L8 Design Document for Topographic Downscaling (land side)
- Peter Thornton
- Renata McCoy
- Teklu Tesfa
- Ruby Leung
cThe Design Document page provides a description of the algorithms, implementation and planned testing including unit, verification, validation and performance testing. Please read Step 1.3 Performance Expectations that explains feature documentation requirements from the performance group point of view.
Design Document
In the table below 4.Equ means Equations and Algorithms, 5.Ver means Verification, 6.Perf - Performance, 7. Val - Validation, 
- competed, 
- in progress, 
- not done
Overview table for the owner and an approver of this feature
1.Description | Topographic downscaling, as implemented within the land component |
| 2.Owner | Ruby Leung |
| 3.Created | |
| 4.Equ | |
| 5.Ver | |
| 6.Perf | |
| 7.Val | |
| 8.Approver | Peter Thornton, William Riley (Unlicensed) |
| 9.Approved Date | |
| V1.0 | Accepted |
Title: Implement topographic downscaling within ACME Land Model, with connections to atmosphere effort through coupler
Requirements and Design
ACME Land Group
Date:
Summary
This task includes:
1) Testing two approaches to represent subgrid topography (elevation, slope, and aspect) and select an optimal approach to balance the ability to capture subgrid variability of topography with computational efficiency.
2) Discuss with the land and atmosphere team the selection of approaches to inform the design of model data structure and coupling method to support the implementation of subgrid topography in ALM and CAM.
3) Develop input datasets needed to implement subgrid topography in ALM and CAM. These include (a) delineation of watersheds at 1/8 degree resolution (for the original design), (b) fractional area, mean elevation, slope, and aspect of each subgrid topographic landunit within each model grid (which can be the 1/8 degree watershed or the CAM-SE grid), (c) dominant soil texture for each subgrid topographic landunit, (d) fractional area of each PFT within each subgrid topographic landunit, (e) elevation profile within each subgrid topographic landunit to support inundation modeling, and (f) information needed to map between the ALM and CAM subgrid topographic classes (not needed for the fallback plan in which CAM and ALM share the same grid and subgrid topographic landunits).
Requirements
Requirement: Testing and comparison of subgrid classification schemes
Date last modified: // date
Contributors: Ruby Leung; Teklu Tesfa
Two subgrid classifcation schemes, global and local, were implemented over the Columbia River Basin based on the HydroShed DEM data at 90m resolution. The global method used fixed intervals for the subgrid elevation classes, while the local method defines class intervals by percentile areas using elevation profiles. We also compared two types of subgrid structures: geolocated (subbasins/grids divided into contiguous units) and non-geolocated (subbasins/grid divided into spatially non-contiguous units - i.e., percentage areas). Several metrics were used to evaluate the effectiveness of the two methods in capturing the subgrid variability of surface elevation while balancing computational requirements needed to model processes in each subgrid class. The local and non-geolocated approach was selected based on the metrics.
Requirement: Discuss approaches with the atmosphere and land teams
Date last modified: // date
Contributors: Ruby Leung; Teklu Tesfa
Prepared presentations to discuss the approaches and results.
Requirement: Development of input datasets for the subgrid classification schemes
Date last modified: // date
Contributors: Ruby Leung; Teklu Tesfa
First a global DEM dataset was developed by patching together the HydroSHED DEM with DEM data for the high latitude regions, Arctic, and Antarctic, while ensuring smooth transition across data boundaries.
Watersheds equivalent to 1/8th degree were delineated using TauDEM software. Python scripts were developed to implement the local and non-geolocated subgrid classification approach and to automate the process.
The scripts were applied to generate input data by continents. The first set of input data was developed for the same ALM and CAM grids based on the CAM-SE grid at 1-degree resolution.
Algorithmic Formulations
Design solution: Algorithm to determine subgrid topographic landunits
Date last modified:// date
Contributors: Teklu Tesfa; Ruby Leung
Algorithmic details are provided in an attached file:
Tesfa_LocalMethod_LandUnits_ACME_CLM_Model.docx
Design solution: Algorithm to calculate metrics to evaluate the subgrid classification schemes
Date last modified:// date
Contributors: Teklu Tesfa; Ruby Leung
Algorithmic details are provided in an attached file:
Tesfa_LocalMethod_ElvClasses_for_ACME_Atm_Model.docx
Design solution: Algorithm to define soil texture and PFT classes for each subgrid topographic landunit
Date last modified:// date
Contributors: Teklu Tesfa; Ruby Leung
Design and Implementation
Implementation: subgrid topographic landunits, evaluation metrics, soil textures and PFT classes
Date last modified: // date
Contributors: Teklu Tesfa; Ruby Leung
The local non-geolocated subgrid classification scheme has been implemented in Python.
The scripts use multiple ArcGIS tools and functions from the Geospatial Data Abstraction Library (GDAL). Details on implementation of the algorithms are in the file attached:
Tesfa_LocalMethod_LandUnits_ACME_CLM_Model.docx
Planned Verification and Unit Testing
Verification and Unit Testing: Verify subgrid topographic landunits
Date last modified:
Contributors:Teklu Tesfa; Ruby Leung
Plot the subgrid topographic landunits over selected continents to visually check for accuracy.
Calculate various statistical measures as a sanity check and to compare the methods.
Planned Validation Testing
Validation Testing: Validate subgrid topographic landunits
Date last modified:
Contributors: Teklu Tesfa; Ruby Leung
Ability to capture topographic patterns and sensitivity to area threshold values will be used to evaluate the two schemes. More specifically, maps of the number of subgrid units per grid has been compared against maps of topographic slope over topographically complex regions to evaluate their ability to capture topographic patterns. Plots of statistical measures have been used as a sanity check. Moreover, maps of subgrid units over selected continents will be used to check accuracy visually.
Planned Performance Testing
Performance Testing: topographic landunits
Date last modified:
Contributors: (add your name to this list if it does not appear)
Statistical metrics such as total number of subgrid units, average size and standard deviation in size, across multiple area threshold values has been used to evaluate computational efficiency and sensitivity to changes in area threshold.