Summary statistics.

The main result will be a quantitative comparison of modeled energy partitioning among different land-cover types.

3. Required Inputs

Input Purpose source

Qle Latent heat flux (LE) NCA-LDAS

Qh Sensible heat flux (H) NCA-LDAS

Qg Ground/soil heat flux NCA-LDAS (G)

Swnet Net shortwave radiation NCA-LDAS

Lwnet Net longwave radiation NCA-LDAS

Land-cover map Identify surface types User-provided spatial dataset

Time range Select analysis period TimeRange

Space range Select study area SpaceRange

DEM, if needed Supporting spatial DEMAgent information

Observations, if Model evaluation User-provided available observations

NCA-LDAS V2.0 is produced with the Noah 3.3 land-surface model at 1/8° spatial resolution and provides daily model outputs. The dataset covers 1979--2016. These limits will be considered when selecting the simulation period.

NCA-LDAS provides modeled surface-energy variables including Qle, Qh, Qg, Swnet, and Lwnet. Qle, Qh, and Qg will be used directly as LE, H, and G.

Net radiation will be calculated as:

Rn = Swnet + Lwnet

The land-cover map is the main additional spatial input because the analysis will compare the NCA-LDAS fluxes among land-cover classes.

4. CyberWater Components That Already Exist

4.1 NCALDASAgent

The NCALDASAgent will be the main Data Agent for this workflow.

It will be used to retrieve the available NCA-LDAS model outputs needed for the analysis.

The required variables are:

• Qle
• Qh
• Qg
• Swnet
• Lwnet

Qle, Qh, and Qg will be used directly as LE, H, and G. Swnet and Lwnet will be used to obtain Rn.

If the existing NCALDASAgent does not provide Qle, Qh, and Qg, an additional data-access step may be needed to obtain these existing NCA-LDAS outputs.

4.2 TimeRange

The TimeRange module will define the start and end dates for the analysis.

4.3 SpaceRange

The SpaceRange module will define the geographic study area. If SpaceRange cannot select multiple locations at once, the NCA-LDAS data will be downloaded individually for each location. Each dataset will then be given to the workflow as a separate input file, similar to the forcing-file method in Activity 3.

4.4 Data-processing tools

Existing CyberWater data-processing tools can be used to:

• Calculate Rn.
• Calculate LE/Rn.
• Calculate H/Rn.
• Calculate G/Rn.
• Calculate the energy-balance residual.
• Prepare data for comparison.

4.5 DEMAgent

The DEMAgent can provide elevation data if it is needed for spatial processing or interpretation. It is optional for the main analysis.

4.6 Visualization tools

Existing CyberWater visualization tools can be used to display:

• Rn, LE, H, and G time series.
• Energy-partitioning fractions.

5. Missing CyberWater Components or Development Needs for This Workflow

The main development needs are related to accessing the required NCA-LDAS variables and linking them to land-cover classes.

5.1 Access to NCA-LDAS energy-flux variables

The NCA-LDAS product provides Qle, Qh, Qg, Swnet, and Lwnet. The existing NCALDASAgent should be checked to determine whether it can retrieve these variables directly.

If some variables are not currently exposed by the agent, the workflow NCALDASAgent would need to be extended.

5.2 Land-cover and NCA-LDAS grid matching

A preprocessing step may be needed to match the user-provided land-cover map with the NCA-LDAS 1/8° grid.

For example:

Land-cover map ↓ NCA-LDAS grid ↓ Land-cover class for each grid cell

The method should also handle grid cells that contain more than one land-cover type.

5.3 Land-cover-based energy analysis

A post-processing step may be needed to group NCA-LDAS outputs by land-cover class.

For each class, the workflow will calculate:

• Rn
• LE
• H
• G
• LE/Rn
• H/Rn
• G/Rn
• Rn - (LE + H + G)

If existing CyberWater tools cannot perform the grouping directly, a small analysis component or script may be needed.

5.4 Land-cover comparison visualization

If existing visualization tools cannot compare the fluxes by land-cover class, a simple additional visualization step may be needed.

6. Proposed End-to-End CyberWater Workflow

Conceptual workflow

Study Area + Time Period | v SpaceRange + TimeRange | v NCALDASAgent | v NCA-LDAS Outputs | +----------+----------+ | | | v v v Qle/Qh/Qg Swnet/Lwnet Land-Cover Map | | | v v v LE/H/G Rn Match to NCA-LDAS Grid | | | +----------+----------+ | v Energy-Partitioning Analysis | +--------+--------+ | | | v v v LE/Rn H/Rn G/Rn | | | +--------+--------+ | v Compare Land-Cover Classes | v Plots / Maps / Statistics

Step 1 --- Define the study

The user selects the study area and simulation period.

CyberWater components: SpaceRange and TimeRange.

Output: spatial and temporal domain.

Step 2 --- Retrieve NCA-LDAS outputs

The NCALDASAgent retrieves the required NCA-LDAS variables.

Required outputs:

• Qle
• Qh
• Qg
• Swnet
• Lwnet

Step 3 --- Prepare the energy variables

Use:

• Qle → LE
• Qh → H
• Qg → G

Calculate:

Rn = Swnet + Lwnet

Step 4 --- Prepare the land-cover map

Provide a land-cover map for the study area.

Step 5 --- Match land cover with the NCA-LDAS grid

Match each NCA-LDAS grid cell with a land-cover class.

If a grid cell contains several land-cover types, a defined rule will be used, such as the dominant land-cover class or land-cover fractions.

Step 6 --- Calculate energy partitioning

Calculate:

• LE/Rn
• H/Rn
• G/Rn

Also calculate:

Rn - (LE + H + G)

Step 7 --- Compare land-cover classes

Group the results by land-cover class and compare the mean, median, range, or other selected statistics.

Step 8 --- Visualize the results

Produce:

• Time-series plots of Rn, LE, H, and G.
• Energy-partitioning plots.
• Land-cover maps.
• Comparison plots by land-cover class.
• Summary statistics.

The final workflow will show how NCA-LDAS-modeled energy partitioning differs among land-cover types.

7. Risks and Uncertainty

7.1 NCA-LDAS spatial resolution

NCA-LDAS has a 1/8° grid. One grid cell can therefore contain more than one urban land-cover type.

7.2 NCA-LDAS temporal resolution

NCA-LDAS V2.0 provides daily output. It therefore cannot provide hourly energy-partitioning patterns.

7.3 Dataset time period

NCA-LDAS V2.0 covers 1979--2016. The selected analysis period must be within this period.

7.4 Land-cover classification

Errors in the land-cover map can affect the land-cover grouping and the final comparison.

7.5 Mixed land-cover cells

A single NCA-LDAS cell may contain several land-cover types. The method used to assign a class to a mixed cell can affect the results.

7.6 Model-derived fluxes

Qle, Qh, and Qg are model outputs from the NCA-LDAS Noah 3.3 land-surface model. They are not direct observations.

7.7 Data-agent limitations

The existing NCALDASAgent may not currently expose all required NCA-LDAS variables. This should be tested before implementation.

7.8 Energy-balance residual

The modeled fluxes may not exactly satisfy:

Rn = LE + H + G

The residual will therefore be checked and reported.

8. Questions for the CyberWater Team

1. Is it possible for the existing NCALDASAgent to retrieve Qle, Qh, Qg, Swnet, and Lwnet? directly?
2. If not, could the present NCALDASAgent be modified in order to provide these? variables?
3. Is there currently a CyberWater tool which can be used for matching a land-cover Does the raster match the NCA-LDAS grid?
4. What is the suggested way of dealing with grid cells that have multiple land-cover types?
5. Is it possible for the existing dataset-operation tools to calculate Rn and the energy-partitioning fractions?
6. Which CyberWater tool is recommended for grouping the model outputs? by land-cover class?
7. What visualization tool is best for comparing the energy fluxes? among land-cover classes?
8. Is there any existing instance of the NCA-LDAS model outputs being used directly?